Toward the integration and convergence between 5G and TSN technologies and architectures for industrial communications: A survey

IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025 259
Towa d he In eg a ion and Con e gence Be ween
5G and TSN Technologies and A chi ec u es
o Indus ial Communica ions: A Su ey
Jo ge Sasiain , Da id F anco , Asie A u xa , Jasone As o ga ,
and Edua do Jacob ,Senio Membe , IEEE
Abs ac —Communica ion ne wo ks a e in cons an e olu ion
o adap o he e e -changing demands o mode n se ices
and applica ions. On he one hand, echnologies wi hin he
amewo k o he i h gene a ion o communica ion ne wo ks
(5G), cu en ly in a s eady cou se owa ds 6G, a e being designed
wi h unp eceden ed pe o mance and lexibili y capabili ies o
empowe he digi al ans o ma ion o he e ogeneous e ical
indus ies. On he o he hand, Time-Sensi i e Ne wo king (TSN)
is being de eloped o enable con e ged ne wo k in as uc u es
able o sa is y he mos s ingen communica ion equi emen s
in ega ds o bounded low la ency and eliabili y. In e icals
such as manu ac u ing, a join usage o 5G and TSN is e y
p omising o enable e olu iona y use cases ha coexis wi h he
cons ain s imposed by indus ial communica ions and sys ems.
In his pape , we exhaus i ely su ey he echnology landscape
conce ning he in eg a ion be ween 5G and TSN echnologies,
including he pe spec i es o bo h indus y and academia. In
he i s pa o his pape , we p o ide a comp ehensi e e iew
o ele an s anda ds and ou comes om indus y ini ia i es.
We i s examine he capabili ies p o ided by TSN and hei
applica ion o indus ial communica ions. Likewise, we con ex u-
alize he 5G a chi ec u e and i s key enabling echnologies, and
la e p o ide a comp ehensi e e iew o he a chi ec u e and
mechanisms o enable he in eg a ion be ween TSN and 5G. In
he second pa o his pape , we ex ensi ely su ey he s a e
o he a om academic li e a u e conce ning p oposals ha
con ibu e o enabling, imp o ing, and/o demons a ing di e en
aspec s o he in eg a ion o in e wo king be ween 5G and TSN
echnologies.
Index Te ms—Time-sensi i e ne wo king (TSN), 5G, 5G-TSN,
ime-sensi i e communica ions, ime synch oniza ion, i ual-
iza ion, so wa e-de ined ne wo king (SDN), managemen and
o ches a ion, Indus y 4.0.
Manusc ip ecei ed 12 July 2023; e ised 16 Janua y 2024 and
15 May 2024; accep ed 21 June 2024. Da e o publica ion 3 July 2024; da e
o cu en e sion 17 Feb ua y 2025. This wo k was suppo ed in pa by
he Na ional P ojec unded by MICIU/AEI/ 10.13039/501100011033 i led
“Enabling Na i e-AI Secu e De e minis ic 6G Ne wo ks o Hype -Connec ed
En i onmen s” (6G-INSPIRE) unde G an PID2022-137329OB-C44, and
in pa by he Basque Go e nmen h ough he P ojec “Social Ne wo k
o Machines” (SONETO) o he ELKARTEK P og am unde G an
KK-2023/00038. (Co esponding au ho : Jo ge Sasiain.)
The au ho s a e wi h he Depa men o Communica ions Enginee ing,
Facul y o Enginee ing, Uni e si y o he Basque Coun y UPV/EHU,
48013 Bilbao, Spain (e-mail: [email p o ec ed]; da[email p o ec ed];
asie [email p o ec ed]; [email p o ec ed]; [email p o ec ed]).
Digi al Objec Iden i ie 10.1109/COMST.2024.3422613
I. INTRODUCTION
A. Backg ound and Mo i a ion
THE FOURTH indus ial e olu ion, known as Indus y
4.0, [1] encompasses he digi al ans o ma ion o
he manu ac u ing sec o . I d i es he e olu ion owa ds
ully in e connec ed scena ios in he manu ac u ing indus y
be ween he physical and i ual wo lds o achie e g ea e
au oma ion and sma iza ion o he indus ial p ocesses. This
end ecei ed he e m o Sma Fac o ies; a ision o sma
p oduc ion sys ems and p oduc ion lines in e connec ed by
wi ed o wi eless means, and cha ac e ized by lexibili y,
econ igu abily, cos -e iciency, and agili y [2].I isalso
commonly e e ed o as Sma Manu ac u ing [3].
In his con ex , Indus y 4.0 emb aces a se ies o ech-
nologies and pa adigms p ima ily ocused a ound ubiqui ous
communica ion scena ios d i en by In e ne o Things (IoT)
and Indus ial In e ne o Things (IIoT), and a ound he
me ging o physical and i ual asse s h ough concep s like
Cybe -Physical Sys ems (CPS) and digi al wins [4]. Fu he ,
Indus y 4.0 sys ems can be empowe ed by ad anced ech-
nologies such as A i icial In elligence (AI), Machine Lea ning
(ML), Big Da a analy ics, and Augmen ed Reali y and Vi ual
Reali y (AR/VR) [5]. Al oge he , he Indus y 4.0 e olu ion is
d i ing he e olu ion o indus ial au oma ion in he manu ac-
u ing sec o [6]. Howe e , his echnological ecosys em has o
coexis wi h he eal- ime and de e minis ic la ency cons ain s
imposed by he use cases wi hin he indus ial au oma ion
ield including closed-loop con ol sys ems o p ocess and
ac o y au oma ion, connec ed obo ics, mo ion con ol o
machines and obo s, emo e ope a ion, e c. This landscape
implies complex communica ion scena ios wi h s ingen ye
he e ogeneous equi emen s ha demand a highly lexible,
adap able, and e icien ne wo k subs a e.
T adi ionally, he indus y has elied on p op ie a y ieldbus
p o ocols, such as PROFINET, E he ne /IP, Modbus, and
E he CAT, o achie e he equi ed le els o de e minism o
ime-sensi i e communica ions. Howe e , he lack o in e -
ope abili y and he closed na u e o hese solu ions make
hem e y limi ed in hei abili y o suppo mode n Indus y
4.0 applica ions [7]. To b idge his gap, he pas decade has
seen he eme gence o Time-Sensi i e Ne wo king (TSN),
which encompasses a se o open s anda ds, con inuously
being olled ou as pa o IEEE 802.1, o enable de e minis ic
c
2024 The Au ho s. This wo k is licensed unde a C ea i e Commons A ibu ion-NonComme cial-NoDe i a i es 4.0 License.
Fo mo e in o ma ion, see h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
260 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
communica ions o e E he ne . O e all, TSN has achie ed
e y signi ican p og ess owa ds a echnology ha is capa-
ble o con e ging mul iple ypes o a ic o e a sha ed
in as uc u e whils p o iding pe o mance gua an ees o he
mos c i ical ime-sensi i e applica ions. Fu he mo e, he open
and in e ope able na u e o TSN, along wi h i s e sa ile
con igu a ion capabili ies, g ea ly acili a es i s coexis ence
wi h ex e nal ne wo ks and i s in eg a ion wi h o he echnolo-
gies. This enablemen o a con e ged ne wo k in as uc u e
is key o mee ing he ubiqui ous connec i i y equi emen s
demanded by mode n Indus y 4.0 applica ions [8].
In pa allel, he i h gene a ion o communica ion ne wo ks
(5G) has eme ged wi h he aim o sa is ying a highly demand-
ing ye highly he e ogeneous se o equi emen s, which
include ul a-high bandwid h (enhanced Mobile B oadband;
eMBB), ul a-low la ency and eliabili y (Ul a-Reliable and
Low-La ency Communica ion; URLLC), and ul a-high con-
nec ion densi y (massi e Machine-Type Communica ions;
mMTC). A key di e ence wi h p e ious gene a ions is ha he
a ge is no almos exclusi ely human use s, bu also e ical
indus ies. In o de o be able o ailo 5G ne wo ks o his wide
ange o e ical use cases, 5G ne wo ks a e being designed o
suppo an unp eceden ed le el o lexibili y. As such, many
echnologies ha ha e eme ged in ecen yea s ha e become
key 5G building blocks [9], including Ne wo k Func ions
Vi ualiza ion (NFV), So wa e De ined Ne wo king (SDN),
ne wo k slicing, and Mul i-access Edge Compu ing (MEC).
These echnologies ep esen he end owa ds he i ual-
iza ion, so wa iza ion, and p og ammabili y o ne wo ks and
se ices, pa ing he pa h owa ds a ne wo king and compu ing
in as uc u e ha can be speci ically ailo ed o mee he
a ying equi emen s and demands o each indi idual e ical.
In o de o uly maximize he po en ial o Indus y 4.0
and beyond sys ems, bo h 5G echnology—including i s u u e
e olu ions—and TSN echnology a e expec ed o play a key
ole. In his way, 5G is no a eplacemen o TSN; ins ead,
hey a e complemen a y echnologies. On he one hand, 5G
enables wi eless scena ios ha a e no possible o e a wi ed
E he ne in as uc u e [10], whe he i is o enable mo e cos -
e icien and lexible deploymen s and educe main enance
cos s, o o in eg a e wi eless equipmen in o he p oduc ion
lines, such as mobile obo s, Au oma ed Guided Vehicles
(AGVs), and wi eless ield de ices. The URLLC capabili ies
de eloped in 5G ha e opened up he possibili y o ealiz-
ing hese wi eless communica ion scena ios main aining an
unp eceden ed le el o la ency and eliabili y gua an ees [11].
On he o he hand, TSN’s pu sui o ne wo k con e gence
is undamen ally aligned wi h 5G a chi ec u es; i ualiza ion
and edge compu ing pa adigms enable he coexis ence o
mixed-c i icali y applica ions in a sha ed ne wo k and compu e
in as uc u e, whe eas no el ne wo k o ches a ion ame-
wo ks [12] p esen a solu ion o he e e -inc easing complexi y
o managing and ope a ing such he e ogeneous ne wo ks.
The in e es in 5G and TSN by bo h indus y and academia
is e iden , ha ing p opelled ex ensi e s anda diza ion and
esea ch wo k alike. Besides, gi en he o eseen po en ial o
enabling de e minis ic low-la ency wi eless communica ions
o e icals such as indus ial au oma ion, he in eg a ion
and con luence o hese wo echnologies is also ecei ing
conside able a en ion. No ably, he 3 d Gene a ion Pa ne ship
P ojec (3GPP), he pa ne ship o S anda ds De elopmen
O ganiza ions (SDOs) cu en ly esponsible o he s anda d-
iza ion o 5G, has been s anda dizing he a chi ec u e and
mechanisms o enable he in eg a ion wi h TSN in i s la es
eleases. Toge he wi h hese s anda diza ion e o s, endea o s
ca ied ou by se e al o ganiza ions and indus ial alliances,
and ex ensi e esea ch wo k by academia, ha e demons a ed
signi ican b eak h oughs in ecen yea s, a end ha is
expec ed o con inue in esponse o he equi emen s o he
con inuously e ol ing e ical indus ies.
B. Pape O ganiza ion
The emainde o his pape is s uc u ed as ollows:
Sec ion II ca ego izes and examines he s a e o he a
ega ding su eys and u o ials in he scope o in eg a ion
be ween 5G and TSN echnologies, and ou lines he con i-
bu ions o his pape . Sec ion III e iews TSN echnology
acco ding o he main IEEE s anda ds, and con ex ualizes
TSN in he indus ial au oma ion and manu ac u ing e -
ical. Sec ion IV e iews he a chi ec u e, capabili ies, and
key echnology pilla s o 5G ne wo ks, emphasizing on he
enabling echnologies and s anda diza ion e o s o make
possible de e minis ic communica ions complian wi h TSN
o e 5G ne wo ks. Sec ion Vp o ides a comp ehensi e s a e-
o - he-a su ey o 5G and TSN in eg a ion, analyzing and
discussing li e a u e ha co esponds o he in eg a ion o
5G and TSN ne wo ks and o he in e wo king be ween 5G
echnology pilla s and TSN. Sec ion VI p esen s he mos
signi ican esea ch challenges and di ec ions ega ding he
opics co e ed in his su ey. Finally, a b ie conclusion o
he pape is p o ided in Sec ion VII. The s uc u e o co e
sec ions o his pape is ep esen ed in Figu e 1. A lis wi h
he main ac onyms used in his pape is p o ided in Table I.
II. RELATED WORK AND CONTRIBUTIONS
Bo h TSN and especially 5G ha e been ho oughly in es-
iga ed by he scien i ic communi y. Because o his, se e al
esea ch wo ks exis ha comp ehensi ely su ey ei he o
he wo echnologies, such as [13],[14], and [15]. Among
he ela ed su ey and/o u o ial wo ks ound in he li e a u e
in ol ing hese opics, Table II e lec s he co e age o hose
ha a e deemed o ha e he mos o e lap wi h ou wo k.
As pe he scope o his pape , he lis o esea ch pape s
included in Table II has been na owed down o he ones ha
include in e wo king aspec s be ween TSN and 5G, and/o
be ween TSN and 5G- ela ed echnologies. Table II hin s he
dep h a which a ious high-le el scopes a e co e ed by he
esea ch pape s, and whe he hey con ibu e wi h a e iew o
u o ial o echnologies and s anda ds, and/o wi h a su ey o
esea ch li e a u e. I should be no ed ha he p ima y scope
o some o he included esea ch pape s may be ou side o
he ex en o Table II and hus no indica ed, and ha some
pape s join ly ca y a heo e ical e iew and some s a e-o - he-
a p oposal, meaning ha hei con ibu ions may be discussed
in la e sec ions o his pape ins ead o , o , in addi ion o,
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 261
Fig. 1. O ganiza ion o he co e sec ions o his pape . O ange: e iew o echnologies and s anda ds; g een: su ey o academic li e a u e; blue: pe spec i es.
appea ing he e. The emainde o his sec ion expands on he
con ibu ions o hese su eys, and ou lines he con ibu ions
o ou wo k in compa ison o said s a e o he a and he gaps
encoun e ed.
A. Su ey and Tu o ial Wo ks Rela ed o he In eg a ion
be ween 5G and TSN Ne wo ks
A numbe o esea ch pape s add ess he in eg a ion
be ween TSN and 5G ne wo ks o di e en deg ees. Some
wo ks such as [19],[21],[23],[24], and [28] speci ically assess
he e olu ion o 5G s anda ds o suppo ime-sensi i e com-
munica ions. Among he i s pape s su eying his opic, [19]
and [21], desc ibe ini ial s anda diza ion ac i i ies by 3GPP
o suppo TSN se ices in 5G ne wo ks. Re e ences [23]
and [24] also analyze he 3GPP s anda ds in his con ex ,
in his case in a mode a ely mo e ma u e s a e. On he
o he hand, [28] comp ehensi ely e iews 5G URLLC ea u es
in oduced in he s anda ds, and analyzes how hey can con-
ibu e o suppo low la ency and high eliabili y in indus ial
au oma ion.
Con inuing his end, addi ional exis ing pape s desc ibe
di e en aspec s o he a chi ec u e-le el in eg a ion be ween
5G and TSN, possibly ocusing on speci ic aspec s such as
la ency, synch oniza ion accu acy, o con e ged con igu a ion.
Fo example, [22] p o ides a p elimina y analysis on he usage
and pe o mance o he p ocedu es and QoS mechanisms
a ailable in 5G o he suppo o ime-sensi i e communica-
ions. In his line, [34] ocuses on he con igu a ion aspec s
o TSN s eams, likewise p o iding an o e iew o he 5G
QoS mechanisms, and analyzing hei mapping o TSN a ic
classes. Re e ences [36] and [42] p o ide a concise o e iew
o he sys em in eg a ion be ween 5G and TSN.
In ega ds o [42], his pape also p esen s a b ie su ey
o implemen a ions o 5G-TSN in eg a ion in he li e a u e.
Re e ences [26] and [27] p o ide li e a u e e iew add ess-
ing u u e esea ch di ec ions o 5G and TSN, bu do no
in es iga e he in eg a ion o he wo echnologies. O e all,
comp ehensi e co e age o li e a u e con ibu ions on his
opic appea s o be la gely missing om he a ailable su ey
pape s. No ably, howe e , [47] illus a es an exhaus i e ca -
ego iza ion o esea ch wo ks in he scope o 5G and TSN
in eg a ion wi h he aim o quan i ying esea ch ends, bu
does no del e in o hei con ibu ions.
Re e ence [45] analyzes de e minis ic communica ions wi h
an eye on he upcoming 6G ne wo ks. This pape en isions
262 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
TABLE I
LIST OF ACRONYMS
he ole o se e al echnologies in he e olu ion owa ds
6G, including TSN, De Ne , URLLC, edge compu ing, and
also co e ing he applica ion laye and end- o-end secu i y.
Insigh ul akeaways can be ound in his a icle, bu due o i s
mo e gene alis na u e, de ailed TSN in eg a ion aspec s a e
spa se. O he wo ks such as [25],[31],[33],[38], and [48]
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 263
TABLE II
RELATED SURVEY AND TUTORIAL RESEARCH WORKS R: REVIEW OR TUTORIAL OF TECHNOLOGIES AND STANDARDS;S:SURVEY OF ACADEMIC
LITERATURE;BOLD NOTATION:SIGNIFICANTLY COVERED
also p o ide gene alis iews on he ma e , con ex ualizing
he opic o de e minis ic communica ions in he scope o
p i a e 5G deploymen s o he indus y e ical. Al hough
discussed, he in eg a ion be ween TSN and 5G is no he co e
con ibu ion o hese pape s, so i is no co e ed in ex ensi e
de ail.
The enablemen o end- o-end ime synch oniza ion o
suppo TSN o e 5G ne wo ks is an inhe en equi emen
o hei in eg a ion, so many wo ks, including some o he
a o emen ioned ones, add ess ime synch oniza ion suppo
oge he wi h o he a chi ec u e and con igu a ion aspec s.
The e a e, howe e , se e al a icles ha speci ically concen a e
on e iewing his opic: [16],[23],[24],[29],[35],[40],
and [46].[16] and [29] e alua e he challenges and equi emen s
o clock synch oniza ion accu acy o wi eless TSN, and
discuss how no el ad ances and capabili ies in wo wi eless
echnologies—Wi-Fi 802.11AX, and 5G—con ibu e o he
accu acy o he synch oniza ion achie ed. The o he pape s
a ge exclusi ely 5G sys ems. Re e ences [23] and [24] p o ide
a de ailed pe spec i e om 3GPP s anda ds in ega ds o
imesynch oniza ion osuppo ime-sensi i ecommunica ions.
Re e ences [35] and [40] ocus on su eying mechanisms o
enhance P opaga ion Delay Compensa ion (PDC), which can be
decisi e o he accu acy o he ime synch oniza ion dis ibu ed
h ough he wi eless 5G ne wo k. Finally, [46] discusses ecen
ad ances in 3GPP s anda ds in ega ds o ime synch oniza ion
esiliency in a mo e gene alis manne , and looks ahead o he
challenges o be sol ed in 6G.
Ano he scope o he in eg a ion be ween 5G and TSN
echnologies co esponds o he use o TSN wi hin he 5G
T anspo Ne wo k (TN). The ollowing a icles a e conside ed
o e iew his opic o di e en deg ees, o en wi hou being
he main con ibu ion o he pape : [11],[19],[20],[33],[36],
and [50]. Re e ence [20] does ocus exclusi ely on his ma e ,
b ie ly s udying he applicabili y o di e en TSN s anda ds
and ea u es o anspo on haul a ic wi h de e minis ic

264 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
equi emen s in 5G. Addi ionally, a b oade iew on he ma e
om an s anda ds pe spec i e is one o he con ibu ions
o [11]. Finally, [50] also concen a es on his opic, as i
p esen s an in e es ing pe spec i e on he oppo uni ies and
challenges o in eg a ing TSN in o he O-RAN a chi ec u e.
O e all, howe e , ewe wo ks wi h a su ey o u o ial na u e
can be ound in he li e a u e co e ing he use o TSN
in he 5G TN, in compa ison wi h hose ha analyze he
use o 5G o ex end a TSN ne wo k, co esponding o he
p e ious pa ag aphs o his subsec ion. This is especially ue
conce ning esea ch pape s published om 2020 onwa ds.
B. Su ey and Tu o ial Wo ks Rela ed o he In eg a ion o
TSN Wi h Key 5G Enabling Technologies
On he con a y, despi e he inc easing ecogni ion o he
bene i s o enabling ime-sensi i e communica ions wi hin he
amewo k o echnologies such as SDN and i ualiza ion,
pu ely su ey o u o ial wo ks conce ning he landscape and
s a e o he a o he use o hese echnologies oge he
wi h TSN appea o be highly sca ce. TSN oge he wi h
SDN is add essed by [17],[18],[27], and [32]. Re e ences
[17] and [27] s udy he applicabili y o TSN and SDN o
ne wo ks in Indus y 4.0 o in e ical indus ies, bu discuss
he con ibu ions o each o he wo echnologies indepen-
den ly om each o he , as opposed o hei in eg a ion. The
p ima y con ibu ions in his ega d a e p obably [18] and [32].
The o me e iews challenges and esea ch di ec ions o
so wa e-de ined ime-sensi i e ne wo ks o be applied o he
ope a ion and managemen o CPSs. Re e ence [32] also
ske ches a p elimina y a chi ec u e, so i is co e ed in a la e
sec ion o his pape . In addi ion, [11] p esen s a b ie li e a u e
e iew su eying he adop ion o he SDN pa adigm o TSN.
Howe e , o e all, he co e age o s a e o he a conce ning
he usage o TSN and SDN oge he is limi ed and no
comple ely up- o-da e. Finally, a ela ed pape wo h poin ing
ou is [51], which s udies ime synch oniza ion mechanisms
ha can be bo owed o he coo dina ed managemen o SDN
ne wo ks.
In ega ds o in eg a ing TSN and i ualiza ion, [43] is
cen e ed on e iewing he a ailable capabili ies o enable
a cloudi ica ion o speci ic TSN unc ionali y o enhance
eliabili y. Re e ence [52] p o ides a echnology and li e a u e
e iew owa ds eal- ime con aine -based i ualiza ion. I does
no di ec ly add ess TSN echnology, so i has been le ou
o Table II. E en hough [49]’s p ima y con ibu ion is o
p esen expe imen a ion esul s, i b ie ly e iews challenges
and exis ing mechanisms o TSN suppo in con aine -based
applica ions. O e all, no esea ch wo k wi h a su ey na u e
has been ound o widely in es iga e app oaches o combine
TSN and i ualiza ion, e en hough he opic is gaining
ema kable ac ion in he li e a u e in o de o suppo he
equi emen s o new echnological ends such as i ualized
con olle unc ionali y in CPSs and in indus ial au oma ion
ne wo ks, and digi al wins.
Re e ence [39] add esses he p og ess owa ds ze o- ouch
cogni i e managemen o indus ial ne wo ks. The use o TSN
is conside ed only e y b ie ly and no di ec ly add essed.
Re e ences [19] and [33] desc ibe analy ics exposu e o
applica ions in 5G ne wo ks, and ze o- ouch p ac ices o he
managemen o p i a e indus ial 5G ne wo ks, espec i ely.
Bo h o hese pape s ha e a gene alis na u e and also
in oduce he suppo o TSN in 5G ne wo ks. Howe e , hese
opics a e only p esen ed as independen capabili ies o 5G
ne wo ks a he han del ing in o a mo e holis ic pe spec i e.
O e all, i has been ound ha he applicabili y o no el
in elligen , ze o- ouch, and analy ics-powe ed o ches a ion
app oaches and echnologies om he amewo k o 5G o he
managemen o complex 5G-TSN ne wo ks is ba ely p esen
in su ey con ibu ions o he li e a u e, e en hough i could
be a key ma e in he e olu ion owa ds 6G.
C. Addi ional Rema ks
Among he esea ch pape s in Table II,[11],[19],[20],
[26],[29],[32],[35],[36],[40],[43],[45], and [47] do no
ocus on any speci ic e ical use case. Meanwhile, he es
a ge he indus ial au oma ion and manu ac u ing e ical
like ou wo k does. S ill, while he a ge ed en i onmen may
ha e implica ions, he echnology and a chi ec u e aspec s a e
la gely applica ion scena io-agnos ic.
I should also be no ed ha due o ex ension easons, gi en
he b oad and o e a ching scope o he opics a hand, he
ollowing ela ed aspec s a e na owly co e ed, o no co e ed,
by his pape . Example a icles ha p ima ily co e such opics
a e included in he lis below o he eade ’s e e ence.
•Taxonomy o applica ions and a chi ec u es in Indus y
4.0 and in indus ial ne wo ks [4],[53],[54].
•Modeling and e alua ion o TSN mechanisms and algo-
i hms whe e 5G in eg a ion does no in e ene [55],[56],
[57],[58],[59].
•Secu i y aspec s o TSN and o 5G-TSN in eg a ion [60],
[61].
•In eg a ion o TSN wi h wi eless access ne wo k ech-
nologies o he han 5G [62],[63],[64].
•P o ocol and a chi ec u e aspec s ha speci ically a ge
use cases no ela ed o indus ial au oma ion in manu-
ac u ing [65],[66].
•P i a e 5G ne wo k a chi ec u es o indus ial en i on-
men s [67] [33].
•De e minis ic Ne wo king (De Ne ) [68],[69].
D. Con ibu ions o This Su ey
The ou line o he con ibu ions o his pape is desc ibed
as ollows:
•An in-dep h e iew, om he pe spec i e o SDOs and
indus ial o ganiza ions, o he echnologies and ne wo k
a chi ec u es ha a e shaping a new gene a ion o inno-
a i e indus ial applica ions, ocusing on TSN s anda ds,
on he 5G echnology ecosys em, and on he enabling
echnologies o hei in eg a ion.
•A comp ehensi e s a e-o - he-a su ey o academic
pape s on he in e wo king be ween TSN and key pilla s
o 5G ne wo ks o e he pe iod be ween 2016 and
Sep embe 2023. This includes con ibu ions in he ma -
e s o SDN and ne wo k p og ammabili y, i ualiza ion
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 265
echnologies and amewo ks, edge compu ing a chi ec-
u es, end- o-end o ches a ion, ze o- ouch au oma ion,
and ne wo k capabili y exposu e.
•A comp ehensi e s a e-o - he-a su ey o academic
pape s on he in eg a ion be ween 5G ne wo ks and
TSN o suppo end- o-end ime-sensi i e communica-
ions o e wi ed and wi eless in as uc u es, spanning he
pe iod be ween 2018 and Sep embe 2023. This includes
con ibu ions in he a eas o : pe o mance e alua ion
and op imiza ion o 5G-TSN in eg a ed a chi ec u es,
enablemen o ime synch oniza ion o e 5G, con e ged
con igu a ion o 5G and TSN domains, and TSN o he
5G TN.
•An ou line o he o e all ision and he challenges ahead
o end- o-end communica ions ha in eg a e 5G and TSN
echnologies, based on syn hesized insigh s and lessons
lea ned de i ed om he li e a u e su ey.
To he bes o ou knowledge, acco ding o he ela ed wo k
explo ed in his sec ion, his is he i s wo k p o iding a
holis ic comp ehensi e e iew o he in eg a ion and con e -
gence o TSN wi h he 5G a chi ec u e as well as wi h a
wide spec um o enabling echnologies in he 5G ecosys em.
Mo eo e , we conside ha no exis ing wo k su eys he s a e
o he a o such opics in an exhaus i e manne and om
a TSN echnology-cen ic pe spec i e, despi e he spike in
in e es and in he numbe o esea ch wo ks unde gone by he
scien i ic communi y since he la e pa o he las decade.
III. TIME-SENSITIVE NETWORKING AND ITS ROLE IN
INDUSTRIAL COMMUNICATIONS
This sec ion p o ides a comp ehensi e e iew o
he TSN ea u es om a s anda ds pe spec i e om
Sec ions III-A–III-E. The emaining Sec ions III-F–III-H
in oduce he basic a chi ec u e o indus ial ne wo ks and
he communica ion pa e ns in ol ed in such ne wo ks. Key
con ibu ions in he ma e om SDOs and o he en i ies
a e examined in o de o analyze how TSN echnology can
be implemen ed and con igu ed o sa is y he he e ogeneous
communica ion equi emen s p esen in hese scena ios.
A. TSNO e iew
TSN s anda ds a e de eloped by he Time-Sensi i e
Ne wo king Task G oup, which is one o he ou subg oups
o he IEEE 802.1 Wo king G oup [70]. TSN has i s o igins in
2004, when he IEEE 802.3 Residen ial E he ne S udy G oup
was es ablished. This S udy G oup was a yea la e mo ed
o he IEEE 802.1 Audio Video B idging (AVB) Task G oup,
be o e being enamed in 2012 o he cu en Time-Sensi i e
Ne wo king Task G oup o be e e lec an scope ha goes
beyond he o iginally concei ed audio and ideo ansmission
by including con ol applica ions in indus y and au omo i e
a eas [71].
TSN communica ion occu s be ween end s a ions (Talke s
and Lis ene s) ac oss a laye 2 ne wo k o TSN-capable
de ices, gene ally e e ed o as TSN b idges. Each commu-
nica ion channel be ween a Talke and one o mo e Lis ene s
is known as a TSN s eam o TSN low, which is associa ed
o a gi en TSN a ic class cha ac e ized by a se o ea u es
(pe iodici y, payload size, e c.) and Quali y o Se ice (QoS)
p ope ies (la ency, bandwid h, ole ance o loss, e c.). T a ic
classes a e assigned o queues acco ding o he P io i y Code
Poin (PCP) in he VLAN ag wi hin he E he ne heade . As
he PCP ield has a leng h o 3 bi s in he VLAN heade , his
enables he dis inc ion o up o 8 di e en a ic classes o
p io i ies. The me e use o PCP o p io i ize he queuing o
ames is e e ed o as s ic p io i y. Howe e , TSN s anda ds
de ine addi ional mechanisms o he ea men o E he ne
a ic. In many cases, a TSN s eam is implemen ed h ough
he assignmen o a pe iodic ime window in which he a ic
belonging o said s eam is allowed o be ansmi ed ac oss he
TSN ne wo king de ices. The end s a ions epo hei s eam
equi emen s o he TSN con ol plane, which akes ca e o
compu ing he global schedule ha accommoda es all s eams,
and o con igu ing he b idges in ol ed in he communica ion
pa h acco dingly.
TSN s anda ds can be p ima ily g ouped in ou ca ego ies
acco ding o he ea u es hey p o ide: iming and synch o-
niza ion, bounded low la ency, high eliabili y, and esou ce
managemen . Timing and synch oniza ion s anda ds p o ide
he means o es ablish a common sou ce o ime be ween
all en i ies in ol ed in a TSN communica ion, in o de o
pa e a way o con igu e consis en a ic scheduling policies.
Bounded low la ency s anda ds p o ide a ic shaping mech-
anisms, o en le e aging ime synch oniza ion, ha allow o
apply a di e en ea men o di e en a ic s eams acco d-
ing o hei cha ac e is ics and equi emen s. This ensu es ha
he deli e y o c i ical ames is ca ied wi hin a bounded
imespan. Besides, p eemp ion mechanisms included in his
ca ego y o s anda ds can u he educe in e e ence. High
eliabili y s anda ds include pa h edundancy mechanisms o
mi iga e he e ec s o ne wo k ailu es o subop imal ne wo k
condi ions, and policy en o cemen mechanisms o p o ec
he ne wo k om non-complian a ic. This g oup o s an-
da ds a ge high a ailabili y, ul a- eliabili y, and ame loss
a oidance o c i ical a ic. Finally, esou ce managemen
s anda ds deal wi h he mechanisms and in e aces ha make i
possible o alloca e TSN esou ces o a ic s eams and push
he necessa y con igu a ion on o he in ol ed en i ies. TSN
has s anda dized bo h dis ibu ed and cen alized models.
As TSN p o ides a oolbox o s anda ds, he usage o each
o hese indi idual ea u es in a speci ic applica ion scena io
can be selec ed based on i s cha ac e is ics and i s equi e-
men s [72]. Among he base se o TSN s anda ds, IEEE
S d 802.1Q-2022 [73] s anda dizes mul iple aspec s ega ding
b idge o wa ding, scheduling, and con igu a ion. P e ious
e isions o 802.1Q, such as IEEE S d 802.1Q-2018 [74] and
IEEE S d 802.1Q-2014 [75], ha e been p og essi ely olled
in o i . IEEE S d 802.1Q-2022 has had some amendmen s
co esponding o enhancemen s o speci ic ea u es since i s
elease. O he TSN base s anda ds include, bu a e no limi ed
o, 802.1AS in ega ds o ime synch oniza ion, and 802.1CB
p o iding ame eplica ion mechanisms.
A la ge, TSN de ines h ee main a ic classes in he
802.1Q s anda d. Bes E o (BE) is used on low-p io i y
a ic ha does no equi e any o m o a ic shaping and has
266 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
TABLE III
CLASSIFICATION OF MAIN TSN STANDARDS
no ime-based cons ain s. Nex , S eam Rese a ion (SR) is
used on a ic wi h in e media e p io i y. SR was in oduced
by AVB s anda ds, and is enabled by c edi -based shaping
mechanisms. SR subclasses a e also de ined as SR Class A,
SR Class B, e c., om highe o lowe p io i y. Howe e ,
e en hough a ic shaping based on c edi can be adequa e
o audio and ideo a ic, TSN usage in indus ial con ol
and au omo i e scena ios equi e addi ional mechanisms o
sa is y he equi emen s o pe iodic applica ions ha demand
a ic o be ansmi ed acco ding o a ime schedule [76].
As such, as ime-based shaping and ime synch oniza ion
mechanisms we e s anda dized in TSN, a new a ic class
wi h he highes p io i y, Scheduled T a ic (ST), was also
in oduced. ST p o ides he leas in e e ence and he g ea es
empo al gua an ees in he deli e y o ames, as sepa a e
s eams in he ST class can be assigned isola ed ime windows
o ansmission.
Subsequen subsec ions co e he mos signi ican IEEE
802 TSN s anda ds and hei ea u es classi ied in o he
ou a o emen ioned ca ego ies. Table III classi ies he main
s anda ds in o de o appea ance in his sec ion. The eade
in e es ed in lea ning mo e abou a speci ic s anda d is e e ed
o he e e enced TSN s anda d documen s published by IEEE.
B. Timing and Synch oniza ion S anda ds
Time synch oniza ion is c ucial o enabling de e minis ic
communica ions and hence is a p e equisi e o he TSN mech-
anisms ha enable ime-sensi i e a ic shaping. TSN ime
synch oniza ion is based on he IEEE 1588 P ecision Time
P o ocol (PTP) s anda d, which is a p o ocol o p o iding
p ecise clock synch oniza ion in he sub-mic osecond ange,
p ima ily a ge ing E he ne -based LANs. The o iginal PTP
s anda d da es om 2002, wi h he mos ecen e ision, IEEE
1588-2019, o en e e ed o as PTP 2.1, being eleased in
2019 [77].
A PTP domain is composed o se e al in e connec ed PTP-
enabled nodes, o ming a PTP ne wo k. Each o hese nodes
is known as a PTP clock. PTP clocks a e cha ac e ized by
ha ing one o mo e PTP-enabled po s h ough which he ime
synch oniza ion in o ma ion is p opaga ed. Each indi idual
po o a PTP clock can be in h ee possible s a es a any gi en
ime, which a e Mas e Po ,Sla e Po , and Passi e Po .
A Passi e Po is simply a po ha is no PTP-enabled. On
he o he hand, ime synch oniza ion in o ma ion consis s on
imes amped PTP messages ha a e ansmi ed be ween he
Mas e Po o a clock and he Sla e Po o ano he clock, in a
p ocedu e ha allows he la e clock o become synch onized
o he o me .
The IEEE 1588 s anda d de ines se e al ypes o clocks
acco ding o he unc ionali y hey implemen and o he po s
hey con ain. A he op o he hie a chy o a PTP domain,
aG and Mas e (GM) clock is elec ed, ei he manually,
o h ough nego ia ion by execu ing he Bes Mas e Clock
(BMC) algo i hm. The BMC algo i hm compa es he accu acy
o wo gi en clock sou ces o de e mine he mos accu a e one
acco ding o se e al a ibu es. The BMC algo i hm de e mines
he GM his way, and hen cons uc s a ime-synch oniza ion
spanning ee. The GM clock ac s as he oo o he ee
and p o ides he p ima y ime sou ce in he PTP domain,
pe iodically ini ia ing he synch oniza ion p ocedu e owa ds
he downs eam clocks h ough one o mo e Mas e Po s.
De ices wi h one PTP-enabled po in a domain a e known
as O dina y Clocks, and hose wi h mul iple PTP-enabled
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 267
po s in a domain a e known as Bounda y Clocks. The single
Sla e Po o an O dina y Clock allows i o synch onize i s
clock om an ups eam clock. Meanwhile, a Bounda y Clock
has one Sla e Po connec ed o an ups eam Mas e Clock
om which i ecei es he clock e e ence, and one o mo e
Mas e Po s h ough which i p o ides i s clock e e ence o
downs eam clocks. Finally, a T anspa en Clock co esponds
o a de ice ha nei he p o ides a clock sou ce no upda es
hei own, bu is s ill capable o elaying PTP messages while
applying a co ec ion o hei ime e e ence (i.e., upda ing he
co ec ionField ield) o compensa e o he ansi ime in he
de ice.
The PTP s anda d suppo s cus omiza ion h ough he de i-
ni ion o p o iles. The TSN s anda d o ime synch oniza ion
is IEEE 802.1AS, known as gene alized P ecision Time
P o ocol (gPTP), which is p ecisely a p o ile o he PTP
s anda d wi h adap a ions and simpli ica ions o i s use in
ime-sensi i e applica ions o e IEEE 802 laye 2 ne wo ks.
The la es e ision o he 802.1AS s anda d is 802.1AS-
2020 [78], published in 2020. 802.1AS s anda dises a PTP
Relay Ins ance, co esponding o a TSN b idge ac ing as a
Bounda y Clock, and a PTP End Ins ance, co esponding o
a TSN end s a ion ac ing as an O dina y Clock. Acco ding
o 802.1AS, a g oup o in e connec ed gPTP-capable de ices
o ms a ime-awa e ne wo k. These de ices a e e e ed o
as ime-awa e sys ems. A ime-awa e sys em can suppo
mul iple PTP ins ances and hus can be pa o mul iple gPTP
synch oniza ion domains.
The 802.1AS s anda d p o ides addi ional obus ness,
demanding each node o suppo gPTP in he MAC le el.
802.1AS also in oduces eliabili y imp o emen s, such as he
abili y o ha e mul iple gPTP ime domains o allow nodes
o be synch onized o mul iple GM clocks, as well as he
possibili y o make mul iple edundan connec ions o hese
GM clocks. Fo u he insigh in o he di e ences be ween
PTP and gPTP, he eade is e e ed o Clause 7.5 o 802.1AS.
Time synch oniza ion in a gPTP domain is achie ed h ough
he execu ion o wo sepa a e messaging p ocedu es be ween
each pai o PTP ins ances sepa a ed by one hop. The i s
p ocedu e consis s on measu ing he link p opaga ion delay
be ween wo ins ances, and is pe o med by exchanging imes-
amped messages as depic ed in he example o Figu e 2.The
ini ia o o he exchange e en ually collec s ou imes amps
ha enable i o calcula e he link delay as shown in he igu e.
I he equencies o he clocks o each ins ance di e , an
addi ional Ra eRa io ac o is aken in o accoun , calcula ed as
he a io be ween he clock equency o he ini ia o ela i e
o he clock equency o he esponde . I can be quan i ied
by execu ing he p ocedu e mul iple imes.
Once he p opaga ion delay o each link has been calcula ed,
clock synch oniza ion can ake place. Sync—and Follow_Up
i ope a ing in wo-s ep mode—messages pe iodically sen by
he GM clock a e ecei ed on Sla e Po s and o wa ded ia
Mas e Po s as depic ed in Figu e 3. Be o e a de ice o wa ds
synch oniza ion messages o he nex de ice his way, i
upda es i s Co ec ionField ield by adding o i he calcula ed
p opaga ion delay o he link as well as he esidence ime in
he de ice. The esidence ime is calcula ed as he di e ence
Fig. 2. Example o calcula ion o Ra eRa io and link delay be ween wo
clocks. The alues a e calcula ed by he ini ia o and he delay alue is
de e mined ela i e o i s clock equency. Symme ic link and cons an link
delay is assumed.
be ween a imes amp a he eg ess o he de ice and a
imes amp a he ing ess. The de ice also ecalcula es he
message’s Cumula i eRa eRa io ield, which is he p oduc o
Ra eRa io ac o s du ing he ansi ac oss all de ices.
C. High Reliabili y S anda ds
The wo p ima y s anda ds conce ning he p o ision o
eliabili y gua an ees in TSN a e 802.1CB and 802.1Qci.
The 802.1CB-2017 s anda d [79] de ines he F ame
Replica ion o Elimina ion and Reliabili y (FRER) p o ocol.
FRER is in ended o p o ide edundancy o c i ical a ic
by p o iding mechanisms o eplica e he ames o a s eam
in o mul iple membe s eams ha can be sen h ough pa hs
wi h he leas possible o e lap. P ecisely, he execu ion o an
op imal pa h selec ion s a egy o ansmi he s eams is key
o maximizing he e ec i eness o FRER when ully disjoin
pa hs a e no a ailable [57].
FRER also ensu es ha he edundan duplica e ames a e
elimina ed in he des ina ion o in in e sec ion poin s o he
274 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
TABLE IV
CHARACTERISTICS OF TRAFFIC TYPES FOR INDUSTRIAL COMMUNICATIONS
TABLE V
CHARACTERISTICS OF TRAFFIC TYPE CATEGORIES IN INDUSTRIAL AUTOMATION SYSTEMS
IIC [84] conside s ha a ic scheduling based on ime
synch oniza ion may s ill be used o cyclic a ic o gua an ee
bandwid h and bounded la ency, gi en ha he p esence o
isoch onous a ic al eady demands i .
These a ic ypes ha ep esen indus ial au oma ion
communica ions can be mapped o TSN equi emen s and
con igu a ion. The mappings be ween a ic ypes and a ic
classes (i.e., he a ic p io i ies ma ked in he PCP ield o
he VLAN ag) p oposed by hese sou ces, also compa ed
o he lis o a ic ypes s anda dized in 802.1Q (Annex
I), is p esen ed in Table VII. On he o he hand, Table VIII
shows he mapping be ween each a ic ype and he usage
equi emen s o he p ima y TSN s anda ds.
As e lec ed in Table VIII, 802.1Qb along wi h 802.1AS
a e used o o e dedica ed cycle imes o he a ic ypes
wi h he mos c i ical equi emen s in e ms o bounded low
la ency. Mo eo e , ime-based 802.1Qci gua an ees ha no
in e e ing ames occupy hei ese ed cycle imes. 802.1Qbu
and 802.3Qb could also be used in complemen a y ashion o
802.1Qb as discussed in Sec ion III-D. On he o he hand,
asynch onous and spo adic a ic classes end o ha e mo e
loose equi emen s, elying p ima ily on ame p eemp ion
om 802.1Qbu and 802.3Qb and on a e-based TSN mech-
anisms ha do no equi e ime synch oniza ion. The mos
c i ical a ic ou o hese a ic classes, such as Cyclic-
Asynch onous and E en s: Con ol, shall be con igu ed as
exp ess a ic. Bandwid h ese a ion based on 802.1Qa can
be applied mainly o ideo and oice a ic classes ha ha e
he lowes la ency cons ain s, o p o ide hem wi h bandwid h
gua an ees o e bes e o a ic. As a as edundancy is
conce ned, he usage o ame eplica ion ia 802.1CB is
p ima ily conside ed acco ding o he ne wo k opology as
well as o he c i icali y and loss ole ance o he applica ions.
Finally, 802.1Qcc applies o mos indus ial communica ion
a ic classes, as i p o ides he mechanisms o con igu e
ea u es om 802.1Qa , 802.1Qb , 802.1Qci, and 802.1CB,
ac oss he TSN domain.
IV. THE 5G TECHNOLOGICAL LANDSCAPE IN THE
CONTEXT OF TIME-SENSITIVE NETWORKING
The cu ing-edge capabili ies in oduced by 5G ne wo ks
ha e caused a pa adigm shi in how e icals can le e age he
ne wo k o enable inno a i e and g oundb eaking applica ions.
To add ess he exponen ial inc ease o demands aised by hese
he e ogeneous e ical applica ions, an unp eceden ed le el o
lexibili y, scalabili y, and agili y is equi ed om he ne wo k
and om he se ice p o isioning mechanisms. This sec-
ion i s e iews, in Sec ions IV-A and IV-B, 5G mechanisms
o suppo de e minis ic and eliable communica ions o e
he wi eless medium, as well as p ocedu es o acili a e he
in e ope a ion be ween 5G ne wo ks and TSN ne wo ks. Then,
Sec ions IV-C–IV-E e iew key echnologies and amewo ks
ha ha e d i en he e olu ion owa ds ne wo k a chi ec u es
ha can accommoda e he a o emen ioned he e ogenei y and
complexi y, oge he wi h he exis ing echnological suppo
o adop hese a chi ec u es and pa adigms in TSN ne wo ks.
A. An Open and P og ammable Se ice-Based 5G
A chi ec u e Tha Mee s S ingen Requi emen s
The necessi y o sa is y an e e -g owing se o di e se
communica ion equi emen s highligh ed se e al limi a ions o

SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 275
TABLE VI
CLASSIFICATION OF TRAFFIC TYPES FOR INDUSTRIAL COMMUNICATIONS
TABLE VII
TSN TRAFFIC CLASSES FOR INDUSTRIAL COMMUNICATION TRAFFIC TYPES
TABLE VIII
APPLICABILITY OF TSN STANDARDS TO INDUSTRIAL COMMUNICATION TRAFFIC TYPES
he monoli hic, s a ic, and p op ie a y black box app oach o
he 4G ne wo k a chi ec u e. The in lexibili y o 4G ne wo ks
c ucially limi s hei scalabili y and hei abili y o p o ide
c i ical capabili ies o no el use cases [122], and leads o
highe CAPEX/OPEX cos s and highe complexi y [123].
Release 15 o he 3 d Gene a ion Pa ne ship P ojec (3GPP)
276 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
TABLE IX
OVERVIEW OF MAIN 5GC NFS
ma ks he ansi ion be ween 4G and 5G. As such, his
3GPP elease s anda dizes he i s e sion o he 5G Sys em
(5GS), de ined as he “3GPP sys em consis ing o 5G Access
Ne wo k (5G-AN), 5G Co e (5GC) Ne wo k and UE” [124].
5G e olu ion has con inued wi h he s anda diza ion o Release
16 and Release 17, and i s cu en ly in i s e olu ion owa ds
5G Ad anced wi h he s anda diza ion o Release 18.
In o de o o e come he limi a ions o 4G, 5G s anda d-
iza ion e o s ha e been execu ed s a ing om he na i e
adop ion o so wa iza ion, cloudi ica ion, and disagg ega ion
o he unc ionali ies o he 5GS a chi ec u e. The i s
s ep owa ds disagg ega ion s a ed in 3GPP Release 14
wi h he in oduc ion o Con ol and Use Plane Sepa a ion
(CUPS), which enables mo e lexible deploymen s by physi-
cally decoupling o con ol and use plane unc ionali y in he
Co e Ne wo k. 5G ne wo ks, howe e , ha e ully e amped
he a chi ec u e h ough he de ini ion o he Se ice-Based
A chi ec u e (SBA). The 5GC SBA is buil upon he design
p inciples o modula i y and open in e aces, a ge ing se ice
agili y and adap abili y. Eme ging echnological ends such
as NFV, SDN, cloud-na i e, and in elligen end- o-end o ches-
a ion a e a pe ec i o he SBA a chi ec u e. Wi h SBA,
he 5GC is implemen ed in he o m o lexibly deployable
and con igu able Ne wo k Func ions (NFs) ha in e ac ia
se ice-based in e aces. An o e iew o he p ima y NFs ha
make up he 5GC since Release 15 is p o ided in Table IX.
The 5G-AN also ollows simila ends o pu suing open-
ness and le e aging i ualiza ion echnologies. Building upon
he concep o Cloud-RAN (C-RAN) in oduced in 4G
ne wo ks o decouple base band p ocessing unc ions ca ied
by a cen alized pool o Base Band Uni s (BBUs) om adio
unc ions ca ied by he Remo e Radio Heads (RRHs)—known
as Radio Uni s (RUs) in 5G—, he 5G-AN has in oduced
a unc ional spli in he BBUs. This unc ional spli enables
he eloca ion o lowe laye BBU unc ions o he an enna
si e, elie ing he on haul link. As such, a BBU wi hin he
5G base s a ion (gNodeB, gNB) can be u he spli in o a
Dis ibu ed Uni (DU) a he an enna si e, and a Cen alized
Uni (CU), which handle lowe and highe laye s o he 5G
New Radio (5G NR) p o ocol s ack, espec i ely. A CU can be
connec ed o a pool o mul iple DUs and hus suppo mul iple
gNBs. O e all, his imp o es he lexibili y and cos -e iciency
o Radio Access Ne wo k (RAN) deploymen s, and, in his
line, acili a es he i ualiza ion o he RAN in an a chi ec u e
known as Vi ual RAN ( RAN). Based on RAN and seeking
an open, in e ope able, and in elligen mobile RAN, he O-
RAN Alliance [125] is wo king on he de elopmen o he
O-RAN a chi ec u e, which is also s eadily gaining indus y
adop ion.
The alloca ion o RAN esou ces o ansmi ing da a (i.e.,
scheduling) in 5G en ails challenges inhe en o he wi eless
medium gi en i s ime- a ian ansmission cha ac e is ics,
which has implica ions on he p opaga ion la ency and ji e
and on he eliabili y o he ansmissions. URLLC ea u es
in oduced in he 5G NR speci ica ions p o ide la ency and
eliabili y enhancemen s o 5G communica ions in o de o
suppo new e ical applica ions wi h s ingen equi emen s,
such as in indus ial au oma ion [126]. In o he wo ds, URLLC
capabili ies o 5G ne wo ks con ibu e o he suppo o wi e-
less da a ansmission wi h ul a-low bounded deli e y imes
and wi h ul a-high eliabili y gua an ees. Hence, URLLC
capabili ies a e syne gis ic wi h he capabili ies ha TSN has
been concei ed o p o ide o wi ed, E he ne -based commu-
nica ions, and a e complemen a y o he sys em a chi ec u e
aspec s la e e iewed in Sec ion IV-B o in eg a ing 5G and
TSN ne wo ks.
5G NR uses he O hogonal F equency-Di ision Mul iple
Access (OFDMA) modula ion scheme. The ansmission
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 277
esou ces a e quan ized in o Resou ce Blocks (RBs). In he
equency domain, each RB consis s o a o al o 12 o hogonal
sub-ca ie equencies, allowing simul aneous ansmissions
om di e en use s on di e en sub-ca ie s. In he ime
domain, RBs a e di ided in o mul iple ime slo s called
T ansmi Time In e als (TTIs). In he de aul ansmission
scheme, each TTI uses 14 O hogonal F equency-Di ision
Mul iplexing (OFDM) symbols. The leng h o a TTI depends
on he numbe o symbols pe TTI and on he du a ion o such
OFDM symbols.
Flexible Sub-Ca ie Spacing (SCS) and mini-slo schedul-
ing a e wo signi ican Release 15 URLLC capabili ies ha
a ge la ency imp o emen s. While ixed a 15 kHz in 4G, he
SCS can exis a 15, 30, 60, 120, and 240 kHz in 5G. This
ea u e can educe ansmission la ency hanks o dec easing
OFDM symbol du a ion. Addi ionally, while in 4G he de aul
slo -based scheduling o Physical Downlink Sha ed Channel
(PDSCH) and Physical Uplink Sha ed Channel (PUSCH) uses
a sepa a ion o 14 OFDM symbols, 5G suppo s scheduling
packe s o ansmission in mini-slo s ha use 2, 4, o
7 symbols. This is also known as sub-slo -based ansmission.
Combining lexible SCS wi h sub-slo scheduling, ansmis-
sion in e als can be scaled be ween 8.93 µs and 1 ms, which
allows o mo e s ingen and g anula pe iodici ies in he
scheduling o TSN a ic s eams [127].
Semi-s a ic scheduling based on Con igu ed G an (CG) o
uplink, o on Semi-Pe sis en Scheduling (SPS) o downlink
o o uplink, can educe ansmission la ency. By de aul ,
each Dynamic G an (DG) uplink ansmission equi es a
Scheduling Reques om he UE and a g an om he
gNB i s , which accumula es la ency. Suppo o CG uplink
ansmissions o pe iodic a ic a e in oduced in 5G, elimi-
na ing he need o send Scheduling Reques s by p ealloca ing
esou ces o he UE ins ead. Release 16 enhances CG in
uplink and SPS in downlink wi h suppo o con igu ing
mul iple CG and SPS esou ces wi h di e en scheduling
pa ame e s and pe iodici ies, which is key o suppo ing
cyclic indus ial con ol a ic lows wi h he e ogeneous cha -
ac e is ics [28].
O he Release 16 URLLC ea u es conce ned wi h imp o -
ing la ency a e: PUSCH enhancemen s ha allow a CG o DG
can schedule mul iple PUSCH epe i ions in consecu i e slo s,
which elimina es he ime gap be ween epe i ions; sub-slo -
based Hyb id Au oma ic Repea Reques acknowledgemen
(HARQ-ACK) eedback suppo ing mo e han one Physical
Uplink Con ol Channel (PUCCH) ca ying HARQ-ACK
wi hin a slo ; enhancemen s o Logical Channel P io i iza ion
(LCP) es ic ion, allowing o es ic he mapping o logical
channels o speci ic CG con igu a ions; and in a-UE p i-
o i iza ion ha enables a UE o d op an al eady-scheduled
ansmission whose esou ces o e lap in ime wi h a mo e
ime-c i ical low.
In ega ds o URLLC capabili ies ha a ge eliabili y
imp o emen s ia P o ocol Da a Uni (PDU) edundancy,
Release 16 includes Packe Da a Con e gence P o ocol
(PDCP) Packe Duplica ion and highe -laye mul i connec-
i i y. Wi h PDCP packe duplica ion, PDUs a e duplica ed
a he PDCP laye and can be submi ed ia independen
ansmission pa hs. To enable his, addi ional Radio Link
Con ol (RLC) en i ies a e added o he adio bea e , leading
o one p ima y logical channel and mul iple seconda y ones.
A e he ansmission o a PDCP PDU is acknowledged, o h-
e s a e disca ded. Highe -laye mul i connec i i y consis s on
duplica ing ansmissions o e disjoin use plane pa hs. These
edundan pa hs can be se up by ha ing he UE es ablish wo
PDU sessions using Dual Connec i i y, o ia a single PDU
session ha uses edundan unnels be ween he RAN and he
UPF o be ween he RAN and wo In e media e UPFs. The
abili y o se up edundan ansmission pa hs can be le e aged
o implemen TSN ea u es, pa icula ly FRER [128].
O he URLLC capabili ies o enhanced eliabili y include
Release 16 imp o emen s o Mul iple T ansmission and
Recep ion Poin (Mul i-TRP) echnology, which can enhance
Signal o In e e ence plus Noise Ra io (SINR) in p esence o
in e e ence o poo channel condi ions h ough join ansmis-
sions and ecep ions. In Release 15, a new Channel Quali y
Indica o (CQI) able wi h a a ge Block E o Ra e (BLER)
o 10−5 o channel s a e epo ing, and a new 64-QAM
Modula ion and Coding Scheme (MCS) able o e y low
spec al e iciency a e in oduced. Addi ionally, o imp o e
obus ness o Physical Downlink Con ol Channel (PDCCH),
Release 16 in oduces Downlink Con ol In o ma ion (DCI)
o ma s wi h con igu able sizes, and a highe PDCCH
Agg ega ion Le el (AL), which inc eases decoding capabili y
o PDCCH by he UE.
As a as a chi ec u al aspec s o 5G a e conce ned, a
co ne s one echnology in oduced by 5G ne wo ks o simul a-
neously comply o di e ging use case equi emen s is ne wo k
slicing [129]. Fundamen ally, ne wo k slicing means ha a
sha ed physical ne wo k in as uc u e can be pa i ioned—
sliced—in o mul iple isola ed logical ne wo ks, each ailo ed
o gua an ee a QoS ha mee s he equi emen s o a speci ic
applica ion o e ical use case [130] [131] [132]. Relying on
he ounda ions p o ided by NFV and SDN echnologies, hese
ne wo k slices a e comp ised by a se o NFs and associa ed
p og ammable esou ces, and can be dynamically c ea ed,
scaled, and e mina ed acco ding o demand. La gely, ne wo k
slices can be used o p o ide dedica ed logical ne wo ks wi h
g anula and cus omized con igu a ion acco ding o speci ic
unc ional (e.g., URLLC), adminis a i e, o business equi e-
men s [133], h ough p ocedu es ca ied wi hin bo h he Co e
Ne wo k and he RAN [134].
Finally, ano he key capabili y ha dis inguishes 5G ech-
nology om i s p edecesso s is ne wo k p og ammabili y,
which is c i ical o acili a ing e ical cus ome s o ansla e
hei QoS equi emen s in o ailo ed ne wo k se ices and
con igu a ions. Tigh ly linked wi h he SBA, 3GPP has de ined
5GC exposu e capabili ies h ough s anda d APIs o e he
Ne wo k Exposu e Func ion (NEF), enabling ex e nal sys ems
o communica e wi h he 5GC in a secu e manne and in lu-
ence ne wo k se ices and esou ces. In his line, amewo ks
such as Common API F amewo k (CAPIF) [135] and Se ice
Enable A chi ec u e Laye (SEAL) [136], in oduced in
Release 15 and Release 16 espec i ely, augmen hese expo-
su e capabili ies h ough he p o ision o en iched in e aces,
p ocedu es and auxilia y se ices, which, o example, opens
278 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
TABLE X
LOWEST BOUNDS OF MAXIMUM END-TO-END LATENCY FOR SIGNIFICANT INDUSTRIAL VERTICAL APPLICATIONS ACCORDING TO TS 22.104
up new oppo uni ies o indus ial applica ions in 5G non-
public ne wo ks (NPN) [137]. These openness capabili ies o
5G ne wo ks ha e p opelled a new ecosys em o Ne wo k
Applica ions, o Ne Apps o sho . By consuming he APIs
exposed by he 5G con ol plane, Ne Apps cons i u e a
middlewa e laye o e ical applica ions owa ds 5G unc-
ionali ies, simpli ying he deploymen o e ical applica ions
wi h enhanced in elligence by i ue o aking ull ad an age
o 5G capabili ies [138].
B. S anda ds o Enabling Time-Sensi i e Communica ions
O e 5G Ne wo ks
The p ima y e e ence o in eg a ing 5G and TSN ne wo ks
is 3GPP s anda diza ion ac i i ies. The need o suppo ing
TSN in eg a ion and Time Sensi i e Communica ions (TSC)
in 5G sys ems is e lec ed in 3GPP Technical Speci ica ion
(TS) 22.104 [139],[140]. This documen discusses he
communica ion equi emen s d i en by indus ial e ical
domains, including: ac o y au oma ion, p ocess au oma ion,
and human-machine in e ac ion o he Fac o ies o he Fu u e;
elec ic-powe dis ibu ion and cen al powe gene a ion; and,
in he Release 17 e ision, connec ed hospi als. This TS a i es
o an ex ensi e lis o end- o-end pe o mance equi emen s
ha he 5GS shall comply o in o de o suppo he needs o
cybe -physical con ol applica ions in hese e ical domains,
which ea u e communica ion se ices aligned wi h hose
a ge ed by TSN sys ems as desc ibed in Sec ion III-I.
Key Pe o mance Indica o s (KPIs) p o ided wi h TS
22.104 o di e en kinds o 5G indus ial applica ions include
end- o-end la ency, a ailabili y and eliabili y, UE densi y,
se ice a ea, clock synch oniza ion, and posi ioning accu acy.
In ega ds o clock synch oniza ion, a maximum e o o 900
ns is a ge ed o bo h Wo king Clock and Global Time, as
well as suppo o up o ou simul aneous Wo king Clock
synch oniza ion domains on a UE. La ency and eliabili y
KPIs a e p o ided acco ding o e ical applica ion. Se ice
a ailabili ies o up o 8 nines (99.999999%) and mean ime
be ween ailu es o up o 10 yea s a e p esen . Some o he
mos s ingen la ency KPIs de ined in TS 22.104 a e shown
in Table X.
The speci ica ions o achie ing hese ime-sensi i e capa-
bili ies in 5G ne wo ks a e mainly co e ed in TS 23.501
and TS 23.502. Ex ensions o p o ide suppo o TSC a e
in oduced in 3GPP 5G Release 16 [124],[141],by he
means o he explici in eg a ion be ween he 5GS and TSN.
Fu he enhancemen s o TSC a e added in Release 17 [142],
[143]. In ac , he capabili ies and ea u es o TSC including
5G-TSN in eg a ion a e expec ed o con inue e ol ing wi h
Release 18 and u u e eleases, as 5G echnology ma u es
in o 5G Ad anced and, e en ually, 6G. The emainde o his
subsec ion del es in o he a chi ec u e aspec s, mechanisms,
and p ocedu es s anda dized as o he w i ing o his a icle.
1) 3GPP S anda ds Suppo o an In eg a ed 5G-TSN
A chi ec u e: Release 16 in oduces an a chi ec u e o suppo
TSC by de ining mechanisms and p ocedu es ha pe mi
o expose he 5GS as a logical TSN b idge. This b idge
can be used o in e connec ield de ices, AGVs, mobile
obo s, came as, e c. o ex e nal TSN ne wo ks wi elessly
h ough a 5G UE, hence seeking o emo e he dependency
om in lexible wi ed connec ions o in e connec he asse s
o p oduc ion lines in indus ial au oma ion ne wo ks. The
dedica ed en i ies ha enable his modelling o he 5GS a e
he De ice-Side TSN T ansla o (DS-TT), he Ne wo k-Side
TSN T ansla o (NW-TT), and he TSN Applica ion Func ion
(TSN AF). The TSN AF is esponsible o he con ol plane
in e ac ions wi h he CNC in he TSN domain, pe o ming
he exchange o po and b idge managemen in o ma ion
be ween bo h domains. DS-TT and NW-TT a e loca ed a
he edges o he 5GS, in he UE side and in he UPF side,
espec i ely. Being synch onized o bo h he 5G ime domain
and o he TSN ime domain, hey a e he en i ies esponsi-
ble o he in e ope a ion and synch oniza ion be ween bo h
domains.
A 5GS TSN b idge (5GS b idge o sho ) is composed o
a single UPF and associa ed NW-TT, and o one o mo e UEs
and associa ed DS-TTs. Each UE es ablishes a PDU session o
ype E he ne e mina ing in he UPF o he b idge. Mul iple
PDU sessions om he same UE e mina ing in di e en UPFs
a e modelled as sepa a e b idges, wi h a di e en DS-TT being
pa o each b idge o each sepa a e PDU session es ablished
by he UE (Figu e 8). Hence, he e is a one- o-one ela ionship
be ween B idge ID and UPF ID. Meanwhile, he TSN AF
is esponsible o s o ing he associa ion be ween UE/DS-
TT po s, hei MAC add esses—which a e known by he
SMF a e PDU session es ablishmen —and B idge ID. This
in o ma ion is used o binding TSN a ic wi h a espec i e
UE’s PDU session.
The in o ma ion model encompassing a 5GS b idge ha can
be con igu ed in he 5GS by he TSN ne wo k and/o epo ed
om he 5GS o he TSN ne wo k includes he a ibu es lis ed
below. In o ma ion ecei ed in he 5GS om he TSN ne wo k
can be mapped o 5GS in o ma ion o con igu e 5GS QoS o
TSN a ic.
•B idge desc ip ion, including B idge ID and lis o po s
o DS-TT and NW-TT.
•Capabili ies o he 5GS b idge ha a e epo ed o he
TSN ne wo k, as de ined in 802.1Qcc. This comp ises
b idge delay pe po pai pe a ic class, p opaga ion
delay pe po , and VLAN con igu a ion in o ma ion.
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 279
Fig. 8. The 5G Sys em ac ing as one o mo e logical TSN b idges allowing o connec wi eless ield de ices in a p oduc ion line o a wi ed TSN ne wo k.
An UE ha es ablishes edundan PDU sessions owa ds sepa a e UPFs se s up connec i i y h ough mul iple logical TSN b idges.
•LLDP con igu a ion in o ma ion, and neighbo disco e y
in o ma ion and con igu a ion.
•Scheduled a ic in o ma ion acco ding o he 802.1Qb
s anda d, including a ic classes and hei p io i ies pe
po and ga e con ol in o ma ion.
•In o ma ion ela ed o he 802.1Qci s anda d, including
he S eam Pa ame e Table ha desc ibes PSFP suppo ,
and PSFP s eam il e s and s eam ga es.
To allow o he ansmission o TSN scheduled a ic o e
a 5GS b idge, he 5GS shall also suppo a hold and o wa d
bu e ing mechanism implemen ed in DS-TT and NW-TT
po s. This mechanism a ge s achie ing he beha io p o ided
by TAS as de ined in he 802.1Qb s anda d, ansmi ing
o holding back ames acco ding o he s a e o he co e-
sponding ansmission ga e. Howe e , bo h he Release 16 and
17 s anda ds do no u he speci y on he de ails, lea ing hem
up o implemen a ion.
Apa om he po and b idge managemen in o ma ion
exchange be ween CNC and TSN AF, he 5GS also shall
suppo anspa en ans e o his in o ma ion be ween TSN
AF and DS-TT o NW-TT. This is done h ough s anda dized
Po Managemen In o ma ion Con aine s (PMIC) and B idge
Managemen In o ma ion Con aine s (BMIC), which con ain
much o he in o ma ion desc ibed abo e. Release 17 enames
BMIC o Use plane node Managemen In o ma ion Con aine
(UMIC).
The suppo o TSC and o in eg a ion wi h TSN ne wo ks
was enhanced in Release 17 wi h he addi ion o new capa-
bili ies and ea u es. Many o hese a ge he suppo o
non-TSN TSC, which is no conside ed he ocus o his
esea ch pape . Fo example, Release 17 adds suppo o TSC
PDU sessions o ype IP and suppo o IEEE 1588-based
ime synch oniza ion. In his con ex , Release 17 in oduces a
new Ne wo k Func ion called Time Sensi i e Communica ion
and Time Synch oniza ion Func ion (TSCTSF). The TSCTSF
enables he new ime synch oniza ion mechanisms, as well
as non-TSN TSC scena ios. In Release 17, Managemen
In o ma ion Con aine s may be exchanged be ween NW-TT
and TSCTSF and also include ime synch oniza ion pa ame e s
o he pu pose o enabling he TSCTSF o manage DS-
TT and NW-TT in he absence o TSN AF. Las ly, Release
17 also in oduces suppo o UE-UE TSC communica ions,
ea ing he UE-UE TSC s eam as a combina ion o one uplink
s eam and one o mo e downlink s eams and ca ying hei
p ocedu es sepa a ely.
2) 3GPP S anda ds Suppo o TSN Time Synch oniza ion:
Suppo ing TSN ime synch oniza ion is a c i ical equi emen
o enabling he in eg a ion be ween 5G and TSN. Wi hin he
3GPP s anda diza ion ac i i ies, suppo o TSN ime syn-
ch oniza ion is in oduced in Release 16 and u he expanded
in Release 17.
Release 16 in oduces he exposu e o he 5GS as a logical
TSN b idge modelled as a ime-awa e sys em complian o he
802.1AS s anda d, as ep esen ed in Figu e 9. Wi hin he 5GS,
5G en i ies including he UE, gNB, and UPF a e synch onized
o he 5G GM clock acco ding o 5G-AN synch oniza ion
mechanisms [144] ha u ilize he ITU-T G.8275.1 p o ile o
elecommunica ion applica ions based on PTP, and a e e ence
ime signaling mechanism om he gNB o synch onize an
UE. Meanwhile, TSN en i ies ou side he 5GS a e ins ead
synch onized o he TSN GM in he TSN domain. The TSN
ansla o s (i.e., NW-TT and DS-TT) a he edges o he
5GS a e synch onized o bo h GM clocks, and, implemen ing
suppo o 802.1AS ope a ions, a e esponsible o linking he
5G ime domain and he TSN ime domain.
Release 16 suppo s he elec ion o he GM ins ance and
he de e mina ion o he ime synch oniza ion ee ia manual
con igu a ion. DS-TT po s a e con igu ed as Mas e Po s

280 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
Fig. 9. Modeling o he 5G Sys em o suppo downlink TSN ime synch oniza ion om a TSN domain behind he UPF owa ds an end s a ion behind an
UE, including he anspo a ion o iming messages om Mas e Po s o Sla e Po s. NW-TT and DS-TT a e he en i ies esponsible o linking he 5GS
ime domain and he TSN ime domain.
and o wa d iming in o ma ion o he downs eam TSN nodes
behind he UEs, whe eas each NW-TT po is con igu ed as
a Sla e Po o ecei e iming in o ma ion om he ups eam
gPTP ime-awa e ne wo k o which he UPF o he b idge is
connec ed o.
The p ocedu es o enable TSN ime synch oniza ion ac oss
he 5GS a e implemen ed acco ding o he 802.1AS s anda d.
In Release 16, gPTP synch oniza ion is always ini ia ed
downlink om he GM in he TSN domain. When a NW-
TT ecei es a gPTP message, i needs o calcula e he
message’s upda ed Co ec ionField aking in o accoun he
link delay om he ups eam TSN node, and he upda ed
Cumula i eRa eRa io accoun ing o he equency o se
be ween 5GS ime and TSN ime. Such ields a e upda ed
in he gPTP messages ha he NW-TT o wa ds h ough he
5GS owa ds he DS-TT. The esidence ime in he 5GS is
calcula ed ia he c ea ion o ing ess and eg ess imes amps
(TSi and TSe) by he NW-TT and DS-TT espec i ely. The
TSi is added o he gPTP messages by he NW-TT and la e
emo ed by he DS-TT. Finally, his esidence ime is used
o calcula e he inal alue o he Co ec ionField when he
DS-TT o wa ds he gPTP message o he downs eam TSN
node. In Release 16, he connec ion o mul iple TSN domains
is also suppo ed, wi h he NW-TT and DS-TT en i ies being
able o handle he imes amping and o wa ding o gPTP
messages acco ding o he abo e p ocedu e o one o mo e
domains.
Release 17 expands he suppo o TSN ime synch o-
niza ion capabili ies in he 5GS. Apa om ope a ing as a
802.1AS ime-awa e sys em, he 5GS can also ope a e as
o he ypes o PTP ins ances, including IEEE 1588 complian
Bounda y Clock, End- o-End T anspa en Clock, and Pee -
o-Pee T anspa en Clock (Figu e 9). Hence, om Release
17, he 5GS sys em suppo s bo h gPTP and PTP-based ime
synch oniza ion wi h ex e nal ime domains. Addi ionally, in
he gPTP and PTP Bounda y Clock schemes, he 5GS can
be con igu ed o ope a e as he GM clock o he gPTP o
PTP domain, as opposed o he clock sou ce being loca ed
in he wi ed TSN ne wo k. Suppo o uplink ime synch o-
niza ion, including he UE-UE scena io, is also in oduced,
co esponding o a GM loca ed behind an UE a he han in
he ne wo k behind he UPF. Ha ing mul iple TSN domains
behind he UE each wi h hei own GM o uplink ime
synch oniza ion is also suppo ed. Addi ionally, Release 17
in oduces suppo o he BMC algo i hm o de e mining he
GM clock and he ime synch oniza ion ee, as opposed o
eso ing o only p econ igu ed GM and po s a es. Finally,
ime synch oniza ion exposu e capabili y owa ds an a bi a y
AF is in oduced in Release 17, allowing he AF o eques
ime synch oniza ion se ices (indi ec ly ia NEF i he AF is
non- us ed) ha may include a speci ic ime synch oniza ion
e o budge .
To mee he s ic ime synch oniza ion equi emen s o
ime-sensi i e communica ions, TR 38.330 [145] in Release 17
es ablishes he use o wo echniques ha help compensa e he
p opaga ion delay wi hin he 5GS in he esidence ime ha is
added o he Co ec ionField o he gPTP o PTP messages.
On he one hand, Round-T ip Time (RTT)-based PDC employs
5G NR e e ence signals ansmi ed be ween gNB and UE in
bo h di ec ions. Bo h gNB and UE measu e he ime di e ence
in he e e ence signal ecei ed om he o he en i y, and ei he
o he wo p o ides he o he en i y wi h i s own measu emen ,
so ha i can pe o m PDC based on he measu emen s om
i sel and om he o he en i y. On he o he hand, Timing
Ad ance (TA) sen om gNB o UE o adjus he iming o
uplink ansmissions based on he es ima ed p opaga ion delay
can also be used o compensa e his delay.
3) 3GPP S anda ds Suppo o he QoS Con igu a ion
and T ansla ion o TSN T a ic: 5G s anda ds suppo he
in eg a ion wi h TSN ne wo ks ha apply he ully cen alized
model as de ined in he 802.1Qcc s anda d. QoS con igu a ion
o suppo TSN a ic in he 5GS in ol es he use o
TSC Assis ance In o ma ion (TSCAI) and o 5G QoS lows.
The aim is o achie e a ansla ion o pa ame e s om he
wi ed, ime-domain-based E he ne , o he wi eless, equency-
domain-based 5G h ough he mos app op ia e mapping o
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 281
TSN s eam equi emen s o 5G adio esou ces. Fo his
pu pose, 3GPP has s anda dized suppo o QoS con igu a ion
ha may be implemen ed in wo ways acco ding o a ailabili y
o PSFP in o ma ion p o ided by he CNC, o he lack he eo .
•I PSFP in o ma ion is always p o ided o he TSN AF
by he CNC, hen his in o ma ion is used o se up he
QoS lows in he 5GS. PSFP in o ma ion can also be used
o de i e TSCAI o each QoS low.
•I PSFP in o ma ion is no p o ided by he CNC, p econ-
igu ed QoS lows ha do no equi e PSFP in o ma ion
a e used, and may be se up du ing PDU session es ab-
lishmen . The means o de i e TSCAI a e no speci ied
o his case. I may howe e be s ill possible o de i e
QoS in o ma ion (p io i y and delay) o TSN s eams
om scheduled a ic in o ma ion (i.e., he TAS GCL)
p o ided by he CNC.
TSCAI includes a desc ip ion o TSC low a ic pa e ns
ha can be p o ided o he gNB o allow i o e icien ly
con igu e 5G-AN esou ces o he scheduling o de e minis ic
a ic lows ( o example ia CGs o SPS). TSCAI pa ame e s
a e de i ed om he PSFP in o ma ion ha he TSN AF
ecei es om he CNC. TSN AF o wa ds pe - low TSC
Assis ance Con aine (TSCAC) wi h a ic pa e n pa ame e s
calcula ed om his in o ma ion o he SMF. These pa ame e s
a e Flow Di ec ion (uplink o downlink), Pe iodici y ( ime
pe iod be ween ansmissions, o bu s s), and Bu s A i al
Time. Release 17 in oduces a new pa ame e , Su i al Time,
which e e s o he ole ance agains he lack o a i al o
an icipa ed messages. The SMF builds he TSCAI by adjus ing
hese pa ame e s acco ding o he di e ence be ween 5G and
TSN clocks, and o wa ds he esul o he 5G-AN.
Whe eas TSCAI con ains QoS con igu a ion used speci -
ically o he a ic cha ac e is ics o TSC lows, all 5G
QoS lows a e iden i ied by a QoS Flow Iden i ie (QFI)
and cha ac e ized by a QoS P o ile. A QoS P o ile i s ly
classi ies lows acco ding o Resou ce Type, ha is, be ween
Gua an eed Bi Ra e (GBR) ones and Non-Gua an eed Bi
Ra e (Non-GBR) ones. These QoS P o iles desc ibe se e al
QoS cha ac e is ics, which, o GBR lows, include a leas
he ollowing: Alloca ion and Re en ion P io i y (ARP), which
de e mines p io i y le el and p eemp ion capabili y, uplink and
downlink Gua an eed Flow Bi Ra e (GFBR) and Maximum
Flow Bi Ra e (MFBR), and a 5G QoS Iden i ie (5QI). To
de e mine he igh QoS P o ile o each TSN s eam, he TSN
AF is p econ igu ed wi h a QoS mapping able ha associa es
TSN a ic classes o b idge delays—which can be calcula ed
a e PDU session es ablishmen —and o p io i y le els.
A 5QI e e ences a combina ion o addi ional QoS cha ac-
e is ics ha impac he o wa ding ea men o a gi en low
in he RAN. P ecisely, QoS P o iles and 5QIs can be consid-
e ed mechanisms o implemen 3GPP RAN ne wo k slicing
in 5G ne wo ks [146]. 3GPP has s anda dized se e al 5QI
alues [124], including some wi h a Resou ce Type o Delay-
c i ical Gua an eed Bi Ra e (Delay-c i ical GBR) aimed a use
cases such as ime-sensi i e indus ial communica ions. The
5GS suppo s he use o bo h s anda dized o cus omized 5QIs,
and, as such, TSC QoS lows may also use ei he op ion.
Apa om he a o emen ioned Resou ce Type cha ac e is ic
ha de e mines he alloca ion o esou ces, addi ional QoS
cha ac e is ics a e included in he 5QI, such as P io i y Le el
o he scheduling o esou ces, uppe bounds o he Packe
Delay Budge (PDB) and o he Packe E o Ra e (PER), and
Maximum Da a Bu s Volume (MDBV). These can be mapped
acco dingly, ia a PCF QoS mapping able, om TSC QoS
pa ame e s such as he maximum bu s size o he agg ega ed
TSC s eams alloca ed o a 5G QoS low ha is p o ided by
he CNC, and he maximum low bi a e de i ed om he
calcula ion o TSCAI. The PDB o a 5QI has o sa is y—
i.e., be equal o o highe han— he b idge delay epo ed by
he 5GS b idge o he CNC o i s associa ed a ic class.
PDB-based QoS can be applied using he hold and o wa d
mechanism.
Figu e 10 summa izes he s eps in ol ed in he con igu a-
ion o a TSN s eam, ocusing on he con igu a ion o 5GS
b idges ha a e included in he pa h. This ollows p ocedu es
de ined by ei he he 802.1Qcc s anda d o he ully cen-
alized model, o by 3GPP s anda ds. Fo addi ional de ails
on he 5GS p ocedu es in ol ing TSN b idges, including he
in e en ion o 5GC NFs like he PCF, SMF, and AMF, he
eade is e e ed o Annex F o TS 23.502 [143].
Based on he 3GPP s anda ds, 5G-ACIA whi e pape [147]
p o ides a de ailed desc ip ion o he 5G QoS model and o
5G p ocedu es o enabling he con igu a ion and managemen
o QoS ea u es o ime-sensi i e lows. Some key akeaways
o he indus ial au oma ion use case in pa icula a e:
•The abili y o a UE, ha could o example be connec ed
o a ield de ice, o eques he modi ica ion o a PDU
session in o de o assign di e en QoS pa ame e s o
addi ional QoS lows.
•The abili y o an ex e nal applica ion, such as
Manu ac u ing Execu ion Sys em (MES) and En e p ise
Resou ce Planning (ERP), o lexibly manage QoS
P o iles and moni o he QoS pa ame e s ia he con-
sump ion o he NEF APIs o he 5GC.
•The suppo o adap i e QoS applica ions in which wo
o mo e al e na i e QoS P o iles wi h di e en p io i ies
may be speci ied, so he applica ion can su i e e en s
o empo a y QoS deg ada ion in he ne wo k p o iding
a lowe quali y se ice (e.g., adjus ing he esolu ion o
a ideo s eam).
•The in e wo king wi h di e en deploymen con igu a-
ions o 5G ne wo k slicing o he p o isioning o
ailo ed QoS lows.
4) Time-Sensi i e Ne wo king o F on haul: While one
di ec ion o s anda diza ion and esea ch wo k in ol ing he
usage o 5G and TSN oge he is he 5GS b idge o in e con-
nec ing TSN de ices o e 5G, ano he di ec ion add esses he
usage o TSN o achie e de e minis ic low-la ency ansmis-
sions in he 5G TN. As such, a he han s udying how a 5G
ne wo k can be inco po a ed as a wi eless subcomponen o
an indus ial TSN ne wo k, his esea ch di ec ion le e ages
he adop ion o TSN wi hin he wi ed in as uc u e—i.e.,
be ween RAN and UPF—o a 5GS ha could i sel ac as
a 5GS b idge as pa o an end- o-end TSN ne wo k. The
282 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
Fig. 10. Fully cen alized TSN ne wo k con igu a ion o a TSN s eam spanning a TSN b idge and a 5GS b idge.
co ne s one s anda d o using TSN as a building block o
5G TNs is he 802.1CM-2018 s anda d [148], a collabo a i e
e o be ween Common Public Radio In e ace (CPRI) and
IEEE 802.1. 802.1CM is he TSN P o ile o F on haul, and,
much like he case o he TSN-IA p o ile, i selec s TSN
ea u es and con igu a ions o , in his case, anspo ing ime-
sensi i e on haul s eams. E en hough his app oach may
no ge as much ocus as he 5GS b idge in he con ex o
enabling ime-sensi i e indus ial communica ions o e 5G,
using TSN in he TN can be a key enable o achie ing
end- o-end de e minis ic communica ions [149], speci ically
o he on haul connec ion, which has e y s ingen packe
loss and delay equi emen s. Fu he mo e, bo h app oaches o
combining 5G and TSN can be employed in complemen a y
ashion.
802.1CM desc ibes how o mee he s ic equi emen s o
E he ne -based on haul ne wo ks using TSN mechanisms.
Following he 5G unc ional spli , a on haul link connec s
a RU wi h a DU, and is implemen ed h ough he CPRI
p o ocol, o i s enhanced e sion, e ol ed CPRI (eCPRI). Bo h
CPRI and eCPRI a e suppo ed in he 802.1CM s anda d,
in which hey a e called Class 1 a ic and Class 2 a ic
espec i ely. CPRI and eCPRI de ine di e en unc ional spli
op ions, which indica e he line in which he di ision be ween
DU and CU unc ionali ies akes place.
Di e en kinds o in o ma ion lows a e ansmi ed h ough
he on haul link, as pe he CPRI and eCPRI speci ica-
ions. This includes Use da a, Con ol and Managemen
(C&M) da a, and Synch oniza ion da a. Use da a includes In-
Phase/Quad a u e (I/Q) da a, and, in he case o Class 2 a ic,
also o he da a ela ed o he Use plane. Synch oniza ion
da a is p o ided sepa a ely om he o he wo. Fo ime
synch oniza ion, he s anda d speci ies suppo o he ITU-T
G.8275.1 elecom p o ile and one o mo e o he ela ed clocks
as a equi emen .
TABLE XI
CHARACTERISTICS OF CLASS 1 (CPRI) INFORMATION FLOWS
TABLE XII
CHARACTERISTICS OF CLASS 2(ECPRI) INFORMATION FLOWS
The espec i e CPRI/eCPRI speci ica ions de ine he
equi emen s o he emaining wo ypes o da a, in he o m
o maximum la ency and ame loss p obabili y. This leads o
h ee ypes o on haul a ic classes acco ding o p io i y,
which a e High P io i y F on haul (HPF) wi h he lowes one-
way la ency equi emen o 100 µs, Medium P io i y F on haul
(MPF) wi h a one-way la ency equi emen o 1 ms, and Low
P io i y F on haul (LPF), wi h he mos elaxed equi emen o
100 ms one-way la ency. Table XI and Table XII summa ize
he equi emen s o he in o ma ion lows and hei mapping
o on haul low ypes, o Class 1 a ic and Class 2 a ic
espec i ely.
To sa is y such equi emen s, 802.1CM de ines wo possible
p o iles, called P o ile A and P o ile B. The wo p o iles a e
applicable o bo h Class 1 and Class 2 a ic, and essen ially
di e in he TSN ea u es adop ed. Speci ically:
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 283
•P o ile A: he s ic p io i y algo i hm, as de ined in he
802.1Q s anda d, is employed in o de o ea Use Plane
da a (highe p io i y) and Con ol and Managemen Plane
da a (lowe p io i y) di e en ly. The highes p io i y is
assigned o HPF a ic, ollowed by MPF a ic.
•P o ile B: apa om making use o he s ic p io i y
algo i hm like in P o ile A, ame p eemp ion mecha-
nisms o 802.1Qbu and 802.3b a e also used. This allows
non- on haul a ic o be p eemp able by on haul
a ic.
I is wo h no ing he exis ence o ammendmen
802.1CMde-2020 [150] o he 802.1CM s anda d. I in o-
duces enhanced synch oniza ion ea u es and o he upda es
ha speci ically add ess 5G on haul in e aces s anda dized
a e he elease o 802.1CM-2018.
C. Vi ualiza ion and Cloudi ica ion o Time-Sensi i e
Applica ions
Vi ualiza ion has p o en o be a echnology ha can
signi ican ly imp o e lexibili y and cos -e iciency in com-
pa ison o monoli hic app oaches, being speci ically a ocal
echnology in he indus ial au oma ion e ical ha conce ns
his wo k o de elop no el Indus y 4.0 and beyond appli-
ca ions [151],[152]. As indus ial con ol unc ions mo e
o p i a e edge cloud in as uc u es enabled by 5G [153],
pa adigms such as NFV, MEC, and cloud-na i e can be
adop ed in he design o he a chi ec u e o indus ial ne wo ks
and applica ions. Howe e , i ualiza ion pla o ms also en ail
addi ional complexi y and o en come wi h a dis ega d o
op imized pe o mance ha may hampe hei iabili y o hos
ime-c i ical applica ions wi h low la ency and bounded deli -
e y ime equi emen s in he communica ions. Fu he mo e,
gene al-pu pose sys ems do no ocus on p o iding eal- ime
capabili ies, equi ing speci ic uning o example in embedded
sys ems o indus ial con ol applica ions ha a e commonly
based on Linux [154],[155]. This subsec ion i s e iews
a ailable echnologies and mechanisms ha can add ess hese
challenges, and la e in oduces he i ualiza ion and edge
compu ing pa adigms men ioned abo e.
1) Technologies and Mechanisms o De e minis ic End
S a ions wi h Vi ualiza ion: The Linux ecosys em p o ides
se e al TSN ea u es o accommoda e TSN-complian applica-
ions. Beginning wi h VLAN agging suppo , mul iple i ual
in e aces wi h he desi ed VLAN PCP can be associa ed
o a physical in e ace. In e nal Linux packe p io i ies, se
wi h socke op ion SO_PRIORITY, can be lexibly mapped
o VLAN PCPs. Fo enabling ime synch oniza ion, Linux
PTP [156] is he mos popula implemen a ion o PTP. I
includes suppo o he gPTP p o ile and suppo s ha dwa e
and so wa e imes amping; he o me also equi es NIC
and d i e suppo . Linux PTP includes wo u ili y daemons,
p p4l and php2sys, o synch onize he ha dwa e clock o
he NIC and he sys em clock, espec i ely. T a ic shaping
unc ionali y is implemen ed h ough T a ic Con ol (TC)
Queuing Disciplines (Qdiscs), which a e associa ed o a
speci ic ne wo k in e ace o con ol he scheduling, shap-
ing, and policing o ansmi ed a ic. CBS unc ionali y is
implemen ed in he Mul iQueue P io i y (MQPRIO) Qdisc,
whe eas TAS unc ionali y is implemen ed in he Time
Awa e P io i y (TAPRIO) Qdisc. The con igu a ion o hese
Qdiscs includes ea u e-speci ic pa ame e s, and he mappings
be ween Linux p io i ies, a ic classes, and he ansmis-
sion queues o he ne wo k in e ace. Addi ionally, he
Ea lies TxTime Fi s (ETF) Qdisc can be used alongside he
TAPRIO Qdisc o p o ide ime-based scheduling using he
SO_TXTIME socke op ion. The TSN Documen a ion P ojec
o Linux [157] ga he s comp ehensi e documen a ion and
con igu a ion ins uc ions ega ding he ea u es in oduced in
his pa ag aph.
To p o ide eal- ime capabili ies o he s anda d Linux
ke nel, he PREEMPT_RT pa ch [158] was de eloped. This
eal- ime p eemp ion pa ch a ge s a maximiza ion o he
amoun o poin s in he ke nel code whe e i s execu ion can be
in e up ed, hus educing he delay uppe bound ha eal- ime
asks may expe ience. PREEMPT_RT is being p og essi ely
mainlined in he Linux ke nel. An al e na i e app oach o
achie e eal- ime capabili ies is he use o a eal- ime co-
ke nel ha uns in pa allel o he Linux ke nel and speci ically
execu es he ime-c i ical wo kloads. Examples o he co-
ke nel app oaches a e Xenomai [159] and RTAI [160]. Las ly,
a po en ially less in usi e al e na i e o sa is y so e eal-
ime equi emen s is he low la ency Ubun u ke nel [161],
which is con igu ed wi h he maximum p eemp ion and ime
g anula i y a ailable in he mainline ke nel.
Task scheduling in Linux is ca ied ou acco ding o poli-
cies. The SCHED_FIFO, SCHED_RR (Round Robin), and
SCHED_DEADLINE policies a e called eal- ime policies as
asks scheduled wi h hese policies always ha e p io i y o e
hose scheduled wi h no mal ones. The SCHED_DEADLINE
policy in pa icula inco po a es awa eness o he deadline o
he ask. Vi ualiza ion echnology po en ially inc eases he
amoun o mixed-c i icali y applica ions execu ing p ocesses
in he same hos sys em, highligh ing he need o use eal-
ime scheduling policies ha can gua an ee empo al isola ion
be ween he i ualized componen s [162],[163].
Ano he po en ial sou ce o la ency in i ualized applica-
ions is ela ed o he ne wo k s ack. Vi ualiza ion in oduces
a i ual swi ching laye esponsible o di ec ing he packe s
o and om he espec i e i ual componen s ha use an
emula ed i ual NIC, all o which can be causes o addi ional
o e head [164]. Enhanced Pla o m Awa eness [165] con ig-
u a ion such as CPU and in e up a ini y can o example
op imize he se icing o packe p ocessing in e up s. In his
con ex , PCI Pass h ough and SR-IOV Pass h ough echniques
can be used o p o ide a i ualized componen wi h di ec
access o a physical ne wo k in e ace. SR-IOV allows he
pa i ioning o he NIC a ha dwa e le el in o Vi ual Func ions
ha can be sha ed be ween gues s. S ill, lexibili y is some-
wha educed as he gues emains ied o a pa icula NIC.
Fu he mo e, ypical NICs a ailable in he ma ke usually do
no suppo bo h TSN and Pass h ough [166].
Las bu no leas , accele a ion echnologies such as Da a
Plane De elopmen Ki (DPDK) [167] and eXp ess Da a Pa h
(XDP) [168] can also be used o ci cum en o e head in o-
duced by he ne wo k s ack, especially when a i ualiza ion
290 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
TABLE XIII
TAXONOMY RELATED TO THE DESIGN AND EVALUATION OF TSN-OVER-5G ARCHITECTURES
he enablemen o ime synch oniza ion on in eg a ed 5G-TSN
ne wo ks, nei he hose ha p edomina ely ackle p ocedu al
aspec s o he con igu a ion o TSN a ic in such ne wo ks,
as hey will be su eyed in subsequen subsec ions.
Table XIII shows a axonomy o he esea ch pape s su -
eyed in his sec ion. Rega ding QoS mapping be ween TSN
and 5G, he e a e se e al complemen a y scopes esea ched
in he li e a u e, such as b idge delay es ima ion and he
con igu a ion o 5QIs. When i comes o op imizing la ency
and de e minism o e 5G, mos p oposals can be iden i ied
unde semi-s a ic scheduling schemes, such as hose based
on SPS and CG, hough o he app oaches can also be ound.
Meanwhile, mechanisms p oposed o enhancing eliabili y
include NR-DC and o he edundan use plane pa h con igu-
a ions, CoMP, and Mul i-TRP.
1) Ne wo k A chi ec u es o In eg a ing TSN in o 5G
Ne wo ks: Au ho s o [234] p opose di e en hyb id scena ios
ha combine 5G and Indus ial E he ne based on TSN
in o de o add ess eliabili y and la ency equi emen s o
indus ial communica ion sys ems. P eda ing much o he
s anda diza ion wo k by SDOs such as 3GPP and IEEE/IEC,
his a icle e alua es how di e en ne wo k opologies can be
applied o di e en wi eless communica ion use cases in he
Sma Fac o ies o he u u e, anging om mobile obo ics
and closed loop con ol o massi e senso ne wo ks. Such
5G-TSN in eg a ion scena ios in oduce aspec s such as he
i ualiza ion o a cen alized con olle in an edge cloud
o wi eless in e connec ion be ween emo e p oduc ion si es.
Conduc ed in pa allel o 3GPP Release 16 s anda diza ion
wo k, [235] explo es p elimina y app oaches o an a chi-
ec u e, and p o ocol design aspec s, o he in eg a ion o
5G wi h TSN. Key ea u es ha allow he 5GS o ope a e
as a TSN swi ch a e desc ibed, as well as a chi ec u e
and echnology-le el conside a ions o enabling end- o-end
ime synch oniza ion, ansmission scheduling, and 802.1Qb
and 802.1Qcc-based QoS managemen o TSN a ic wi h
s ingen la ency and eliabili y equi emen s. Re e ence [236]
p oposes an ha monized usage o se e al echnologies o he
ealiza ion o a 5G ne wo k a chi ec u e o a manu ac u ing
ac o y, ha comp ises ou - ac o y and in- ac o y domains.
The use o edge compu ing, URLLC, and TSN echnologies
is p oposed o he la e , whe e QoS moni o ing and dual-
connec i i y edundan ansmission mechanisms a e used o
enhance end- o-end delay and eliabili y. Au ho s e alua e
hei a chi ec u e wi h a p ac ical emo e con ol use case.
Likewise, di e en op ions o deploying 5G NPN ne wo ks
using edge compu ing a chi ec u es a e p esen ed in [237],
whe e he in e wo king wi h TSN is also explo ed. In [238],
au ho s p opose a 5G a chi ec u e ailo ed o indus ial TSN
se ices. Thei con ibu ions a e mainly wo old: on he
one hand, adio access echnologies including New Radio
Dual Connec i i y (NR-DC), and Mul i-TRP, a Coo dina ed
Mul ipoin (CoMP) capabili y in oduced in 5G, a e s udied o
imp o ing ne wo k quali y a di e en le els, p ima ily wi h
ega ds o eliabili y and ansmission a e; on he o he hand,
in o de o imp o e da a secu i y and agili y, an on-p emises
MEC deploymen is used o implemen a disagg ega ed RAN
and he use plane o he 5G ne wo k.
2) Analysis and Es ima ion o 5G-TSN Sys em La ency:
In [22], au ho s look in o how he 5GS b idge delay can
be es ima ed, in o de o selec 5QIs wi h QoS pa ame e s
ha a e adequa e o suppo he speci ic end- o-end la ency
equi emen s o de e minis ic indus ial applica ions o e 5G.
PDB, GFBR, and MDBV a e es ima ed conside ing e e ence
equi emen s o a closed loop con ol use case. Au ho s
conclude by s a ing he incon enience o no ha ing s an-
da dized 5QI alues wi h lowe PDB in he sub-millisecond
ange, he p oblema ic o pe iodici y misma ch be ween he
eques ed GCL and he suppo ed 5G nume ology, and he
desi abili y o being able o es ima e he b idge delay be o e
he es ablishmen o he QoS low. In [239], au ho s pe o m
and end- o-end la ency pe o mance o an in eg a ed 5G-
TSN sys em using a simula ed 5GS b idge model. Au ho s
quan i a i ely e ualua e he impac o PCP-based p io i iza ion

SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 291
and scheduled a ic based on 802.1Qb on TSN-o e -5G
pe o mance. Thei esul s show ha PCP p io i iza ion o he
ansmission o ames is pa icula ly bene icial o educing
ji e , whe eas he use o 802.1Qb on op o i signi ican ly
educes la ency o high-p io i y a ic, especially o uplink
due o CGs. Thei esul s also indica e ha ji e can be
signi ican ly educed a he cos o sligh ly highe a e age
la ency as he leng h o he GCL ime slo s becomes la ge .
Au ho s conclude he a icle by highligh ing he challenge
p esen ed by he need o ansla e he ime-domain scheduling
o 802.1Qb o he OFDM modula ion used in 5G. In [240],
au ho s ocus on he calcula ion o he PDB o di e en indus-
ial scena ios ha ha e adop ed 5G-TSN communica ions.
Au ho s p opose h ee communica ion scena ios in ol ing a
PLC and a ield de ice: (1) single 5G link om wi eless de ice
o wi ed PLC; (2) wi eless de ice o local wi eless PLC no
c ossing UPF; and (3) wi eless de ice o emo e wi eless PLC
c ossing UPF. Fo elabo a ing he PDB o mulas, au ho s use
TSN a ic classes wi h p ede ined packe size and packe
a e alues, ake in o accoun end- o-end delay including delay
in oduced ou side he 5GS, and conside he possibili y o
mul iple 5GS b idge hops in he end- o-end communica ion.
Au ho s p opose a 5QI able wi h PDB alues de i ed om
hei calcula ions, al hough hey s a e ha he delay in oduced
by TSN ansla o s needs u he in es iga ion.
3) E alua ion o T a ic Scheduling and Fo wa ding o e
5G: As opposed o 3GPP s anda diza ion, he esea ch
communi y has gene ally ocused mo e on he op imal in e-
g a ion TAS-based a ic shaping o 802.1Qb a he han
PSFP de ined in he 802.1Qci s anda d, de i ing QoS and
TSCAI pa ame e s om he o me . In ac , many esea ch
wo ks speci ically ocus on analyzing he pe o mance o
802.1Qb -based a ic scheduling h ough he alloca ion o
5G adio esou ces. In [127], au ho s quan i a i ely e alua e
he pe o mance o anspo ing TSN a ic o e a 5G ne wo k.
They implemen a sys em-le el simula ion ha models a p e-
limina y 5GS b idge, applying se e al URLLC unc ionali ies
ha a e mainly om Release 15. They p esen end- o-end
la ency and h oughpu pe o mance esul s o scheduling
mul iple TAS s eams wi h di e en 5G QoS con igu a ions.
They highligh open esea ch challenges ega ding ade-o s
in he op imal con igu a ion and scheduling o esou ces
be ween he 5G and TSN sys ems. In a simila ein, a simu-
la ion model is p oposed in [241] o in es iga e he challenges
o 5G and TSN in eg a ion in ega ds o uplink and downlink
ansmission scheduling using he 802.1Qb s anda d. Thei
simula ion esul s using p econ igu ed ansmission slo s (such
as wi h CG) indica e ha a misma ch be ween applica ion
pe iodici y and he 5G ime slo s causes deg ada ion in delay
and ji e due o a iable ansmission s a imes. Meanwhile,
in [242], au ho s p o ide a comp ehensi e pe o mance e al-
ua ion and compa ison o semi-pe sis en and dynamic packe
scheduling (DPS) s a egies o suppo downlink ime-sensi i e
lows in 5G. Au ho s highligh he bene i s in e ms o
adap i i y o channel condi ions o DPS me hods; howe e , he
absence o con ol channel o e head using SPS esul s in be e
pe o mance. Finally, in [243], au ho s conside a 5G-TSN
ne wo k ha accep s bo h URLLC/TSN and eMBB a ic, and
ake in o conside a ion bo h la ency and h oughpu using TAS
scheduling. They e alua e an scheme based on he alloca ion o
RBs o TSN s eams ollowed by eMBB s eams in a manne
ha maximizes he o e all h oughpu while sa is ying he end-
o-end delay o TSN a ic.
4) P oposal o Mechanisms o TSN-complian Resou ce
Alloca ion and T a ic Scheduling o e 5G: A o emen ioned
wo k [242] also p oposes op imiza ions o bo h SPS and DPS
cases o sa is y he la ency cons ain s o TSN a ic o e
5G. Au ho s le e age he in o ma ion p o ided in TSCAI and
employ Release 15 URLLC ea u es. Fo pe o ming DPS
esou ce alloca ion, au ho s p esen link adap a ion s a egies
based on CQI epo s om he UE. Fo SPS, au ho s p o-
pose a mechanism o collec SINR samples a he UE, in
o de o es ima e and selec an adequa e MCS. Au ho s also
analyze he impac on pe o mance o he upda e equency
o he MCS in use. The misma ch p oblem be ween 5G CG
and TSN pe iodici y is add essed in [244], whe e au ho s
p opose a scheduling scheme whe e wo o mo e CGs pe
TSN lows ope a e al e na ely. Thei algo i hm in ol es he
ac i a ion and deac i a ion o CGs o each low o e ime.
This CG shi is au onomously igge ed by he UE based
on gNB in o ma ion when he h eshold o he ansmission
la ency equi emen would be su passed due o cumula i e
la encies caused by pe iodici y misalignmen . Also analyzing
TSN s eam scheduling pe o mance in an in eg a ed 5G-TSN
sys em, [245] p oposes packe p io i iza ion mechanisms o
he app op ia e scheduling and ea men o hose s eams.
Au ho s analyze he beha io o a ic o wa ding and a i e o
condi ions ha mee TSN-complian scheduling. As p e iously
men ioned wo ks, hey emphasize on he challenges ha a ise
om mapping TSN a ic scheduling and o wa ding ules o
5G esou ces and mechanisms. Based on hei analysis, au ho s
in oduce and e alua e a scheduling amewo k using dynamic
LCP (dLCP). Thei pu pose wi h dLCP is o o e come he
limi a ions o ime in a ian LCP agains he equi emen o
logical channel p io i y upda es a e e e y GCL ime slo
ansi ion. In [246], au ho s ocus on he esou ce alloca ion
con lic p oblem in a 5G-TSN ne wo k whe e he same
adio esou ces a e alloca ed o mul iple TSN lows due
o di e en packe pe iodici ies. Au ho s in oduce a no el
CG scheduling scheme ha elies on he con igu a ion o
se e al g an s pe low acco ding o pe iodici y and o he
cha ac e is ics de i ed om he in o ma ion p o ided by he
TSN ne wo k. Unlike p e ious wo ks ha ocus on 802.1Qb ,
he speci ic TSN mechanism used o acqui e his in o ma ion
is anspa en o he p oposed scheme so long as TSCAI
and 5QI can be de i ed. Thei esul s e lec an inc ease in
he numbe o TSN lows ha can be se ed whils mee -
ing hei la ency equi emen s sa is ac o ily. Finally, au ho s
in [247], ackle his esou ce alloca ion challenge in an uplink
5G-TSN ne wo k by ins ead in oducing an adap i e SPS
algo i hm. TSN lows a e i s di ided in o clus e s acco d-
ing o QoS cha ac e is ics and u ned in o agg ega e lows
acco ding o his classi ica ion. RB p ealloca ion is pe o med
o ime-sensi i e a ic, whe eas non-c i ical spo adic a ic
uses dynamically alloca ed esou ces. The designed algo i hm
ocuses on op imizing he p ealloca ion o esou ces acco ding
292 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
o low equi emen s in o de o maximize sys em esou ce
u iliza ion.
5) Implemen a ion o Reliabili y Mechanisms o e 5G
Ne wo ks: End- o-end eliabili y and obus ness is p edomi-
na ely imp o ed h ough he p o ision o edundan da a pa hs
ha le e age 5G NR and URLLC ea u es, and he FRER
mechanism o he 802.1CB TSN s anda d. In [236], in o de
o duplica e he ansmi ed da a s eams o e 5G, he use o
edundan links be ween base s a ions and UPFs le e aging
URLLC is b ie ly in oduced; howe e , au ho s do no del e
in o he implemen a ion de ails. Spa ial di e si y ia CoMP is
ho oughly e alua ed in [235] as well o p o iding eliabili y
wi hou elying on packe e ansmission. Au ho s e alua e
CoMP echnology o he p o ision o wi eless URLLC
communica ions equi ed o suppo TSN-o e -5G in indus ial
ac o y au oma ion scena ios. Addi ionally, [238] a ge s a
eliabili y imp o emen in he wi eless da a ansmissions
using CoMP based on Mul i-TRP ope a ion. Au ho s o [237]
p opose he use o FRER o p o ide high eliabili y in TSN-
o e -5G communica ions spanning a ac o y TSN domain and
an edge cloud domain. They analyze di e en a chi ec u al
app oaches o deploy FRER unc ionali y in o de o p o ide
end- o-end edundancy be ween an indus ial end de ice con-
nec ed o a TSN ne wo k, and an applica ion ins ance in he
edge cloud. Fo his scena io, hey p opose he deploymen o
FRER unc ionali y o he edge cloud domain side in h ee
possible ways; in a i ualized manne packaged oge he wi h
he TSN Talke /Lis e ne unc ions o he applica ions; i u-
alized, bu in a sepa a e en i y o Talke /Lis ene unc ions; o
implemen ed in a Sma NIC. This wo k is u he su eyed in
Sec ion V-G in ela ion o i ualiza ion o TSN unc ionali ies.
In [248], au ho s p opose expe imen s o e ac ual sma
manu ac u ing es beds o empi ically e alua e he bene i s o
using FRER o e wi eless 5G. In his case, FRER is used o
ansmi eplica ed a ic o e wo independen 5G use plane
pa hs using wo di e en UEs. They highligh se e al bene i s
ha FRER can p o ide o wi eless ne wo ks such as 5G:
p o ec ion agains ne wo k node ailu es; p o ec ion agains UE
ailu es o UE pe o mance a ia ions; and obus ness agains
wi eless links wi h la ency a ia ions. Au ho s ocus on e alu-
a ing end- o-end la ency obus ness imp o emen s in oduced
by FRER when 5G URLLC unc ionali y is no a ailable,
whe e hey ind ha gua an eed la ency can be educed by
up o se e al milliseconds. Likewise, in [249], au ho s also
in es iga e FRER as a solu ion o he lack o URLLC
a ailabili y in comme cial p oduc s in o de o imp o e wo s -
case la ency in addi ion o signi ican ly inc easing eliabili y.
Thei pe o mance e alua ion in a p oduc ion en i onmen
shows ha hei edundan pa h implemen a ion using FRER
p o ides highe eliabili y agains ou ages. Addi ionally, i
imp o es he o e all la ency, as i mi iga es he e ec o he
addi ional delay in oduced by HARQ e ansmissions in each
pa h.
6) T a ic S ee ing Op imiza ion in Use Equipmen : Wi h
a o emen ioned wo ks ocusing on he a chi ec u e in eg a ion
and e alua ion wi hin he RAN and Co e Ne wo k le els o
he 5GS b idge, in [250], au ho s ocus ins ead on he delay
minimiza ion o a ic s ee ing in edge compu e de ices ha
include he UE componen o he 5GS b idge. Au ho s p opose
he use o XDP p og amming o educe he la ency in oduced
by he ne wo k s ack in he de ice when s ee ing E he ne PDU
and gPTP a ic o he TSN NIC o o use space applica ions,
and o he DS-TT, espec i ely. Thei p oposed algo i hm il-
e s he a ic acco ding o des ina ion MAC add ess. Packe s
wi h a gPTP mul icas des ina ion add ess a e edi ec ed o he
DS-TT in use space bypassing he ke nel ne wo k s ack o he
de ice. Au ho s conduc an expe imen al e alua ion o hei
XDP algo i hm, a i ing o an s ee ing la ency o PDU a ic
100 imes lowe han wi h a con en ional laye 2 so wa e
b idge.
Summa y and Lessons Lea ned:
•The in eg a ion o 5G and TSN is empowe ed by 5G
a chi ec u al enabling echnologies and URLLC:As
demons a ed by se e al p elimina y TSN and 5G in e-
g a ion s udies in indus ial en i onmen s, he possibili y
o in oducing a 5G ne wo k in o an indus ial TSN
deploymen begins wi h he adop ion o a chi ec u al
enable s like p i a e 5G ne wo ks and edge compu -
ing. A chi ec u e-le el echnologies highligh ed in he
li e a u e include he a ailabili y o se e al 5G NPN
con igu a ions, edundan ansmissions based on Dual
Connec i i y, Mul i-TRP, and/o URLLC, he suppo o
a ic s ee ing o local da a ne wo ks, and 5G LAN
echnology o suppo he g ouping o de ices ha can
communica e wi h each o he as hough in a Local A ea
Ne wo k. As such, 5G-TSN is expec ed o h i e in his
kind o echnology ecosys em o sa is y he equi emen s
o lexible and inno a i e indus ial applica ions.
•The cha ac e iza ion o TSN a ic equi emen s in 5G
QoS language is a undamen al equi emen : As he 5GS
does no unde s and con igu a ion pa ame e s o speci ic
TSN unc ionali ies, ensu ing ha i is awa e o he
cha ac e is ics o TSN lows a e sing i is undamen al.
3GPP es ablishes guidelines o es ima ing he b idge
delay and o de i ing 5G QoS pa ame e s like PDB,
GFBR, and MDBV, bu implemen a ion de ails a e no
ho oughly elabo a ed, and, addi ionally, depend on he
end- o-end communica ion use case. Thus, e en when no
accoun ing o ac o s such as b idge delay es ima ion
accu acy o he ep esen a ion o a ic pe iodici y wi h
5G mechanisms, he mapping o QoS pa ame e s be ween
he wi ed TSN ne wo k and he 5GS equi es ca e ul
inke ing and, possibly, he use o 5G QoS con igu a ion,
like 5QI, ha is ailo ed o he speci ic use case a he
han depending on ca ch-all s anda dized 5G QoS p o-
iles.
•The 5G-TSN in eg a ion mus also mee s ingen
pe o mance equi emen s: While 3GPP s anda ds de ail
he a chi ec u e-le el in eg a ion o 5G and TSN, achie -
ing a pe o man in eg a ion equi es no only he suppo ,
bu also he op imal execu ion, o 5G NR and URLLC
ea u es. Wi h some indus ial applica ions demanding
a deli e y ime o e en less han 1 ms, bounding he
delay con ibu ion o all elemen s ac oss he end- o-
end pa h is c i ical. In his ega d, one o he main
challenges is he la ency and ji e minimiza ion o e he
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 293
wi eless 5G medium ( ha is, ac oss he 5G RAN) in
o de o gua an ee he bounded la ency equi emen s o
he TSN lows. Many s udies highligh ha i ing TSN
pe iodici y, such as dic a ed by a 802.1Qb schedule, in o
he 5G scheduling mechanisms based on OFDM, en ails
a ma ching p oblem ha mus be ca e ully add essed.
Among hese issues, he unp edic able condi ions o he
wi eless medium, he po en ial misma ch be ween a ic
pe iodici y alues and 5G ansmission in e als, and
po en ial con lic s du ing he scheduling in he 5GS
o mul iple TSN lows wi h di e en pe iodici ies, a e
highligh ed in he li e a u e.
•Con igu a ions based on 5G edundan use plane ans-
mission can be e ec i ely in eg a ed wi h TSN FRER o
imp o e esilience and wo s -case la ency: In ega ds o
speci ically achie ing he equi ed eliabili y and obus -
ness o he a chi ec u e, one esea ch line in es iga es
he usage o CoMP op ions and Mul i-TRP o spa-
ial di e si y, while ano he esea ch line analyzes he
se up o edundan use plane pa hs h ough NR-DC o
h ough edundan unnels be ween RAN and UPF. The
o me is mo e closely linked o inc easing capaci y and
ansmission a e, whe eas he la e is mo e pe inen o
imp o e eliabili y as i also p o ides he a chi ec u al
suppo o implemen FRER. In his ega d, FRER is
ound o be e ec i e a p o iding addi ional obus ness in
communica ion o e 5GS b idges, wi h he mos c i ical
pa h gene ally conside ed o be he link be ween UE and
gNB, due o i s wi eless na u e. This kind o edundancy
is ound o imp o e wo s -case la encies as a conse-
quence o duplica ing ansmissions h ough edundan
wi eless pa hs and disca ding edundan copies a he
eg ess poin . The wo ks add essing hese con igu a ions
mainly p opose he duplica ion o UEs, a he han he
duplica ion o only he numbe o PDU sessions ha
a single UE es ablishes h ough edundan RANs and
UPFs. This p o ides esilience agains UE ailu e as well,
bu implies duplica ion o he UE esou ces connec ed o
he TSN end s a ions.
•The op imiza ion o TSN scheduling and pe o mance
o e 5G is an open esea ch ques ion ha p og esses
in ha mony wi h he s anda diza ion o new 5G and
URLLC capabili ies: The li e a u e showcases a a ie y
o p oposals o add ess he esou ce alloca ion and
scheduling challenge in 5G. The mos basic app oach
in ol es he ba e use o p econ igu ed ansmission slo s
wi h CG and SPS, which p o ides some gua an ees
h ough he p ealloca ion o esou ces o TSN lows
in compa ison wi h he de aul dynamic scheduling.
Howe e , issues such as he ones poin ed ou in he
p e ious i em need o be mo e e ec i ely add essed. Fo
he pe iodici y misma ch p oblem, one wo k p oposes he
use o a scheduling scheme wi h al e na ing CGs as o
adap o low dynamici y. Ano he wo k looks highe in
he 5G NR s ack, e alua ing he use o dynamic LCP
o he ime- a ian p io i iza ion o logical channels in
he RLC laye o suppo TAS-based TSN. They a e
p omising app oaches o dynamic uplink TSN lows and
o TAS GCLs ha accommoda e he e ogeneous a ic,
espec i ely. Fo he mi iga ion o he esou ce alloca ion
con lic p oblem, one wo k p oposes a scheduling scheme
based on he p o isioning o se e al CGs o each TSN
low o ensu e ha RBs a e no alloca ed o mo e han one
low. Ano he wo k p oposes an adap i e SPS algo i hm
o e agg ega ed TSN lows ha ea s ime-sensi i e and
bes e o a ic di e en ly in o de o op imize he
usage o esou ces. These examples also highligh ha
he e can be di e en op imiza ion goals in he design o
he 5G-TSN ne wo k, anging om TSN low admission
o 5G esou ce usage, which expands e en mo e he
esea ch space. Finally, ano he pape looks a imp o ing
a ic s ee ing delay in he UE h ough use space packe
p ocessing in he o m o XDP. This con ibu ion is
complemen a y o he o he ones in he sense ha he 5G-
TSN in eg a ion shall be app oached om an end- o-end
pe spec i e, including he 5G RAN, he 5G UEs, and he
5G TN, he la e o which is su eyed in Sec ion V-E.
C. QoS Con igu a ion and Mapping S a egies o TSN
Flows O e 5G
This subsec ion ga he s esea ch wo ks p ima ily conce ned
wi h a chi ec u e and p ocedu al aspec s o he 5GS con igu a-
ion and mapping o TSN a ic, as opposed o hose schemes
ocused on pe o mance cha ac e iza ion, which we e su eyed
in Sec ion V-B. S ill, i should be no ed ha he dis inc ion in
some cases may be na ow. A axonomy o his subsec ion is
p oposed in Table XV. Se e al app oaches o mapping TSN
and 5G QoS can be iden i ied, whe eas o he p oposals can
be classi ied unde ne wo k op imiza ion a a b oade end-
o-end ne wo k managemen le el and o en employ lea ning
schemes, such as Rein o ced Lea ning (RL)-based.
1) QoS Mapping and T ansla ion be ween TSN and 5G:
Au ho s o [34] del e in o how a ic class, ame p eemp ion,
and 5G QoS con igu a ion can be join ly ca ied ou in a
5G-TSN in eg a ed ne wo k, acco ding o IEEE TSN and
3GPP s anda ds. Based on he mechanisms and capabili ies
p o ided by he 5GS, au ho s sugges pa ame e mappings
and cons ain s o con igu ing 5G QoS P o iles. In [251],
au ho s design and implemen , ia simula ion, a ansla o ool
be ween TSN and 5G a ic ha is able o map pa ame e s
be ween he wo echnologies acco ding o he 3GPP speci i-
ca ions. In hei wo k, hey desc ibe hei implemen a ion o
he ansla ion wo k lows in he simula o , and inally e alua e
he esul using an indus ial au omo i e use case. They ind
ha hei ool can be use ul o e alua ing he pe o mance
o 5G-TSN ne wo k con igu a ions, pa icula ly in ega ds o
channel p opaga ion delays. Au ho s in [252] p opose and
e alua e a QoS-awa e algo i hm ha sys ema ically maps
a ic lows be ween he TSN and 5G ne wo k domains.
Thei algo i hm uses as inpu he equi emen s o applica ions
wi h speci ic cons ain s ha include deadline, ji e , and
bandwid h. Each applica ion is mapped o a 5G Resou ce
Type acco ding o he bina y equi emen s o hese h ee
ypes o cons ain s. The nume ical alues o bandwid h and
deadline, in addi ion o he applica ion’s PER equi emen ,
294 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
TABLE XIV
TAXONOMY RELATED TO THE QOSCONFIGURATION AND MAPPING STRATEGIES FOR TSN FLOWS OVER 5G
TABLE XV
TAXONOMY OF TIME SYNCHRONIZATION SCHEMES IN 5G-TSN NETWORKS
a e used o calcula e he mapping o a s anda dized 5G QFI
o he gi en Resou ce Type. In [247], au ho s p opose a
dynamic and load-awa e QoS mapping me hod o 5G-TSN
ne wo ks. A clus e ing-based algo i hm ha agg ega es lows
wi h simila QoS cha ac e is ics is designed o educe QoS
mapping complexi y whils ensu ing he consis ency o QoS
classes. Addi ionally, au ho s in oduce an algo i hm ha maps
lows o se e al 5QIs conside ing ne wo k load o imp o e
esou ce alloca ion e iciency.
2) Injec ion o Scheduling Con igu a ion in 5G-TSN
Ne wo ks: Bo h TAS and PSFP mechanisms a e explo ed
in [253], as au ho s p opose an he e ogeneous a chi ec u e
in eg a ing 5G and TSN. The analyzed indus ial scena io
comp ises a se o wi eless ield de ices ha connec o a TSN
GW owa ds a wi ed ne wo k ia 5G. On he one hand, au ho s
design a scheduling mechanism ha can ese e 5G esou ces
in ad ance using SPS o omi signaling delays o he highes -
p io i y spo adic eme gency e en s. On he o he hand, o
o se he ji e in oduced by he 5G ne wo k, hey p opose he
use o PSFP a he ing ess o he wi ed TSN ne wo k. Au ho s
do so aking in o accoun he delay expe ienced by he da a in
he 5G ne wo k oge he wi h i s end- o-end delay cons ain s
o assign a sui able p io i y. As such, unlike o he p e iously
co e ed wo ks, hey concen a e on con igu a ion injec ion in
he di ec ion om 5G o TSN. In [254], au ho s s udy a simila
indus ial scena io ex ending he con igu a ion challenge o
he NFV domain as well, as hey o mula e a mapping and
scheduling p oblem o 5G-TSN esou ces o incoming VNFs.
Admi ed VNFs a e deployed in TSN GWs loca ed in an edge
laye ha also con ains 5G base s a ions. The communica ion
lows occu be ween VNFs and de ices in a ield laye
ha connec ia 5G o he TSN GWs. The a i al ime o
VNFs is mapped o 5G slo s. Apa om GCL compu a ion
o he VNF s eams, au ho s also apply p eemp ion-based
scheduling o he a i ing VNFs accoun ing o hei p io i y.
Au ho s o [255] add ess he challenge o g anula ly mapping
a ic in a con e ged 5G-TSN ne wo k, p ima ily ocusing
on an a chi ec u al model o injec he con igu a ion. Au ho s
p opose o mo e he mapping unc ionali y om he TSN AF
in he 5GS—as de ined in 3GPP s anda ds— o a cen alized
CNC, aking ad an age o i s global ne wo k knowledge. In he
p oposed a chi ec u e, which con ains a da a plane comp ised
o one o mo e 5GS and/o non-TSN b idges, he CNC cal-
cula es he 5QIs in each 5GS b idge o he applica ion lows
be ween end s a ions. 5QIs a e de i ed om o deadline, ji e ,
bandwid h, and PER equi emen s ecei ed om he CUC.
Finally, he CNC communica es he 5QIs o he co esponding
PCFs and join ly con igu es 5GS and non-5GS b idges in he
ne wo k.
3) Join con igu a ion and scheduling o e 5G and TSN
esou ces: In [256], au ho s pu sue he op imiza ion o end- o-
end scheduling in a 5G-TSN ne wo k ca ying he e ogeneous
a ic h ough a join con igu a ion app oach. Fi s ly, au ho s
de i e global and echnology-speci ic cons ain s ha he
scheduling s a egy mus adhe e o. This includes, o example,
5G ansmission cons ain s acco ding o 5G NR TTIs and
RBs. TSN cons ain s conside scheduling based on 802.1Qb
TAS. Based on hese cons ain s, au ho s ackle end- o-end
schedule op imiza ion ac oss he di e en domains, ocusing
on esou ce u iliza ion and suppo o BE a ic, gi en ha he
ga ing mechanisms o TAS gua an ee ime-bounded la ency
equi emen s. Some ecen esea ch wo ks adop lea ning
me hods o he con igu a ion o 5G-TSN ne wo ks acco ding
o cons ain s and op imiza ion goals. In [257] and [258],
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 295
au ho s aim o a dynamic and ze o- ouch con igu a ion o
a 5G NPN ea u ing 5G-TSN in eg a ion based on he ZSM
e e ence model. The su ey o hese wo ks is ca ied ou
in Sec ion V-H. No ably, hei p oposed sma con olle is
designed o ac as a TSN AF ha dynamically con igu es
each 5G QoS low acco ding o compliance wi h expec ed
KPIs. In [259], au ho s p esen an algo i hm based on RL—
he Q-lea ning echnique speci ically—and on pa icle swa m
op imiza ion ha aims o i e a i ely a i e o he con igu a ion
ha maximizes he ime-sensi i e low admission a io in 5G-
TSN ne wo ks. Au ho s use CQF as he TSN a ic shaping
mechanism and es ablish 5GS la ency cons ain s, end- o-end
la ency cons ain s, and CQF ime slo cons ain s. Finally,
au ho s conduc an e alua ion o hei p oposed algo i hm
h ough he simula ion o 5G-TSN ne wo k opologies wi h
speci ic 5G nume ology pa ame e s and assuming semi-s a ic
scheduling. Likewise, in [260], au ho s in es iga e he usage
o CQF in 5G-TSN ne wo ks. Au ho s conside cons ain s
ela ed o CQF and o i s mapping o 5G esou ces, including
low pe iodici y and la ency in he 5GS in ega ds o sa is ying
he scheduling ules o CQF ime in e als. Based on hese
cons ain s, au ho s o mula e low schedulabili y and esou ce
u iliza ion as join op imiza ion objec i es. To add ess he
o mula ed p oblem, au ho s p opose a DRL-based algo i hm
designed o lea n he ela ionship be ween di e en ime
in e als and he u iliza ion o 5G sub-ca ie s, and inally
conduc a simula ion o e alua e i s pe o mance. In ega ds o
he use o lea ning me hods, i is wo h poin ing ou p e iously
su eyed wo k [253] o i s con ibu ion o a isk-sensi i e
lea ning algo i hm aimed a educing he wo s -case delay in
PSFP-based 5G-TSN ne wo ks by adjus ing ese ed RBs and
PSFP p io i ies.
Summa y and Lessons Lea ned:
•The e a e di e en possible TSN mechanisms o ex end
TSN o he 5G domain: When a TSN ne wo k in eg a es
5GS b idges, he li e a u e p oposes 802.1Qb TAS,
802.1Qci PSFP, and 802.1Qch CQF as iable TSN mech-
anism choices; ne e heless, u he obse a ions can be
made. The only p oposal conside ing PSFP employs i in
o de o dynamically manage he injec ion o TSN lows
coming om a 5GS b idge in o he en y poin o a wi ed
TSN ne wo k. In his way, he policing unc ions o PSFP
a e ound use ul o balance he assigned queue p io i ies
agains he ji e in oduced by he 5G ne wo k o each
low. O e all, TAS is he mos popula choice. When a
5GS ac s as a i ual TSN b idge, he concep o GCL
can be ex ended na u ally o he 5G domain o he join
con igu a ion o bo h domains. The main challenges o
his app oach ega ding 5G scheduling we e discussed in
Sec ion V-B. In spi e o 3GPP s anda diza ion sugges ing
he use o PSFP’s s eam iden i ica ion and ga e-based
policing con igu a ion o de i e 5G QoS o TSN lows,
se e al wo ks ha e p e e ed o use cha ac e is ics om
802.1Qb o map TSN QoS o 5G QoS. Howe e , he
QoS cha ac e is ics ha can be de i ed om a GCL a e
po en ially less g anula , so, o example, one p oposal
o a QoS mapping algo i hm has ecou sed o explici ly
p o iding addi ional applica ion equi emen s. Hence, i
could be wo hwhile o conside he use o PSFP o
accu a e QoS and TSCAI de i a ion, e en i a TAS
GCL is also used. Meanwhile, he p oposals o using
CQF suppo i o i s simpli ica ion o he scheduling in
compa ison o TAS, a oiding complex GCL compu a ion,
and allowing he accommoda ion o se e al lows in he
same ime slo . Wi h CQF, he imposed TSN scheduling
cons ain is simpli ied o inding a ime slo leng h ha
sa is ies he esidence ime in he 5GS b idge. Howe e ,
as he a icles ha p opose CQF ocus on maximizing
low admission, mo e esea ch is pe inen ega ding i
he end- o-end low la ency a ained wi h CQF is adequa e
enough o sa is y s ingen indus ial communica ion
equi emen s. Indeed, se e al sou ces, including he TSN-
IA as seen in Sec ion IV, sugges ha isoch onous
and cyclic-synch onous a ic ypes equi e he use o
TAS. S ill, gi en ha CQF gua an ees a ixed wo s -case
delay so long as cons ain s a e me , i s use in 5G-TSN
ne wo ks is p omising in he cases whe e he ime slo
leng h sa is ies he end- o-end la ency equi emen s o he
in ol ed a ic.
•Cogni i e app oaches can be adop ed o op imize he
con igu a ion o 5G-TSN ne wo ks: Se e al wo ks o mu-
la e he con igu a ion o he 5G-TSN as an op imiza ion
p oblem wi h end- o-end cons ain s ela ed o la ency,
pe iodici y, and 5G nume ology, in which he numbe
o scheduled lows, ei he alone o join ly wi h he
esou ce usage e iciency, aims o be maximized. O he
p oposals con ey he p oblem in e sely, aking a 5G-TSN
ne wo k wi h con igu ed QoS lows, and dynamically
applying closed-loop econ igu a ion in o de o op imize
he ne wo k acco ding o he KPIs ha cha ac e ize he
QoS lows. In all cases, au ho s ind RL echniques use ul
o his pu pose, as hey cons i u e AI/ML echniques
sui ed o lea n by obse ing he en i onmen and o ind
he mos adequa e con igu a ion o se e al pa ame e s
ha maximizes he gi en goals. RL a ian s ound in
he a ious algo i hms p oposed in he li e a u e include
DRL, au oma a lea ning, isk-sensi i e lea ning, and Q-
lea ning. Fu he mo e, wo ks ha p opose a closed-loop
managemen o he 5G-TSN ne wo k p opose app oaches
based on he ZSM a chi ec u e, as i p o ides a ame-
wo k ha in eg a es moni o ing, lea ning, p edic ion,
and econ igu a ion capabili ies. Mo e esea ch can be
expec ed going o wa d in he scope o applying lea ning
mechanisms o he design and op imiza ion o 5G-TSN
ne wo ks, especially as mos o he cu en p oposals a e
s ill in heo e ical o simula ion s age.
D. E alua ion o Time Synch oniza ion in 5G-TSN Ne wo ks
Highly accu a e ime synch oniza ion be ween he 5G and
TSN clock domains is an essen ial equi emen o allow o
ime-sensi i e indus ial con ol communica ions ac oss a 5GS
b idge. Resea ch wo ks add essing ime synch oniza ion in
5G-TSN ne wo ks a e su eyed in his subsec ion. As seen in
Table XVI, con ibu ions in he scope o ime synch oniza ion

296 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
TABLE XVI
TAXONOMY RELATED TO THE EVALUATION OF TIME SYNCHRONIZATION IN 5G-TSN NETWORKS
ocus on he e alua ion and op imiza ion o synch oniza-
ion accu acy employing PDC and/o clock d i mi iga ion
schemes. A a ie y o mechanisms, s a egies, and in ol ed
ac o s can be iden i ied.
1) Analysis o O e - he-ai Time Synch oniza ion: P io
o he consolida ion o 3GPP Release 16, a mechanism o
802.1AS-based ime synch oniza ion wi h mobile ne wo ks
is p esen ed in [261], ha is also compa ible wi h 4G.
They le e age he ac ha inne synch oniza ion mechanisms
be ween base s a ion and UEs a e al eady equi ed in 4G and
5G ne wo ks. To synch onize each UE o TSN ime, au ho s
p opose he calcula ion o a ime o se be ween he iming
in o ma ion ecei ed in he base s a ion and he ime o a
Re e ence UE ha is wi edly synch onized o he TSN ne wo k
using 802.1AS. This o se can be dis ibu ed o he UEs ia
a mul icas OPC UA se ice ha he UEs subsc ibe o. They
achie e a 0.55 ms synch oniza ion accu acy o e a 4G es bed.
A join compa a i e analysis o 802.1AS o e Wi-Fi and o e
5G is conduc ed in [262], seeking pe o mance esul s and
he iden i ica ion o ac o s ha impac such pe o mance.
They highligh he asymme ic and medium sha ing na u e
o wi eless links and he lack o ha dwa e imes amping
suppo in wi eless ca ds as pa icula challenges o delay
es ima ion. Mos speci ically o 5G, hey ema k ha he
3GPP a chi ec u e based on TSN ansla o s can b ing c i ical
pe o mance imp o emen s.
2) E alua ion o Time Synch oniza ion Based on 3GPP
A chi ec u e: Se e al wo ks can be ound in he li e a u e
ha make use o he TSN ime synch oniza ion p ocedu es
s anda dized om Release 16 onwa ds. P ecisely, in [263],
au ho s pe o m an exhaus i e analysis o 802.1AS ime
synch oniza ion o con e ged wi ed and wi eless ne wo ks,
a ge ing 5G-TSN in eg a ed ne wo ks. Au ho s model he
beha io o 802.1AS mechanisms and in es iga e synch oniza-
ion accu acy and wo s -case pa ame e s aking in o accoun
equi emen s speci ied by he IEEE 802.1AS s anda d i sel ,
he IEEE/IEC TSN-IA p o ile, and 3GPP. Thei in es iga ion
iden i ies limi ed imes amp esolu ion, physical laye ji e ,
and conca ena ion o ime-awa e sys ems as he p ima y
sou ces o ime synch oniza ion inaccu acy. In [264], au ho s
ca y ou an e alua ion o he TSN ime synch oniza ion
mechanism in oduced in Release 16, using he 802.1AS
ime-awa e sys em model suppo ed in his elease o 3GPP.
They e alua e he ime synch oniza ion accu acy in he p es-
ence o po en ial sou ces o inaccu acy, such as clock d i s
and di e en ai -in e ace e o s associa ed wi h e e ence
ime indica ion. They speci ically e alua e he condi ions ha
need o be me o achie e below 1 µs accu acy. In ega ds
o scalabili y, au ho s also calcula e ha a maximum o 8
TSN domains can be suppo ed. Au ho s o [265] conside a
synch oniza ion scena io o 5G-TSN ne wo ks ha employs
nei he downlink synch oniza ion om he TSN ne wo k, no
uplink synch oniza ion om he UE. Ins ead, au ho s p esen
and e alua e he use o a sha ed GM clock o bo h 5G
and TSN ne wo ks ha is loca ed wi hin he 5GS b idge a
he gNB and has access o a GPS sou ce o ime. Au ho s
iden i y challenges ela ed o he mapping be ween gPTP and
5G RAN synch oniza ion mechanisms and o he execu ion
o he BMC algo i hm, and p opose wo s a egies o add ess
hem: a shallow in eg a ion ha main ains he bounda ies in
NW-TT and DS-TT aligned wi h 3GPP s anda ds, and a mo e
anspa en deep in eg a ion using a modi ied 5GS b idge ha
shi s he synch oniza ion bounda ies o gNB and UE.
3) E alua ion o F equency O se and Clock D i : I has
been ound ha he e alua ion o equency o se s and esul -
ing clock d i s is one o he main esea ch a eas wi hin he
scope o ime synch oniza ion in 5G-TSN ne wo ks acco d-
ing o 3GPP s anda ds. Indeed, a o emen ioned wo k [264]
analyzes he e ec s o clock d i o e he 5G ai in e ace
be ween UE and gNB. Some akeaways a e: highe SCS is
mo e obus o cumula i e synch oniza ion e o ; e e ence
ime indica ion g anula i y mus be in he o de o 10 ns;
and equen ime synch oniza ion is key agains clock d i .
in [266], au ho s conduc a pe o mance e alua ion o he
Release 16 TSN ime synch oniza ion scheme, ocusing speci -
ically on es ima ing he impac o he NW-TT, DS-TT, and
5G GM a e a ios, i.e., equency o se s. Thei measu emen s
conclude ha he main con ibu ion o ime synch oniza ion
e o is dependan on he equency o se be ween 5G GM
and DS-TT, which in ol es he ai in e ace delay. To mi iga e
hese e o s, au ho s sugges a me hod ha akes ad an age
o he ac ha he 5G GM is dis ibu ed o he UE in
o de o compu e he a e a io be ween DS-TT and 5G GM.
Meanwhile, he au ho s in [267] adop Release 17 s anda ds
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 297
as hey e alua e ime synch oniza ion pe o mance, modeling
he 5GS as a T anspa en Clock, and ca ying ou a de ice- o-
de ice synch oniza ion analysis. Au ho s p esen a simula ed
scena io o wo coope a i e mobile obo s and execu e i wi h
di e en pa ame iza ions. Thei esul s highligh ha e en
small equency o se s be ween NW-TT and DS-TT can ha e
a signi ican impac on he calcula ion o he esidence ime
inside he 5GS b idge. They conclude ha , ul ima ely, his can
esul in signi ican synch oniza ion e o s.
4) P opaga ion Delay Es ima ion and Compensa ion
using TA-based o RTT-based me hods: Compa ed o wi ed
ne wo ks, he link p opaga ion delay in a wi eless ne wo k
is much mo e challenging o es ima e, due o a iable ans-
mission cha ac e is ics ha also depend on UE posi ion and
mo emen . As such, some esea ch wo ks speci ically con-
cen a e on his equi emen o 5G-TSN ime synch oniza ion.
Re e ence [268] p esen s an analy ical e alua ion o ime
synch oniza ion e o sou ces ac oss he end- o-end pa h o a
5GS in he con ex o in eg a ed 5G-TSN ne wo ks. Au ho s
analyze how a 5GS modeled as a 802.1AS ime-awa e sys em
can be designed o suppo a 900 ns ime synch oniza ion
e o budge . Au ho s spli hei analysis be ween anspo
and adio ne wo k pa hs. Fo he o me , hey conclude ha
ITU-T ime synch oniza ion s anda ds [269] a e adequa e o
indus ial ne wo k scena ios. Fo he la e , hey analyze RTT-
based and TA-based p opaga ion delay es ima ion me hods and
ind he o me o achie e a lowe ime e o . Likewise, [35]
examines hese wo me hods o PDC acco ding o he ime
synch oniza ion e o budge es ablished in Release 17, aking
in o accoun ha UE-UE communica ions equi e a wice as
igh unidi ec ional synch oniza ion budge . Meanwhile, [270]
p oposes and e alua es a p ocedu e based on UE localiza ion
es ima ion using TA in o de o achie e a mo e accu a e
PDC. Mo e speci ically, hei p oposed mechanism is based on
ob aining TA measu emen s om mul iple gNBs a he han
jus om he one se ing he UE. This esul s in enhanced TA
g anula i y, which leads o a mo e p ecise es ima ion o he
UE localiza ion and i s dis ance o he gNB.
5) P opaga ion Delay Es ima ion and Compensa ion using
o he me hods:
[271] ega ds he 5GS as a T anspa en Clock and adop s a
delay compensa ion me hod based on da a packe elay whe e
he 5G ne wo k is only esponsible o he o wa ding o
imes amp messages, o loading he complexi y o ansla ing
be ween ime domains. The esidence ime o TSN iming
messages o wa ded h ough he 5GS in a PDU session is
es ima ed using a clock domain compensa ion algo i hm ha
uses imes amps in he NW-TT and DS-TT. A delay compen-
sa ion me hod based on Time Di e ence o A i al (TDoA)
is b ie ly p esen ed in pos e [272]. Au ho s p esen an algo-
i hm ha ocuses on o e coming he non-line-o -sigh e o .
Simula ing a ne wo k wi h 7 gNBs, au ho s ob ain a 8.9% ime
synch oniza ion accu acy imp o emen compa ed o he RTT-
based me hod. Finally, in [273], au ho s expand he p oblem
o mula ion o a scena io wi h mul iple 5GS b idges ha
accumula e synch oniza ion e o s, eyeing he use o 5G-TSN
in mul i- ield manu ac u ing sys ems scena ios. Times amp
compensa ion be ween each 5GS b idge is pe o med wi h
a me hod based on cumula i e a e a ios. Au ho s e alua e
he impac o SCS in he synch oniza ion e o , and, ia
simula ion, a i e o 5G-TSN clock synch oniza ion accu acy
esul s o di e en SCS alues and numbe o 5GS b idges in
he ne wo k. Based on hese esul s, au ho s p opose an SCS
op imiza ion scheme ha in eg a es synch oniza ion accu acy,
and bandwid h u iliza ion de e mined by SCS.
Summa y and Lessons Lea ned:
•Ve y accu a e end- o-end ime synch oniza ion is equi ed
in he 5GS o suppo wi eless TSN communica ions:
Many applica ions using TSN equi e highly accu a e
ime synch oniza ion in he sub-mic osecond ange.
Fu he mo e, wi h he in oduc ion o ime-sensi i e
UE-UE communica ions in Release 17, bidi ec ional
synch oniza ion in ol ing wo UE-UPF pa hs in o al is
equi ed. The li e a u e iden i ies he minimiza ion o he
con ibu ion o he 5GS o he end- o-end synch oniza ion
e o as a key challenge. I has been ound ha he
main con ibu ions in his ega d can be classi ied in
wo complemen a y ca ego ies: he es ima ion o he
p opaga ion delay in he 5GS and me hods o i s com-
pensa ion, and he mi iga ion o he di e ence in clock
equency o se s—and esul ing clock d i s—in he 5GS
ha in oduces e o s in he esidence ime compu a ion.
•The in e al be ween ime synch oniza ion e en s is a
decisi e ac o o he e o caused by clock d i s inside
he 5GS: The li e a u e gene ally ag ees ha he bes
solu ion o p e en d i ing be ween he clocks a he
ing ess and eg ess o he 5GS, o be ween he clocks
o he UE and he gNB when ocusing on he wi eless
pa h, is o execu e equen e-synch oniza ion p oce-
du es, as o he wise e en small equency d i s can ha e
a signi ican impac on he esidence ime calcula ion
e o . Some esea ch wo ks analyze ime synch oniza ion
in e als in he anges be ween ens o milliseconds o
hund eds o milliseconds, and he esul s indica e ha
he clock equency o se inc eases linea ly wi h he
synch oniza ion in e al, ega dless o he SCS in use.
On he o he hand, bu o a lesse ex en , a highe SCS
has been ound o lead o lowe equency o se be ween
clocks o he same in e al be ween ime synch oniza ion
e en s. Complemen a ily, he equency o se be ween
TSN GM and 5G GM can be es ima ed in o de o
mi iga e i . The DS-TT could di ec ly make use o
he 5G GM ime acqui ed by he UE—which can be
made mos accu a e h ough equen e-synch oniza ion
e en s as a o emen ioned—and calcula e i s a e a io
wi h he UE, and he same o NW-TT and UPF. The
gPTP p ocedu e ca ied ou h ough he 5GS b idge is
main ained in o de o calcula e he esidence ime in
he 5GS, excep i ins ead le e ages hese a e a io
calcula ions o he con e sion om 5G o TSN ime.
Hence, bo h minimiza ion o he d i ing be ween he
clocks in he 5GS, and he de i a ion o he a e a io
be ween 5G and TSN ime domains, shall be aken in o
conside a ion o a i e o he mos sa is ac o y esul .
•The op imiza ion o he p opaga ion delay es ima ion
and compensa ion mechanisms is c i ical o he ime
298 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
TABLE XVII
TAXONOMY RELATED TO THE USAGE OF TSN IN 5G TRANSPORT NETWORKS
synch oniza ion accu acy: The wo main mechanisms o
es ima e and compensa e he p opaga ion delay be ween
UE and gNB s udied by he esea ch communi y a e he
TA-based me hod and he RTT-based me hod, aligned
wi h 3GPP s anda ds. O e all, he esul s a e mo e
a ou able o he la e , which was speci ically in o-
duced in 3GPP Release 17 o TSC. Fu he mo e, some
au ho s ag ee ha he RTT-based me hod is mo e p omis-
ing due o ha ing mo e oom o op imiza ions. A
heo e ical s udy epo s an imp o emen om 514 ns o
188 ns om TA-based o RTT-based es ima ion. Ano he
s udy p esen s se e al samples o be ween 276 ns and
536 ms depending on he SCS, wi h highe SCS alues
achie ing be e accu acy, bu , in his case, he es ima ion
me hod used is no speci ied. On he o he hand, ano he
wo k p oposes a TA-based app oach ha , by ecei ing
measu emen s om mo e han one gNB, accu acy can
be imp o ed by up o a ound 55 ns. In ano he b ie
s udy ha in ol es mul iple gNBs, he use o TDoA,
no mally used o posi ioning, yields p omising esul s,
bu mo e esea ch is necessa y, especially wi h a lowe
amoun o gNBs. Meanwhile, a me hod ha ea s he
5GS as a T anspa en Clock and is only esponsible
o he o wa ding o imes amped messages in PDU
sessions is ound e ec i e a educing he complexi y
o TSN ansla o s, bu esul s indica e synch oniza ion
accu acy in he mic osecond le el. I is no ewo hy ha
while he e a e se e al ac o s ha con ibu e o he
es ima ion accu acy (e.g., he PDC me hod, he SCS, he
UE mobili y including i s speed, he ne wo k size, he
g anula i y o he epo ed ime e e ence, whe he i is
he UE o he gNB who pe o ms he PDC, e c.), he
su eyed wo ks do no always analyze o epo all o
hese ac o s, so mo e esea ch is a guably necessa y.
An impo an conside a ion is he synch oniza ion e o
budge equi emen s es ablished in 3GPP Releases 16
and 17—900 ns, bu 450 ns o he UE-UE scena io—
, which may dic a e he alidi y o di e en me hods
and con igu a ions. Ano he open ques ion consis s on
de e mining he UE-gNB dis ance h eshold, pe haps
accoun ing o o he ai -in e ace ac o s, ha wa an s
he use o PDC, as, o low dis ances, he p opaga ion
delay could be conside ed negligible.
E. TSN o 5G T anspo Ne wo ks
This subsec ion p esen s esea ch wo ks ha analyze he
usage o TSN unc ionali ies in 5G TNs, and concen a es
speci ically on he con ibu ions p o ided by he use o TSN.
Like he o me subsec ions, his subsec ion su eys ela ed
wo ks published om 2018 onwa ds, aligning in his case wi h
he publica ion da e o 802.1CM. As e lec ed by Table XVII,
he li e a u e showcases esea ch wo ks ha ackle, among
o he aspec s, la ency and ji e minimiza ion in he on haul,
and TN con e gence by b inging TSN o he backhaul and
c osshaul as well.
1) Delay and Ji e Minimiza ion o F on haul T a ic:
In [274], au ho s i s analyze he delay equi emen s in
eCPRI-based 5G on haul ne wo ks and wha a e he com-
ponen s ha con ibu e o end- o-end delay be ween BBU
and RRH. Then, au ho s analyze TSN unc ionali y and he
app op ia e design o on haul ne wo ks using such unc-
ionali y in o de o comply o such equi emen s. Au ho s
ocus on wo s -case delay and packe ji e minimiza ion using
802.1Qbu-based p eemp ion and 802.1Qb . In [275], au ho s
e alua e he pe o mance o a E he ne TSN ne wo k enabled
by 802.1Qbu and 802.1Qb o he anspo o on haul
a ic. Au ho s benchma k he la ency pe o mance o hei
p oposal agains s ic p io i y and ound- obin scheduling.
Thei esul s o e se e al expe imen s demons a e ha he
ei he o TSN s anda ds is adequa e o p o ec ing he high-
p io i y on haul lows when he ne wo k is o e loaded
by low-p io i y a ic. An indi idual compa ison be ween
802.1Qbu and 802.1Qb also e eals ha he o me achie es
he lowes a e age la ency and ji e alues, whe eas he la e
achie es he lowes maximum ji e alue while s ill p o iding
below 100 µs la ency o comply wi h eCPRI.
2) TSN o Spo adic Backhaul T a ic: Au ho s o [276]
p opose he adop ion o he ATS s anda dized in 802.1Qc
o 5G backhaul ne wo ks. In asynch onous TSN, au ho s
encoun e an op imal ade-o be ween la ency gua an ees
and ne wo k scalabili y o he spo adic backhaul a ic wi h
eal- ime cons ain s, a oiding he ne wo k-wide synch oniza-
ion and coo dina ion needs o synch onous a ic shaping.
Au ho s add ess he low alloca ion p oblem in ATS-based
backhaul ne wo ks, which in ol es op imal pa h selec ion and
con igu a ion o each a ic class acco ding o delay, ji e ,
ame loss, and eliabili y cons ain s.
3) Enabling a Con e ged T anspo Ne wo k: Some esea ch
wo ks p opose he use o TSN as a echnology o enable
con e gence in he c osshaul, which e e s o he con in-
uum o on haul, midhaul, and backhaul ha comp ises
he TN. Re e ence [277] desc ibes an E he ne -based 5G
TN ha inco po a es on haul exp ess a ic and lowe
p io i y midhaul/backhaul a ic o e he same E he ne in as-
uc u e. Au ho s in eg a e a de e minis ic schedule called
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 299
FUSION [278], 802.1CM-based p eemp ion, and IEEE 1588
ime synch oniza ion o enable he agg ega ion and con e -
gence o mul iple on haul, midhaul and backhaul s eams
in he same ne wo k. Compa ed o s ic p io i y and ound-
obin schedule s, au ho s achie e on haul delay and PDV
ha is no impac ed in p esence o an inc easing numbe
o midhaul/backhaul s eams. In [279] and in subsequen
wo k [280], au ho s demons a e a 5G pla o m ha can
simul aneously accommoda e pe iodically gene a ed on haul
a ic and massi e IoT backhaul a ic. This is enabled by
he use o ime awa e shaping and adap i e mobile on haul
comp ession. Fo he o me , au ho s p opose he use o a
a ia ion o TAS in he agg ega ion poin s o on haul and
IoT a ic ha does no equi e ime synch oniza ion. Ins ead,
hei mechanism de ec s he ansmission and maximum leng h
o on haul bu s s by obse ing eal- ime ne wo k s a is ics,
hen au onomously se s he cycle ime and a ime slo o
he la ency-sensi i e on haul a ic. This ime slo can
addi ionally be sh unk wi hin each cycle by de ec ing when
each on haul bu s ends, o op imize he ansmission o IoT
a ic. In hei pla o m e alua ion in [280], au ho s achie e
a76µs on haul la ency o e a 10 km TN. Las ly, [149]
also deals wi h he inco po a ion o TSN in he whole 5G
anspo ne wo k, bu as i ocuses on he enablemen o
end- o-end TSN ne wo k slice o ches a ion, i is mainly
su eyed in Sec ion V-H. The a chi ec u e p oposed in his
wo k con empla es he use o TSN o he c osshaul including
wi hin he TN o a 5GS modeled as a TSN b idge.
4) In eg a ion o TSN in o he O-RAN A chi ec u e: Au ho s
o [50] e iew he equi emen s and op ions owa ds a
TSN-enabled O-RAN a chi ec u e. Au ho s highligh he
open na u e and c osshaul con e gence p o ided by O-RAN,
while ocusing he s udy on he on haul—OpenF on haul
in O-RAN, loca ed be ween he O-RU and he O-DU—due
o being he in e ace ha con ains he a ic wi h he mos
es ic i e equi emen s in e ms o la ency. Au ho s p opose
placemen op ions o he TSN con ol plane en i ies wi hin
he O-RAN componen s, and analyze possible mechanisms
o achie e he igh ime synch oniza ion equi ed be ween
O-RUs and O-DUs. Au ho s also analyze he TSN s anda ds
mos app op ia e o i s usage in O-RAN, desc ibing how
he 802.1CM p o ile could be le e aged, and highligh ing he
join usage o 802.1Qb and 802.1Qbu in o de o p o ide
de e minis ic delay whils op imizing he a ailable capaci y o
he di e en kinds o a ic in he TN.
Summa y and Lessons Lea ned:
•TSN mechanisms a e e ec i e o minimizing delay and
ji e in on haul ne wo ks and achie ing he la ency
budge s speci ied in eCPRI: The li e a u e showcases
ha he usage o 802.1Qbu-based ame p eemp ion and
o 802.1Qb o he agg ega ion o mul iple on haul
s eams o e a GCL, leads o be e pe o mance in
ega ds o delay and ji e compa ed o he me e use o
s ic p io i ies o o he use o o he mechanisms. This
quali y is pa icula ly ele an o i s di ec impac on
he size o he on haul link in e ms o nodes, and on
he leng h o he ibe links ha can be used, wi hou
comp omising he 100 µs la ency budge es ablished in
he eCPRI s anda d and he 802.1CM p o ile o he mos
s ingen on haul a ic. To achie e he ime synch o-
niza ion equi ed by 802.1Qb , ei he IEEE 1588 PTP o
802.1AS gPTP is p oposed by di e en wo ks. In ega ds
o 802.1Qb and 802.1Qbu, he li e a u e ag ees ha
he bes pe o mance is achie ed when bo h mechanisms
a e join ly used. Howe e , he use o only one o he
wo mechanisms may su ice o comply wi h he 100 µs
la ency budge depending on he ne wo k cha ac e is ics.
On he o he hand, he e ec i eness o o he mechanisms
such as 802.1Qch o 802.1Qci is unclea , as esea ch
wo ks a ailable so a only explo e hose wo s anda ds
along wi h simply using s ic p io i y.
•Adop ing TSN o he whole TN enables ne wo k
con e gence: While he mos s ingen ime-sensi i e
equi emen s ac oss he TN a e loca ed in he on haul
link, one o he main s eng hs o TSN lies in i s enable-
men o a con e ged ne wo k in as uc u e ha is capable
o simul aneously ea ing a ic wi h mixed cha ac e -
is ics in he mos app op ia e manne . Se e al esea ch
wo ks demons a e ha TSN can be used o con e ge and
agg ega e he on haul, midhaul, and backhaul a ic.
802.1Qb TAS-based mechanisms a e he mos popula
choices o his pu pose, managing he p io i y o each
kind o a ic and he leng h o he GCL schedule and
ime slo s acco dingly, including one p oposal ha adop s
a ic moni o ing and lea ning me hods o op imize TAS
con igu a ion. Fu he mo e, a con e ged c osshaul based
on TSN can be applied o he O-RAN a chi ec u e as
in oduced by one o he esea ch pape s su eyed in
his sec ion, hus enabling a de e minis ic TN wi h open
in e aces. The O-RAN a chi ec u e p o ides so wa e-
de ined con olle s o accommoda e he deploymen o
he TSN con ol plane en i ies as applica ions, while
s anda dizing well-de ined in e aces along he c osshaul
whe e TSN mechanisms can be applied. Howe e , his
is conside ed a no el esea ch line and mo e s udies a e
deemed app op ia e in his di ec ion.
F. So wa e-De ined Time-Sensi i e Ne wo ks
Resea ch ega ding he con e gence o SDN and TSN
a chi ec u es is ex ensi e in he li e a u e. In gene al, hese
ini ia i es le e age he cen alized con igu a ion model s an-
da dized in TSN and hei alignmen wi h he con ol plane
cen aliza ion pilla o SDN. As explained in Sec ion III-E,
in he cen alized con igu a ion models, he CNC en i y is
esponsible o he con igu a ion o he whole TSN ne wo k
based on s eam equi emen s ha come om he ne wo k
de ices, o , in he ully cen alized model speci ically, om
he CUC en i y. Due o i s lexibili y, SDN has been u ilized
in TSN ne wo ks o se e mul iple, o en complemen a y,
pu poses.
The axonomy p oposed o his subsec ion is shown
in Table XVIII. The p ima y scope is he so wa e-de ined
ne wo k con igu a ion, whe e a a ie y o design app oaches
and echniques can be iden i ied. Addi ionally, SDN-aligned
schemes ha ackle he ne wo k sel -con igu a ion and
306 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
•The esou ce alloca ion and VNF placemen challenge o
NFV is augmen ed in TSN ne wo ks: The VNF placemen
p oblem in NFV has been s udied in he li e a u e, aiming
o a i e a op imal s a egies o admi incoming VNFs
by weighing esou ce u iliza ion cos s o compu ing
and ne wo king esou ces whils adhe ing o speci ic
VNF cons ain s. When NFV mee s a TSN ne wo k
whe e VNFs ha e a ying equi emen s o imeliness
and de e minism, a new dimension is in oduced o his
p oblem, as he a ailabili y and op imiza ion o TSN
esou ces has o be aken in o conside a ion as well.
Fo example, an implemen a ion based on 802.1Qb
equi es ha all communica ion s eams be ween VNFs
a e assigned a a ic class and ime slo ha sa is ies
hei equi emen s. In his ega d, ano he ma e o
be add essed is ha , by de aul , NFV conside s lows
be ween VNFs as bidi ec ional and i ual, whe eas TSN,
by de ini ion, de ines unidi ec ional s eams ha a e
mapped o speci ic physical pa hs ac oss he ne wo k.
The e o e, he mapping is no s aigh o wa d.
•Use space ne wo king echniques a e p omising o mini-
mize he la ency o he ne wo king s ack: Se e al wo ks
con e ge o he use o ke nel-bypassing echniques o
accele a ed packe p ocessing in o de o deli e he
la ency and de e minism equi emen s o TSN appli-
ca ions. The mos s udied echnology o his pu pose
is DPDK, o en used oge he wi h OVS. OVS-DPDK
p o ides i ual ne wo king whe e applica ions in he
use space can ecei e ames om he NIC wi hou
con ex swi ching be ween spaces and wi hou CPU in e -
up s. As such, some wo ks ha add ess his opic epo
signi ican la ency imp o emen s when adop ing DPDK
o e ke nel space ne wo king o he communica ion pa h
o VMs and con aine s alike. OVS-DPDK has also been
p oposed o he i ualiza ion o a high-pe o mance
TAS b idge, an app oach ha can be conside ed syne -
gis ic wi h app oaches o encapsula e PSFP su eyed in
Sec ion V-F using ha dwa e accele a ion h ough special-
ized P4 swi ches.
•Vi ualiza ion p o ides a con e ged subs a e o lexibly
suppo 5G unc ions along wi h indus ial con ol unc-
ions in 5G-TSN en i onmen s: As seen in Sec ion IV,
he design o 5G ne wo ks in ol es he use o se ice-
based and mic ose ices-based a chi ec u es ha h i e
wi h he use o i ualiza ion. On he o he hand, se e al
wo ks su eyed in his sec ion ui ully push he use
o i ualiza ion in indus ial con ol applica ions wi h
s ic equi emen s in e ms o de e minis ic communi-
ca ions and eliabili y. This mani es s he iabili y o
deploymen s ha can no only con e ge he communica-
ion in as uc u e h ough TSN, bu also he compu ing
in as uc u e h ough i ualiza ion. A MEC deploymen
close o he p oduc ion lines could, o example, simul-
aneously alloca e a p i a e 5G con ol plane including
use case-speci ic NFs, and indus ial con ol unc ions
h ough PLCs ha suppo TSN-based C2C, C2D,
and D2Cmp communica ions, all in a cos -e icien and
lexible manne . While he li e a u e e lec s ha he
pe o mance esul s achie ed so a migh no be sui able
o he p ocesses wi h he igh es la ency-sensi i e con-
ol loops, many indus ial con ol in as uc u es shall
expec a signi ican lexibili y and scalabili y imp o e-
men wi h he adop ion o he i ualiza ion, cloud-na i e,
and mic ose ices IT pa adigms.
H. O ches a ion and Closed-Loop Au oma ion o TSN
Ne wo ks
This subsec ion su eys esea ch wo ks ha in oduce end-
o-end ne wo k managemen aspec s ha cha ac e ize 5G
ne wo ks and 5G-TSN ne wo ks, such as: end- o-end o ches-
a ion, ne wo k sel -con igu a ion and ze o- ouch closed-loop
au oma ion, ne wo k slicing, and ne wo k capabili y exposu e.
These ne wo k managemen schemes a e e lec ed in he
axonomy o Table XX, whe e con ibu ions can be p ima ily
classi ied be ween ne wo k op imiza ion,ne wo k o ches a-
ion, and ne wo k sel -con igu a ion and au oma ion.
1) End- o-end O ches a ion In ol ing TSN Resou ces:
In [334], au ho s desc ibe an edge cloud collabo a i e in el-
ligen pla o m o indus ial applica ions called Indus Edge
ha in eg a es a TSN-based da a plane subs a e. Indus Edge’s
a chi ec u e consis s o : an in as uc u e laye ha comp ises
ield de ices, a TSN da a plane be ween he de ices and he
edge, and an edge cloud ha p o ides K3s and Kube ne es
nodes; a pla o m laye , which includes an o ches a o ha
coo dina es TSN con igu a ion acco ding o he placemen
and cha ac e is ics o he applica ions, and addi ional ools
o boos e iciency and p og ammabili y such as edge cloud
and edge-edge collabo a ion modes, an SDK, and analy ics;
and an applica ion laye o he mic ose ices. In addi ion o
he p o o ype pla o m, au ho s p opose an emula ion pla o m
o Indus Edge. A pe o mance e alua ion is also ca ied ou
h ough wo indus ial use cases. Re e ence [335] p oposes
a ask de e minis ic ne wo k (TDN) a chi ec u e o enable
bounded la ency and ze o ji e communica ions in scena ios
o he Indus y 4.0 ecosys em ha in ol e 5G and MEC
echnologies. Wi h he au ho s’ ocal poin being he ask
o loading o indus ial applica ions o MEC sys ems, he
p ima y de e minis ic communica ion scena ios con empla ed
a e mul iple- o-one de ice- o-MEC access, and c oss-domain
in e connec ion o MEC sys ems. The p oposed hie a chi-
cal TDN a chi ec u e consis s o : an he e ogeneous access
ne wo k and da a plane, which includes a 5G RAN in he
ac o y ne wo k wi h a local MEC deploymen , and a anspo
ne wo k wi h addi ional MEC deploymen s o e TSN and
De Ne domains; a con ol plane o med by sub-con olle s
ha collabo a e seamlessly o he managemen o he da a
plane including c oss-domain con igu a ion o a ic shaping
mechanisms; and an applica ion plane ea u ing a as pa h
be ween he ne wo k and he applica ions. Au ho s o [336]
p esen AIDA, an AI-powe ed a chi ec u e o da a-d i en IIoT
applica ions. The p oposed o ches a ion a chi ec u e ocuses
on: (1) a TSN ne wo k and con ol plane wi h econ igu a ion
capabili ies aimed a add essing he dynamici y o IIoT and
mic ose ices-based indus ial applica ions in Kube ne es clus-
e s on he edge; (2) a moni o ing a chi ec u e encompassing

SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 307
TABLE XX
TAXONOMY RELATED TO THE ORCHESTRATION AND CLOSED-LOOP AUTOMATION OF TSN NETWORKS
he in as uc u e, pla o m, and con aine ized mic ose ices;
and (3) eal- ime AI/ML pipelines ha a ge he p o ision o
decision-making capabili ies o IIoT p ocesses based on he
p ocessing o collec ed senso da a. Au ho s in oduce p elim-
ina y pe o mance esul s o hei a chi ec u e based on wo
p oposed indus ial con ol use cases. Addi ionally, p e iously
su eyed wo ks [237] and [330] can also be highligh ed he e
o hei o ches a ion con ibu ion based on Kube ne es. O he
esea ch pape s b ie ly p opose he use o TSN o speci ic
domains such as he TN wi hin an end- o-end 5G o Beyond
5G (B5G) ne wo k o en ich he lexibili y o hei p oposed
a chi ec u e and suppo addi ional e ical use cases. Such
is he case o [337],[338] and [339]. Implemen a ion de ails
ega ding TSN a e howe e no elabo a ed.
2) Ne wo k Slicing in TSN Ne wo ks: [149] explo es he
inco po a ion o TSN echnologies in o end- o-end ne wo k
slicing, ocusing on he o ches a ion o a TSN-based TN
ab ic ha can se e 5G and non-5G lows concu en ly. The
au ho s’ p oposed a chi ec u e is conce ned wi h he in eg a-
ion o 3GPP ne wo k slicing managemen unc ions and an
SDN-based TSN con ol plane. A TSN slice o ches a o en i y
is in oduced as an en y poin a op he TSN con ol plane
(CNC/CUC) o p o ide i wi h slice isibili y so ha addi ional
slice isola ion mechanisms can be en o ced p io o TSN
scheduling con igu a ion such as 802.1Qb . This o ches a o
also in e aces wi h an addi ional con olle ha akes ca e
o slice managemen p ocedu es in he 5G ne wo k. Fo
slice de ini ion in he 5G managemen domain, an augmen ed
ne wo k slice empla e is p oposed o accommoda e TSN
ea u es. Au ho s inalize wi h a p elimina y implemen a ion
and e alua ion o hei app oach. In [340], au ho s ocus on
he use o ne wo k slicing in p i a e 5G ne wo ks in o de o
p o ide isola ion among p oduc ion lines in Indus y 4.0. In
his case, a he han seeking he design o a slice-awa e TSN
ne wo k, TSN is used as a ool o i s abili y o ully isola e he
pe o mance o he di e en slices in he TN whils p o iding
de e minis ic low la ency. Au ho s employ asynch onous TSN
based on he ATS o he 802.1Qc s anda d, which is e alua ed
agains ba e E he ne . Au ho s’ expe imen al e alua ion ocus-
ing on he midhaul be ween CUs and gNB-DUs highligh s he
e ec i eness o in as uc u e slicing oge he wi h TSN a ic
egula ion o ealize comple e slice isola ion and minimize
pe o mance deg ada ion in URLLC slices due o a ic excess
in o he slices.
3) Sel -con igu ing TSN Ne wo ks: Some wo ks in he
li e a u e explo e he challenges owa ds sel -con igu ing TSN
ne wo ks, ei he o add ess ne wo k and schedule econ igu-
a ion, o aiming o en i ely emo e he equisi e om end
s a ions o decla e hei s eam equi emen s. The syne gies
wi h SDN a chi ec u es a e clea in hese p oposals. The
amewo k in oduced in [96] is based on an agen ha con-
inuously moni o s ne wo k changes and adap s he ne wo k
con igu a ion acco dingly in eal- ime. Au ho s p opose he
in oduc ion o lea ning capabili ies in he TSN swi ches o
eed he agen in elligence abou ne wo k a ic. Ano he
example is [341], whe e au ho s ocus on he eal- ime econ-
igu a ion o GCL schedules using an a chi ec u e also based
on a con igu a ion agen . No e ha hese wo a icles we e
published a he ime ha 802.1 Qcc was being s anda dized.
Mo e ecen ly, [342] explo es equi ed unc ionali y o an
end- o-end sel -con igu ing TSN ne wo k. Cen alized knowl-
edge o ne wo k opology le e aging he ully cen alized
model o 802.1 Qcc, and specialized TSN agen s in he end
s a ions o he con igu a ion o pa ame e s such as TC Qdiscs
and ime synch oniza ion, a e p oposed. In his line, [343]
p esen s a SDN-based dynamic sel -con igu a ion app oach
o TSN-based IoT ne wo ks. Edge swi ches in he ne wo k
a e enhanced wi h he capabili y o moni o a ic and lea n
a ic cha ac e is ics. Au ho s’ p oposed app oach consis s on
ea ing incoming a ic as BE by de aul , and, once i s a ic
cha ac e is ics a e ex ac ed, he CNC compu es i s op imal
pa h and TAS con igu a ion acco ding o he s eam’s p io i y
308 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
and ime-sensi i e equi emen s. Finally, [344] speci ically
ocuses on he holis ic moni o ing o TSN ne wo ks, which
could be used o eed AI-based decision-making. Au ho s
iden i y h ee main sou ces o in o ma ion ha can be con-
as ed o p o ide insigh s anging om de ice conges ion o
mal unc ion o ne wo k miscon igu a ion: eal- ime ne wo k
a ic, con igu a ion s a us o ne wo k de ices, and cen alized
CNC/CUC in o ma ion. Au ho s design an a chi ec u e o
ex ac , p ep ocess, and s o e his in o ma ion.
4) Ze o- ouch Closed-loop Au oma ion o TSN and 5G-
TSN Ne wo ks: Some ecen wo ks adop ETSI ZSM-based
app oaches o he ze o- ouch con igu a ion o 5G ne wo ks
wi h TSN capabili ies. In [276], au ho s p opose he use o
Deep Rein o cemen Lea ning (DRL) based on ZSM and
on he F amewo k o Abs ac ion and Con ol o T a ic
Enginee ed Ne wo ks (ACTN) [345] o he op imiza ion o
he managemen and o ches a ion o 5G TNs. They employ
asynch onous TSN based on ATS, a ge ing he backhaul link
speci ically wi h he goal o low accep ance maximiza ion. In
he wo k low de ised by he au ho s, incoming low eques s
a e pa sed o de e mine hei cha ac e is ics, and hen a ic
p edic i e da a analy ics a e le e aged o d i e he alloca ion
o incoming low eques s acco ding o ne wo k s a us. In hei
wo k in [257] and [258], au ho s add ess he challenge o
he con igu a ion o a 5G NPN ea u ing 5G-TSN in eg a ion
ollowing he au oma ion p inciples o he ZSM amewo k.
They p opose an au oma a lea ning app oach o cons uc a
beha io al model o he 5G NPN as a digi al win in o de o
ind sui able 5GS con igu a ions o gi en TSN a ic lows
and desi ed KPIs. Au ho s in oduce a sma con olle ha
moni o s he 5G ne wo k and lea ns sequences o obse ed
s a es o gi en 5GS con igu a ions and TSN a ic pa e ns.
Using his model, he con olle can hen a i e o candida e
ne wo k con igu a ions ha can be applied o he eal 5G
NPN when he e a e KPI de ia ions be ween an obse ed and
desi ed s a e. The new ne wo k s a e hen eeds he moni o ing
module o he con olle in a closed loop manne . In [307],
au ho s apply he NDT pa adigm o 5G and 6G ne wo ks
ha in eg a e TSN. To cons uc he model in he NDT,
simula ion, emula ion, and AI/ML o e collec ed da a a e
p oposed in a complemen a y manne . The p oposed NDT si s
be ween he CNC and he CUC. The CNC sha es in o ma ion
abou he physical ne wo k o he NDT and can dynamically
deploy p obes in he physical ne wo k o collec speci ic da a
eques ed by he NDT. Some challenges o he NDT a e b ie ly
p esen ed, including he eal- ime communica ion be ween he
NDT and he physical ne wo k, and he da a collec ion om
he e ogeneous and dynamic da a sou ces.
5) Exposu e and Abs ac ion o TSN Con ol o
Ve icals: [346] showcases he design and implemen a ion o
a TSN Con olle (TSNC) Ne App o he exposu e o e ical
applica ions o he quali y-awa e managemen o a wi eless
TSN ne wo k. The TSNC con ains he TSN con ol plane
including CNC and CUC, bu also eal- ime moni o ing and
eleme y unc ions o ensu e compliance wi h eques ed KPIs.
The de ised da a plane consis s o WiFi ga eways coupled wi h
a TSN Agen (TSNA) ha in e ac s wi h he TSNC Ne App
and applies he ne wo k con igu a ion changes in he wi eless
TSN da a plane. The TSNC ea u es a no hbound in e ace
ha ecei es eques s om e ical applica ions, which a e
collec ed by he CUC in he o m o s eam cha ac e is ics and
expec ed KPIs. Moni o ing ules in he TSNC a e de i ed om
he egis y o e ical applica ions main ained in he CUC.
While no i s co e con ibu ion, a o emen ioned wo k [258]
also in oduces he managemen o e ical applica ion- acing
concep s including con igu a ion pa ame e s, a ic pa e s,
and expec ed KPIs o TSN sessions.
Summa y and Lessons Lea ned:
•TSN is a aluable echnology o he anspo laye
in edge clouds and MECs wi h de e minis ic equi e-
men s: Indus ial applica ions equi ing educed esponse
imes ha e long adop ed edge compu ing a chi ec u es.
Howe e , some au ho s ind ha he a ailabili y o edge
esou ces nea he de ices is no enough o gua an ee
he de e minism in he communica ion o he edge/MEC
node, o be ween collabo a ing nodes in an edge/MEC
sys em. TSN has been ound o be a p omising ech-
nology o ill his gap, including o he ealiza ion o
pe o mance-isola ed ne wo k slices o e such in as uc-
u es (e.g., o he isola ion o indus ial p oduc ion lines).
One o he main challenges, be e ackled wi h he usage
o SDN a chi ec u es, lies on he design o coo dina ed
mul i-domain o ches a ion capabili ies ha can con ol
end de ices, MEC o edge nodes, and TSN esou ces in
an uni ied manne . Fu he mo e, in 5G ne wo ks, RAN
esou ces also become pa o his puzzle. On he o he
hand, while he li e a u e ocuses on a chi ec u al aspec s
in hese p oposals, he unde lying TSN mechanisms o
employ a e no comp ehensi ely examined. In one case,
he use o CQF is b ie ly p oposed. O he wo ks hin a
he use o TAS-based scheduling. S ill, how o dis ibu e
he ime synch oniza ion be ween he di e en domains is
no examined. Con e sely, a p oposal o ne wo k slicing
in he TN using TSN inds ATS sui able gi en he non-
pe iodic a ic pa e ns. I is wo h no ing ha , in 5G,
MEC connec i i y can be conside ed a subcase o he 5G
TN, and, as la e seen in Sec ion V-E, bo h synch onous
and asynch onous a ic shaping mechanisms may be
app op ia e in hese scena ios.
•Add essing con igu a ion complexi y o TSN and 5G-
TSN ne wo ks h ough sel -lea ning and ze o- ouch
app oaches is challenging bu p omising: Concep s
including sel -con igu a ion, ze o- ouch au oma ion, eal-
ime moni o ing, and digi al win, ep esen syne gis ic
ends being adop ed by no el TSN ne wo k managemen
p oposals o add ess he con igu a ion complexi y o such
ne wo ks. Some wo ks adop he ZSM pa adigm o
his pu pose, wi h wo esea ch lines being iden i ied. In
one, his o ical da a combined wi h p edic i e analy ics
on upcoming wo kloads is used o a ic o ecas ing
o manage online TSN low alloca ion. In he o he ,
eal- ime moni o ing o he TSN a ic is used o
au oma ically econ igu e he ne wo k o mee eques ed
KPIs when he e a e de ia ions. Ini ial p oposals op
o RL-based solu ions, including DRL o au oma a
lea ning, hough u he esea ch is needed o de e mine
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 309
he ML me hod ha yields he bes pe o mance. In his
ega d, he u iliza ion o he digi al win pa adigm and
in pa icula o he NDT is also p omising. I a ai h ul
NDT can be cons uc ed, i p o ides a side-e ec ee
scena io o e alua e he beha io o TSN con igu a ion,
o example h ough RL me hods, be o e applying he
op imal con igu a ion candida e o he physical ne wo k.
No ably, NDTs and closed-loop au oma ion impose he
equi emen o eal- ime moni o ing, which has o be
ca e ully implemen ed o ensu e he alidi y and imeli-
ness o he collec ed da a wi hou in oducing o e head
in he ac ual a ic.
•Closed-loop au oma ion based on KPI moni o ing acil-
i a es he abs ac ion o unde lying TSN complexi y o
e ical applica ions: When i comes o abs ac ing end
s a ions om he unde lying mechanics o he TSN
ne wo k, he 802.1Qcc s anda d p o ides a ele an base-
line suppo h ough he de ini ion o he CUC en i y and
he UNI. End s a ions epo hei a ic equi emen s
in e ms o KPIs o a cen alized en i y, a he han
hemsel es execu ing he ne wo k-wide con igu a ion o
suppo hem. As illus a ed by some p elimina y esea ch
wo ks in he li e a u e, i s combina ion wi h eal- ime
moni o ing o said KPIs e ec i ely ma e ializes an in en -
based app oach o o ches a e TSN ne wo ks, acili a ing
i s usage by he e ogeneous e ical applica ions ha can
ocus on hei business logic a he han on speci ic TSN
mechanisms. A p omising complemen a y esea ch line
has also been ound o aim o ully au onomous TSN
ne wo ks, whe e end hos s do no ha e o epo a ic
equi emen s and hus all knowledge abou a ic cha ac-
e is ics and TSN p ocedu es can be o egoed. S ill, mo e
esea ch is needed o examine how his app oach scales
wi h high dynamici y in he alloca ion and dealloca ion
o TSN s eams, and wi h TSN ne wo ks in eg a ed wi h
5G, whe e he classi ica ion and con igu a ion o TSN
lows is mo e complex.
I. Indus ial Applica ion Deploymen s and P o o ypes Using
TSN o e 5G
This subsec ion desc ibes eal-wo ld deploymen s ca ied
ou by he esea ch communi y in which di e en kinds o
indus ial use cases a e enabled by he u iliza ion o TSN ech-
nology o e wi eless 5G connec i i y. This subsec ion includes
wo ks ha ca y ou pe o mance e alua ion o e a eal es bed
o p o o ype, as opposed o, o example, Sec ion V-B, which
co e ed wo ks in which he ocal poin is he execu ion
o ma hema ical and/o simula ion-based analyses. No e ha
some o he wo ks included in his subsec ion ha e also been
e e enced p e iously in his sec ion co e ing o he speci ic
con ibu ions.
Re e ence [248] desc ibes i e use cases o e alua ing
and alida ing 5G o sma manu ac u ing, conduc ed ac oss
h ee si es and o e six di e en 5G ne wo k deploymen s
wi h di e en cha ac e is ics ( equency bands, bandwid h,
SA/NSA a chi ec u e). Such use cases include collabo a i e
obo ics, p ocess moni o ing, and AGVs, and highligh bene i s
o adop ing 5G, such as he o loading o machine con ol
unc ionali y o an edge cloud, and o e all mo e lexible and
pe o man deploymen s. URLLC ea u es, FRER, and TSN-
o e -5G a e e alua ed.
Re e ence [347] p esen s and e alua es a p o o ype se up
in eg a ing 5G in a TSN ne wo k. In he p o o ype, au ho s
design a obo ic applica ion in which a Robo PLC ansmi s
eal- ime senso da a o a ac o y edge cloud ha con ols he
obo ’s beha io and mo emen . A p e-comme cial 5GS wi h
URLLC suppo is used o implemen he 5GS b idge ha
wi elessly in e connec s wo TSN b idges connec ed o he end
s a ions ha ep esen he obo and he edge cloud. Au ho s
e alua e 802.1Qb -based a ic shaping conside ing ST and
BE a ic classes. They desc ibe an app oach o syn hesize
end- o-end GCL con igu a ion in he de ices acco ding o
la ency h esholds, and le e age 802.1Qb in he ha dwa e
TSN de ices o educe he ji e in oduced sepa a ely by he
sending end s a ion and he 5G ne wo k. A me hod based on
Pulse-pe -second signals is used o measu e clock o se s.
Re e ence [249] demons a es an Au onomous Mobile
Robo (AMR) use case in a lineless mobile assembly sys em.
AMRs collabo a i ely pe o m a ious sc ew ins alla ion asks
on a uck chassis, and use 5G connec i i y o communica e
wi h an edge sys em ha execu es compu a ionally demanding
asks such as mo ion planning and mul i- obo coo dina ion.
Apa om e alua ing FRER as discussed in he p e ious
sec ion, au ho s apply he 3GPP-suppo ed hold and o wa d
mechanism o compensa e ji e in he wi eless ansmissions.
Re e ence [348] p esen s a 5G and TSN-enabled indus ial
con ol sys em de eloped wi h 3GPP Release 15 ha dwa e
and e alua es achie able la ency bounds o e such es bed
consis ing on a Ball-Balancing Table (BBT) ha comp ises
elec ical and mechanical dynamics, senso s, and ac ua o s.
Raspbe y Pi-based BBT componen s send senso da a h ough
a TSN ne wo k connec ed o an edge de ice, which sends back
con ol signals o he BBT o con ol he pa h o he ball.
Communica ion om and o he edge de ice is done ia 5G.
802.1Qb -based a ic scheduling suppo ed by 802.1AS ime
synch oniza ion is e alua ed in his indus ial sys em o he
p io i iza ion o ime-sensi i e a ic o e a i icial BE a ic.
VI. OPEN CHALLENGES AND RESEARCH LINES
F om he ex ensi e s a e o he a su ey ca ied ou in his
pape , he leading esea ch di ec ions can be iden i ied, and he
po en ially mos c i ical o challenging ma e s o be add essed
in he nea u u e can be ou lined. This sec ion discusses how
he in eg a ion be ween cu en 5G and TSN echnologies shall
con inue p og essing and e ol ing in o de o pa e he way
o inno a i e applica ions and pa adigms in he indus ial
au oma ion e ical.
A. S anda diza ion and Adop ion o URLLC Capabili ies
Wi h TSN s anda ds being de eloped o ime-sensi i e
communica ions o e wi ed E he ne ne wo ks, hei applica-
ion o wi eless 5G ne wo ks en ails signi ican challenges.
On he one hand, a wi eless communica ion channel ea-
u es dynamic p opaga ion cha ac e is ics and can expe imen
310 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
signal deg ada ion due o in e e ences and o he kinds o
channel impai men s. Thank ully, a signi ican e o o 5G
s anda diza ion and esea ch ocuses on he enablemen o
ul a- eliable and ul a-low la ency communica ions, which
clea ly syne gizes wi h suppo ing he equi emen s o he
mos s ingen TSN applica ions in he indus ial au oma-
ion e ical. In ac , we ha e seen how se e al URLLC
capabili ies a e highly adop ed in ecen esea ch s udies
as a ounda ion o hei p oposals, such as semi-s a ic
scheduling schemes like CG and SPS, lexible nume ology
in he SCS and he ansmission slo s, and schemes and
con igu a ions ha enable edundan ansmissions such as
edundan use plane pa hs based on NR-DC, and Mul i-TRP.
Hence, o ensu e an smoo h echnological de elopmen , SDOs
and indus y ini ia i es shall con inue o dedica e e o s o
expand cu en 5G NR and URLLC capabili ies in he new
eleases o 5G and u u e gene a ions h ough he echnolog-
ical su eillance o e ical indus y equi emen s. Likewise,
u u e esea ch s udies shall ollow s anda diza ion p og ess
closely and s ay ale o no el echnological ad ances in
his line which could un eil new oppo uni ies o op imize
ime-sensi i e communica ions o e 5G and he upcoming
gene a ions.
B. Scheduling and T ansmission o TSN T a ic o e 5G
The encompassing issue o augmen ing a wi ed TSN
ne wo k wi h wi eless 5G-TSN i ual b idges e ol es a ound
how o achie e an op imal scheduling o ime-sensi i e s eams
using 5G esou ces. Based on a gi en 5G NR nume ology,
TSN s eams ha e o be mapped o he a ailable 5G RBs in he
equency domain. F om he su eyed li e a u e, i is e iden
ha he e is no one-size- i s-all solu ion o his challenge, wi h
a iables anging om he pe iodici ies o he s eams, o hei
maximum delay budge , o he a ic he e ogenei y, and o he
end- o-end ne wo k opology. While some a i ma ions can be
made om he li e a u e such as he use ulness o semi-s a ic
scheduling o p e-alloca e esou ces o ime-sensi i e s eams,
and he bene i s o sub-slo scheduling and high SCS o enable
mo e g anula s eam pe iodici ies, di e en kinds o no el
app oaches a e being b ough o ligh .
Indeed, p oposals including adap i e scheduling algo-
i hms [243],[247], CG-based alloca ion schemes [244],[246],
a ic p io i iza ion schemes [245], and modula ion able
adjus men algo i hms [242], a di e en le els o he 5G
NR s ack, showcase ha he e is s ill ongoing inno a ion
in he ma e . Fu he mo e, he complexi y o his challenge
has p opelled ha many esea che s a e aking ad an age o
AI/ML echniques [258], o en based on RL [259],[260],
o ackle his challenge and o a i e a he op imal 5G
QoS con igu a ion by lea ning om he ne wo k beha io .
Wi h se e al con as ing p oposals eme ging om he esea ch
communi y, no only he e is oom o del e u he in o
hei op imiza ion and pe o mance cha ac e iza ion, bu , as
hey o en ea u e complemen a y app oaches o mechanisms
(e.g., le e aging di e en 5G NR and URLLC capabili ies),
assessing he easibili y and e ec i eness o ha moniously
combining hem can also be subs an ially explo ed.
On he o he hand, as a he syn hesis o a global 802.1Qb
GCL is conce ned, which is he mos adop ed app oach o
coo dina e he end- o-end scheduling in 5G-TSN ne wo ks, i
could be aluable o mo e ac i ely in es iga e he sui abili y
o ho oughly esea ched me hods o wi ed TSN ne wo ks.
Fo ins ance, he usage o SDN-based s eam classi ica ion and
clus e ing algo i hms (e.g., [290],[291]) in 5G QoS mapping
s a egies o TSN s eams could acili a e he syn hesis o
global GCLs ha a e also sui able o 5GS b idges. Howe e ,
his can be a double-edged swo d. While emb acing he
ex ensi e GCL compu a ion esea ch in wi ed TSN ne wo ks
can be highly bene icial, i is c i ical o accoun o he inhe en
di e ences in he wi eless 5GS b idges. Consequen ly, an
exp essi e in e play be ween he TSN con ol plane and he
5GS ha can accu a ely cap u e an uni ied ision o he wi ed
and wi eless domains, despi e hei di e ging cha ac e is ics in
beha io and esou ce and QoS managemen , can be a pi o al
ac o in he design o e ec i e end- o-end 5G-TSN pa h and
schedule compu a ion schemes.
C. Highly Accu a e TSN Time Synch oniza ion O e 5G
Join ly wi h he scheduling o ime-sensi i e s eams, he
main challenge o in eg a e TSN in 5G ne wo ks is conce ned
wi h achie ing a highly accu a e ime synch oniza ion o e
he wi eless medium, as indica ed by he numbe o esea ch
pape s ha add ess his opic. S ingen communica ion
equi emen s imposed by indus ial con ol applica ions and
implemen ed upon TSN capabili ies such as TAS, demand o
TSN ime synch oniza ion ac oss he 5GS wi h an accu acy in
he sub-mic osecond ange. The li e a u e iden i ies PDC as he
mos signi ican challenge owa ds his goal o p opaga ion
dis ances abo e a ce ain h eshold. Wi h he p opaga ion delay
es ima ion accu acy depending on a a ying ange o ac o s as
discussed in Sec ion V-D, mo e esea ch is deemed app op ia e
o mo e p ecisely cha ac e ize he impac o all hese ac o s
and he e ec i eness o a ailable and po en ial new me hods
in di e en ne wo k condi ions. Fu he mo e, gi en he na u e
o he p oblem, op imiza ion schemes based on AI/ML and
closed-loop au oma ion could be explo ed, aimed a inding
a sui able con igu a ion o maximize 5G esou ce e iciency
whils he synch oniza ion accu acy h eshold o he gi en
synch onici y budge equi emen s and ne wo k cha ac e is ics
can be a ained.
Ano he ele an aspec o conside is he e olu ion o ime
synch oniza ion schemes suppo ed in 3GPP s anda ds. While
Release 16 in oduced he suppo o downlink ime synch o-
niza ion, Release 17 expanded his suppo by conside ing new
con igu a ions o dis ibu e iming in o ma ion. A la ge, his
leads o ou di e en ime synch oniza ion schemes ha can
be app op ia e o di e en use case scena ios: (1) downlink
om GM behind ne wo k side owa ds UE side; (2) uplink
om GM behind UE side owa ds ne wo k side; (3) UE- o-
UE wi h GM behind one o he UEs; and (4) GM inside
5GS owa ds ne wo k side and UE side. Wi h a signi ican
po ion o he academic con ibu ions a ge ing he downlink
ime synch oniza ion scheme speci ically o being aligned
in imeline Release 16 s anda ds (e.g., [23],[266],[270]),
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 311
i is impo an ha upcoming esea ch e o s analyze he
implica ions o he emaining con igu a ions mo e deeply.
Las ly, he obus ness and esiliency agains ailu es o he
ime synch oniza ion is also a c i ical aspec o be mind ul
o in indus ial applica ions wi h s ingen synch oniza ion
equi emen s. In he scope o wi ed TSN ne wo ks, he
802.1ASdm s anda d, cu en ly in d a phase, is amending
802.1AS o in oduce ho s andby mechanisms o ob aining a
single clock sou ce om o wo gPTP domains o e edundan
pa hs, a p ima y one and an s andby one, wi hou equi ing a
BMC algo i hm execu ion. On he o he hand, p oposals o
adop ing such ime synch oniza ion edundancy mechanisms
in 5G-TSN ne wo ks appea o be missing in he li e a u e.
Howe e , i can be wo hwhile o s udy he easibili y o
suppo ing hem o e edundan 5G ne wo k con igu a ions.
D. Mobili y and Hando e Managemen o Mobile UEs in
5G-TSN Ne wo ks
Inno a i e Indus y 4,0 applica ions employ all kinds o
mobile ehicles and obo s in hei p oduc ion p ocesses, as
seen o example in he p o o ype deploymen s desc ibed in
Sec ion V-I. As he in eg a ion be ween TSN and 5G is a no el
esea ch opic, he communi y has unde s andably ocused on
e alua ing and op imizing he scena ios whe e he UE and DS-
TT ha e low mo emen speeds and hei 5G connec i i y ixed
o a single gNB. Howe e , while one o he main incen i es
o in oducing 5G ne wo k in o Indus y 4.0 is he educ ion
o cabling equipmen , many use cases such as mo ion con ol,
emo e con ol, and coope a i e wo kpiece ca ying, o en
in ol e mobile uni s ha span la ge a eas equi ing co e age
om mul iple cells, and ha can addi ionally each speeds
o up o 75 km/h [140]. On he one hand, how he PDU
session QoS and he ime synch oniza ion quali y is a ec ed
by UE speed can be u he examined, as i di ec ly impac s he
planning o he ne wo k and he choice o 5G QoS p o iles o
be con igu ed o suppo TSN s eams. On he o he hand, he
suppo o la ge co e age dis ances may equi e om hando e
e en s, whe he i s a cell change while emaining ancho ed o
he same CU, a comple e gNB change, o e en a hando e wi h
UPF elloca ion, wi h each ype po en ially ha ing di e en
easibili y limi s in he con ex o 5G-TSN ne wo ks. As such,
i is conside ed imely o in es iga e mechanisms ha can help
mi iga e he impac . Simila ly, he p ocedu es ha would need
o be igge ed in he 5G and TSN con ol planes o espond o
hese e en s can be s udied, in o de o ensu e he con inui y
o he TSN s eams and he synch oniza ion as seamlessly as
possible.
E. Open Sou ce So wa e o 5G-TSN In eg a ion
E en hough open sou ce 5G solu ions ha e exis ed o
yea s and a e cons an ly being upda ed and polished, as
o he w i ing o his pape , he e is no open sou ce 5G
so wa e ha suppo s he in eg a ion wi h TSN. Such suppo
would equi e he implemen a ion o he a chi ec u e and
p ocedu al aspec s o he 5GS b idge model, including he
TSN ansla o en i ies, NW-TT and DS-TT, he TSN AF, and
addi ional p ocedu es ac oss he 5GC and 5G-AN. Among
examples o exis ing 3GPP-complian open sou ce implemen-
a ions, ee5GC [349] and Open5GS [350] p o ide modula
co es, s sRAN [351] o e s a RAN implemen a ion, and he
OpenAi In e ace (OAI) [352] p ojec includes bo h co e and
RAN. Hence, suppo o TSN in eg a ion in any o hese
p ojec s would equi e ei he implemen a ion o e a solu ion
like OAI, o he coo dina ion be ween an open sou ce 5GC
p ojec and an open sou ce RAN p ojec . A i s app oach
could begin wi h he suppo o Release 16 ea u es, such as
downlink ime synch oniza ion and he essen ial in e ac ions
o QoS con igu a ion and managemen be ween CNC, TSN
AF, NW-TT, and DS-TT. I could e en ually be expanded o
suppo he ex ensions s anda dized in newe 3GPP eleases,
po en ially adop ing consolida ed ad ancemen s ha a ise om
he esea ch communi y as well. The a ailabili y o such open
sou ce solu ion could pa e he way o ake he s ep o wa d
om heo e ical e alua ion and simula ion analysis o ai h ul
es bed p o o ypes.
F. In eg a ion o TSN Suppo In o Vi ualiza ion and
O ches a ion A chi ec u es
As shown in his pape , indus y ini ia i es and academic
esea ch wo ks alike e lec a clea spike in he in e es o
adop i ualiza ion echnologies in indus ial ne wo ks. This
is especially ue o he agile con aine -based i ualiza ion,
ollowing he pa adigm shi owa ds mic ose ices. Mo eo e ,
se e al academic wo ks ha e su aced in ecen yea s add ess-
ing he usage o TSN o ime-sensi i e applica ions in end
s a ions wi h i ualiza ion suppo . In his ega d, he li e a u e
showcases p omising i s s eps owa ds he in eg a ion o
TSN in o Kube ne es [329],[330]; con aine -based i u-
aliza ion can be deemed mo e a ou able in his con ex
han hype iso -based i ualiza ion due o he complexi y
o add essing he o e head in oduced by he la e in o de
o comply wi h TSN equi emen s. Howe e , in addi ion o
Kube ne es, he e a e consolida ed a chi ec u al amewo ks
o deploy applica ions wi h ei he ype o i ualiza ion unde
ich and lexible o ches a ion capabili ies, namely NFV and
MEC, ha a e backed up by ex ensi e SDO wo k and indus y
adop ion. Ini ia i es led by SDOs o in eg a e TSN suppo in o
hese a chi ec u es, po en ially leading o he implemen a ion
o such suppo in o open sou ce NFV and MEC pla o ms,
could acili a e he g ow h o inno a i e indus ial applica ions.
An app oach o in eg a e o TSN in o ei he Kube ne es,
NFV, o MEC shall add ess he in e ope abili y and coope a-
ion be ween he a chi ec u es’ con ol planes. Ne e heless, i
shall also gua an ee ha o ches a ion mechanisms a e made
awa e o TSN asse s and eal- ime in as uc u e capabili ies in
o de o adequa ely con igu e he applica ions and he ne wo k
as a con inuum by applying p ocedu es de ined in TSN
s anda ds. Upcoming TSN s anda ds ega ding con igu a ion
enhancemen s, such as 802.1Qdj and 802.1Qcw, could play
an impo an ole. Fu he mo e, speci ic esea ch ques ions
may a ise. Fo example, in he highly dynamic en i onmen s
ha i ualiza ion o en ea u es, explo ing he easibili y o
al e na i e a ic shaping algo i hms mo e iendly o ne wo k
econ igu a ion, such as 802.1Qch CQF, could be in e es ing.

312 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
Ano he po en ial opic o in es iga e could be he applicabili y
o SFC o ep esen TSN s eams, gi en ha , se ice unc ion
chains, which can be implemen ed o example using he
Sou ce Rou ing pa adigm, can be made unidi ec ional [353].
Finally, a complemen a y esea ch di ec ion ha is only begin-
ning o su ace is he in eg a ion o TSN in o he i ualized
and in e ope able O-RAN a chi ec u e [50]. This is an equally
p omising challenge ha can be app oached in an equi alen
manne o he in eg a ion o TSN in o NFV o MEC, in his
case o b inging TSN o he TN o he 5G RAN.
G. Accele a ed SDN o Enhance Time-Sensi i e
Communica ions
Wi h he pe o mance o ime-sensi i e communica ions
being po en ially hampe ed by i ualiza ion, he use o ke nel
bypass and use space packe p ocessing app oaches ha e been
s udied in ecen yea s o accele a e he ne wo king s ack
o i ualized applica ions in eal- ime pla o ms. This end
is slowly mo ing o he TSN ecosys em. Some p omising
p oposals su eyed in his pape include he use o XDP o
he a ic s ee ing in 5G UEs [250], he design o a OVS-
DPDK-based Kube ne es CNI suppo ing TSN [329], and he
design o a TAS VNF ha is buil using DPDK [333]. Wi h
hese esea ch wo ks being mos ly p elimina y s udies and
app oaches, u he esea ch shall be ca ied ou . As mos
su eyed wo ks op o DPDK o e XDP, he la e could be
u he in es iga ed; e en hough exis ing s udies indica e ha
i s packe p ocessing pe o mance may be lowe han DPDK,
i boas s lexibili y ad an ages due o p o iding a ade-o
be ween ke nel bypass and ke nel ne wo king s ack [354].
O e all, an impo an ma e is pe o mance cha ac e iza ion
in di e en scena ios in o de o suppo he equi emen s
o TSN-awa e and i ualized indus ial applica ions. The
conside a ion o ha dwa e o load ia FPGA-based Sma NICs
o accele a e DPDK o XDP could be bene icial, as he
li e a u e showcases so wa e implemen a ions. In addi ion, he
in eg a ion o hese capabili ies in o exis ing pla o ms such as
Kube ne es can be highly impo an owa ds indus y adop ion.
Ano he echnology showing p omising po en ial in his
ega d is he P4 p og amming language. P4 p o ides a
powe ul and exp essi e packe p ocessing language wi h
a ich ecosys em o a ailable ha dwa e a ge s, ha aims
o decouple he ha dwa e om he so wa e implemen a-
ion [355]. Resea ch wo ks so a dealing wi h he use o P4 in
TSN ne wo ks ha e ocused on he implemen a ion o PSFP
unc ionali y in he da a plane [299],[300]. Howe e , i could
be p omising o employ P4 o he encapsula ion o o he
TSN unc ionali ies. This likely wa an s easibili y s udies in
ma e s anging om he clock synch oniza ion in P4 a ge s
and he p opaga ion o ime synch oniza ion messages in P4
ne wo ks, o he managemen o ne wo k econ igu a ions ha
may impac he P4 da a plane.
H. Au onomous Ze o-Touch 5G-TSN Ne wo ks
The li e a u e e lec s complemen a y p oposals ha can
play a c ucial ole in he au onomous con igu a ion and
op imiza ion o TSN and 5G-TSN ne wo ks. This includes
SDN-based ne wo k econ igu a ion schemes (e.g., [343]),
holis ic eal- ime moni o ing amewo ks (e.g., [344]), and
a e sa ile sui e o ML algo i hms o a i e a op imal
ne wo k con igu a ions (e.g., [253],[259]). E en hough u -
he esea ch on hese opics can be del ed in o independen ly,
he bigges po en ial likely comes om hei con ibu ion
as pa o a whole in he a chi ec ing o closed-loop, ze o-
ouch ne wo k managemen amewo ks— o example based
on ZSM—, o in he adop ion o he NDT pa adigm o TSN
ne wo ks. Addi ionally, he e is a signi ican ly challenging s ep
o ake in he enginee ing o hese solu ions so ha hey
con empla e he inclusion wi eless 5G domains in addi ion o
adi ional wi ed domains.
Some ecen esea ch s udies and p oposals, includ-
ing [257],[258],[276], and [307], ake he i s s eps o
add ess his challenge. Gi en he esonance o ze o- ouch
ne wo k managemen in oday’s 5G and he upcoming
6G [356], as well as in cu en and u u e Indus y 4.0
use cases in manu ac u ing and au oma ion [39], subs an ial
p o i s can be gained by con inuing esea ch in his di ec ion
pe aining TSN ne wo ks. While i s un eiling in he u u e
canno be de e mined wi h absolu e ce ain y, some hin s
can be aken. On he one hand, a mo e ai h ul e lec ion
o 5G-TSN in eg a ion mechanisms, such as he accoun ing
o ime synch oniza ion equi emen s, can enhance cu en ly
p oposed solu ions. Likewise, he conside a ion in he sys em’s
decision-making o p e iously discussed URLLC capabili-
ies, which ha e been deemed highly bene icial by o he
esea ch s udies, could be ano he na u al s ep o wa d in
his ma e . Addi ionally, while cu en ZSM-based p oposals
undamen ally app oach he p oblem om he pe spec i e o
s a ic ne wo k cha ac e is ics wi h changing applica ion a ic
and KPI eques s, a u u e esea ch line could look a how
he closed-loop managemen amewo k can lea n o p edic
a ying ne wo k channel condi ions in he 5G domain. This
would allow o an icipa e hei e ec s and ake app op ia e
ac ions (e.g., econ igu a ion o esou ces), in o de o main ain
compliance wi h a ic equi emen s in spi e o e en s like
channel deg ada ion, obs uc ion o he p opaga ion pa h, e c.
I is unclea i achie ing his would equi e ne wo k sensing
capabili ies ha will only a i e wi h 6G echnology [357],
bu a p elimina y app oach could look o le e age he ne wo k
insigh p o ided by cu en 3GPP-s anda dized capabili ies,
such as he NWDAF and MDAF NFs.
I. Exposu e o 5G-TSN Ne wo k Capabili ies Towa ds
Indus ial Applica ions
In o de o achie e a smoo h adop ion o de e minis ic 5G-
TSN hyb id deploymen s in indus ial en i onmen s, mainly
in he con ex o 5G NPNs, de eloping adequa e exposu e
capabili ies om he ne wo k owa ds he indus ial appli-
ca ions and managemen sys ems can play a c ucial ole.
Fo his pu pose, he ne wo k shall be able o p o ide ich
and lexible se ice-based in e aces ha can be used by an
indus ial consume en i y o p o ide applica ion equi emen s,
o eques he ac i a ion o se ices, and o moni o ne wo k
me ics; hese capabili ies shall be implemen ed in a way
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 313
ha he consume is abs ac ed om he in e nal complexi y
o he ne wo k and he unde lying mechanisms i u ilizes.
Wi h his in mind, i could be highly p o i able o esea ch
mo e exhaus i ely how a sa is ac o y exposu e o TSN a ic
managemen capabili ies in 5G ne wo ks can be ul illed based
on s anda dized 5G amewo ks and se ices such as NEF,
CAPIF, and SEAL. In his ega d, he pe spec i e o upcoming
s anda diza ion endea o s is p omising, as he 3GPP Technical
Repo (TR) 23.700-25 [358], which ou lines how Release 18
is expec ed o e ol e wi h ega ds o TSC and URLLC as
5G ma u es in o 5G-Ad anced, iden i ies se e al key issues
o be s udied, including: (1) how he 5GS can epo ime
synch oniza ion s a us and quali y o UEs, o de ices a ached
o UEs, and o hi d-pa y AFs; (2) how o enable an AF o
eques a ime synch oniza ion se ice in a speci ic co e age
a ea; (3) how o enable an AF o explici ly p o ide he equi ed
PER o he NEF o PCF o assis in he selec ion o a 5QI; and
(4) how o allow applica ions wi h low la ency equi emen s
o adap ansmission scheduling acco ding o RAN eedback,
and how o enable he RAN o p o ide such eedback o
he applica ions (AFs). Addi ionally, SDOs and he esea ch
communi y alike should play close a en ion o he epo s
o indus y alliances such as 5G-ACIA [137], gi en hei
p o ound ision on he demands o eme ging Indus y 4.0
applica ions.
J. End- o-End Mul i-Technology O ches a ion Solu ions
As seen h oughou his pape , complex communica ion sce-
na ios demanded by inno a i e indus ial applica ions imply
he necessi y o managing sepa a e echnology domains in a
coo dina ed manne , including:
•Applica ions unning inside o he e ogeneous sys ems
and pla o ms, loca ed ei he in end s a ions in a wi ed
TSN domain, in end s a ions behind a 5G UE, o in an
edge cloud.
•Wi ed TSN sys ems suppo ing p oduc ion lines o he
communica ion wi h cen alized con ol and/o an edge
cloud.
•5G Sys ems ac ing as i ual TSN b idges wi hin he
abo e TSN sys ems o p o ide wi eless connec i i y.
•The TN be ween he RAN and UPF o a 5G Sys em.
•The inclusion o a chi ec u es wi h 5G ne wo k slicing
capabili ies, whe e a single o mul iple slices may be
a ailable o alloca e di e en se ice lows.
Se e al a icles add essing di e en echnology domains
wi hin he scope o his pape sha e his iew on con igu a ion
and o ches a ion complexi y, e.g., [34],[38],[52],[237],
[257], and [340]. Ul ima ely, end- o-end o ches a ion in his
con ex encompasses no only he innume able p ocedu es
and mechanisms ela ed o 5G-TSN scheduling and ne wo k
managemen , bu also he li ecycle managemen o indus ial
CPSs and applica ions unde pa adigms like edge compu ing,
MEC, o NFV. C ucially, all o hese poin s should be ca ied
ou in an uni ied manne as a seamlessly in eg a ed whole,
o sa is y each indi idual equi emen ac oss he end- o-
end in as uc u e. Fu he mo e, he dynamici y in oduced
by 5G scena ios, such as h ough he apid deploymen
and econ igu a ion o VNFs, is unp eceden ed in adi ional
de e minis ic indus ial ne wo ks. This demands o obus
and complex o ches a ion capabili ies, including a p ope
econ igu a ion s a egy o TSN mechanisms, and a ne wo k
opology-awa e managemen o TSN esou ces. Fo all hese
easons, he s a e o he a shall con inue ad ancing owa ds
pe asi e mul i-domain o ches a ion solu ions such as he
p oposals in [149],[334],[335], and [336], building upon he
e e -e ol ing ecosys em o 5G amewo ks and capabili ies
de eloped by SDOs.
VII. CONCLUSION
In his pape , we ha e comp ehensi ely e iewed he main
echnologies and s anda ds pe aining he in eg a ion o TSN
and 5G echnologies in he con ex o indus ial au oma ion in
manu ac u ing. Addi ionally, we ha e ho oughly su eyed he
exis ing li e a u e in he ma e , showcasing he eme gence o
ex ensi e esea ch s udies and p oposals o imp o e he cu en
s a e o he a . Finally, we de i ed open challenges and se e al
p omising esea ch di ec ions ha a e complemen a y owa ds
he sha ed goal o suppo ing he e e -e ol ing equi emen s
o inno a i e Indus y 4.0 applica ions. Wi h encou aging
esea ch space in di e ging bu complemen a y scopes anging
om he TSN-complian scheduling o 5G esou ces, o he
suppo o TSN in pla o ms wi h i ualiza ion, and o he
end- o-end o ches a ion and ze o- ouch managemen o 5G-
TSN ne wo ks, we belie e ha he in eg a ion and con e gence
be ween TSN and 5G will con inue o be a highly ele an
opic in he coming yea s. Fu he mo e, we belie e ha a
join e o om SDOs, indus y alliances, and he esea ch
communi y, will lead o he bes ou comes. We hope ha ou
wo k in his pape can con ibu e o p o iding u he guidance
o he con inua ion o esea ch in his a ea in he oad owa ds
upcoming pa adigms and echnologies like Indus y 5.0
and 6G.
REFERENCES
[1] H. Lasi, P. Fe ke, H.-G. Kempe , T. Feld, and M. Ho mann, “Indus y
4.0,” Bus. In . Sys . Eng., ol. 6, no. 4, pp. 239–242, 2014.
[2] E. Hozdi´
c, “Sma ac o y o indus y 4.0: A e iew,” In . J. Mod.
Manu . Technol., ol. 7, no. 1, pp. 28–35, 2015.
[3] A. Kusiak, “Sma manu ac u ing,” In . J. P od. Res., ol. 56, nos. 1–2,
pp. 508–517, 2018.
[4] D. G. Pi o o, L. F. de Almeida, R. da Rosa Righi, J. J. Rod igues,
A. B. Lugli, and A. M. Albe i, “Cybe -physical sys ems a chi ec u es
o Indus ial In e ne o Things applica ions in indus y 4.0: A
li e a u e e iew,” J. Manu . Sys ., ol. 58, pp. 176–192, Jan. 2021.
[5] A. Angelopoulos e al., “Tackling aul s in he indus y 4.0 e a—A
su ey o machine-lea ning solu ions and key aspec s,” Senso s, ol. 20,
no. 1, p. 109, 2019.
[6] M. Wollschlaege , T. Sau e , and J. Jaspe nei e, “The u u e o indus ial
communica ion: Au oma ion ne wo ks in he e a o he In e ne o
Things and indus y 4.0,” IEEE Ind. Elec on. Mag., ol. 11, no. 1,
pp. 17–27, Ma . 2017.
[7] S. Vi u i, C. Zunino, and T. Sau e , “Indus ial communica ion sys ems
and hei u u e challenges: Nex -gene a ion E he ne , IIoT, and 5G,”
P oc. IEEE, ol. 107, no. 6, pp. 944–961, Jun. 2019.
[8] J. Fa kas, B. Va ga, G. Miklós, and J. Sachs, “5G-TSN in eg a ion
mee s ne wo king equi emen s o indus ial au oma ion,” E icsson
Technol. Re ., ol. 96, no. 7, pp. 45–51, 2019.
[9] B. Blanco e al., “Technology pilla s in he a chi ec u e o u u e 5G
mobile ne wo ks: NFV, MEC and SDN,” Compu . S and. In e aces,
ol. 54, pp. 216–228, No . 2017.
314 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 27, NO. 1, FEBRUARY 2025
[10] Z. M. Temes á i, D. Ma os, and P. Kádá , “Re iew o mobile com-
munica ion and he 5G in manu ac u ing,” P ocedia Manu ., ol. 32,
pp. 600–612, 2019, doi: 10.1016/j.p om g.2019.02.259.
[11] A. Nas allah e al., “Ul a-low la ency (ULL) ne wo ks: The IEEE
TSN and IETF De Ne s anda ds and ela ed 5G ULL esea ch,” IEEE
Commun. Su eys Tu s., ol. 21, no. 1, pp. 88–145, 1s Qua ., 2018.
[12] C. Benzaid and T. Taleb, “AI-d i en ze o ouch ne wo k and se ice
managemen in 5G and beyond: Challenges and esea ch di ec ions,”
IEEE Ne w., ol. 34, no. 2, pp. 186–194, Ma ./Ap . 2020.
[13] Y. Seol, D. Hyeon, J. Min, M. Kim, and J. Paek, “Timely su ey o
ime-sensi i e ne wo king: Pas and u u e di ec ions,” IEEE Access,
ol. 9, pp. 142506–142527, 2021.
[14] L. Deng, G. Xie, H. Liu, Y. Han, R. Li, and K. Li, “A su ey o
eal- ime E he ne modeling and design me hodologies: F om AVB o
TSN,” ACM Compu . Su ., ol. 55, no. 2, pp. 1–36, 2022.
[15] A. A. Ba akabi ze, A. Ahmad, R. Mijumbi, and A. Hines, “5G
ne wo k slicing using SDN and NFV: A su ey o axonomy, a chi-
ec u es and u u e challenges,” Compu . Ne w., ol. 167, Feb. 2020,
A . no. 106984.
[16] D. Ca alcan i, J. Pe ez-Rami ez, M. M. Rashid, J. Fang, M. Galee ,
and K. B. S an on, “Ex ending accu a e ime dis ibu ion and imeliness
capabili ies o e he ai o enable u u e wi eless indus ial au oma ion
sys ems,” P oc. IEEE, ol. 107, no. 6, pp. 1132–1152, Jun. 2019.
[17] L. Sil a, P. Ped ei as, P. Fonseca, and L. Almeida, “On he adequacy
o SDN and TSN o indus y 4.0,” in P oc. IEEE 22nd In . Symp.
Real-Time Dis ib. Compu . (ISORC), 2019, pp. 43–51.
[18] N. K. Hau and T. S. Chin, “Challenges and u u e di ec ion o
ime-sensi i e so wa e-de ined ne wo king (TSSDN) in au oma ion
indus y,” in P oc. 12 h In . Con . Secu ., P i acy, Anonym. Compu .,
Commun., S o ., 2019, pp. 309–324.
[19] A. Ghosh, A. Maede , M. Bake , and D. Chand amouli, “5G e olu ion:
A iew on 5G cellula echnology beyond 3GPP elease 15,” IEEE
Access, ol. 7, pp. 127639–127651, 2019.
[20] J. Zou, S. A. Sasu, J. Messenge , and J.-P. Elbe s, “Op ions o ime-
sensi i e ne wo king o 5G on haul,” in P oc. 45 h Eu . Con . Op .
Commun. (ECOC), 2019, pp. 1–3.
[21] T. S i le , N. Michailow, and M. Bah , “Time-sensi i e ne wo king
in 5 h gene a ion cellula ne wo ks-cu en s a e and open opics,” in
P oc. IEEE 2nd 5G Wo ld Fo um (5GWF), 2019, pp. 547–552.
[22] A. La añaga, M. C. Lucas-Es añ, I. Ma inez, I. Val, and J. Gozal ez,
“Analysis o 5G-TSN in eg a ion o suppo indus y 4.0,” in P oc. 25 h
IEEE In . Con . Eme g. Technol. Fac . Au om. (ETFA), ol. 1, 2020,
pp. 1111–1114.
[23] I. Godo e al., “A look inside 5G s anda ds o suppo ime synch o-
niza ion o sma manu ac u ing,” IEEE Commun. S and. Mag., ol.4,
no. 3, pp. 14–21, Sep. 2020.
[24] S. Jun, Y. Kang, J. Kim, and C. Kim, “Ul a-low-la ency se ices in
5G sys ems: A pe spec i e om 3GPP s anda ds,” ETRI J., ol. 42,
no. 5, pp. 721–733, 2020.
[25] A. Aijaz, “P i a e 5G: The u u e o indus ial wi eless,” IEEE Ind.
Elec on. Mag., ol. 14, no. 4, pp. 136–145, Dec. 2020.
[26] L. C. Mu alemwa and S. Shin, “Achie ing bounded ul a-low la ency
in 5G and beyond: Challenges and u u e esea ch di ec ions,” in P oc.
Ko ean Ins . Commun. Sci. Con ., 2021, pp. 38–41.
[27] S. Scanzio, L. Wisniewski, and P. Gaj, “He e ogeneous and dependable
ne wo ks in indus y–A su ey,” Compu . Ind., ol. 125, Feb. 2021,
A . no. 103388.
[28] F. Hamidi-Sepeh e al., “5G URLLC: E olu ion o high-pe o mance
wi eless ne wo king o indus ial au oma ion,” IEEE Commun. S and.
Mag., ol. 5, no. 2, pp. 132–140, Jun. 2021.
[29] Ó. Seijo, I. Val, M. Lu iso o, and Z. Pang, “Clock synch oniza ion
o wi eless ime-sensi i e ne wo king: A ma ch om mic osecond o
nanosecond,” IEEE Ind. Elec on. Mag., ol. 16, no. 2, pp. 35–43,
Jun. 2022.
[30] Y. Kang, S. Lee, S. Gwak, T. Kim, and D. An, “Time-sensi i e
ne wo king echnologies o indus ial au oma ion in wi eless commu-
nica ion sys ems,” Ene gies, ol. 14, no. 15, p. 4497, 2021.
[31] J. Sachs and K. Lande näs, “Re iew o 5G capabili ies o sma
manu ac u ing,” in P oc. 17 h In . Symp. Wi eless Commun. Sys .
(ISWCS), 2021, pp. 1–6.
[32] H. Chahed and A. J. Kassle , “So wa e-de ined ime sensi i e ne wo ks
con igu a ion and managemen ,” in P oc. IEEE Con . Ne w. Func .
Vi ualiz. So w. De in. Ne w. (NFV-SDN), 2021, pp. 124–128.
[33] J. P ados-Ga zon, P. Ameigei as, J. O donez-Lucena, P. Muñoz,
O. Adamuz-Hinojosa, and D. Camps-Mu , “5G non-public ne wo ks:
S anda diza ion, a chi ec u es and challenges,” IEEE Access, ol.9,
pp. 153893–153908, 2021.
[34] L. Ma en o m elde, A. Neumann, L. Wisniewski, and
L. Sch eckenbe g. “Co-con igu a ion o 5G and TSN enabling end- o-
end quali y o se ice in indus ial communica ions.” 2021. [Online].
A ailable: h ps://openda a. uni-halle. de//handle/1981185920/41505
[35] H. Xu, J. Xin, S. Xu, and H. Zhang, “RAN enhancemen o suppo
p opaga ion delay compensa ion o TSN,” in P oc. IEEE 9 h In . Con .
In ., Commun. Ne w. (ICICN), 2021, pp. 179–182.
[36] X. Li, Y. Chi, X. Jia, and Y. Xie, “Resea ch on in eg a ion o 5G and
TSN,” in Mecha onics and Au oma ion Technology. Ams e dam, The
Ne he lands: IOS P ess, 2022, pp. 128–135.
[37] T. Fedullo, A. Mo a o, F. T ama in, L. Ro a i, and S. Vi u i, “A
comp ehensi e e iew on ime sensi i e ne wo ks wi h a special ocus
on i s applicabili y o indus ial sma and dis ibu ed measu emen
sys ems,” Senso s, ol. 22, no. 4, p. 1638, 2022.
[38] A. Mahmood, S. F. Abedin, T. Sau e , M. Gidlund, and K. Lande näs,
“Fac o y 5G: A e iew o indus y-cen ic ea u es and deploymen
op ions,” IEEE Ind. Elec on. Mag., ol. 16, no. 2, pp. 24–34,
Jun. 2022.
[39] M. F iesen, L. Wisniewski, and J. Jaspe nei e, “Machine lea ning
o ze o- ouch managemen in he e ogeneous indus ial ne wo ks-a
e iew,” in P oc. IEEE 18 h In . Con . Fac . Commun. Sys . (WFCS),
2022, pp. 1–8.
[40] J. Xin, S. Xu, H. Zhang, and S. Xiong, “A e iew o ime sensi i e
communica ion echnology,” in P oc. In . Con . Compu . Eng. A i .
In ell. (ICCEAI), 2022, pp. 844–848.
[41] M.-A. Kou is e al., “5G NPN pe o mance e alua ion o I4.0
en i onmen s,” Appl. Sci., ol. 12, no. 15, p. 7891, 2022.
[42] F. Be e, “O e iew o 5G new adio and ime-sensi i e ne wo king
b idge a chi ec u e,” Sep. 2022, doi: 10.13140/RG.2.2.23244.49288.
[43] M. Abuibaid e al., “Cloudi ica ion o ime-sensi i e ne wo king eli-
abili y unc ions: Challenges and po en ial solu ion di ec ions,” IEEE
Commun. S and. Mag., ol. 6, no. 4, pp. 30–37, Dec. 2022.
[44] S. Senk, H. K. Naza i, H.-H. Liu, G. T. Nguyen, and F. H. Fi zek,
“Open-sou ce es beds o in eg a ing ime-sensi i e ne wo king wi h
5G and beyond,” in P oc. IEEE 20 h Consum. Commun. Ne w. Con .
(CCNC), 2023, pp. 1–7.
[45] G. P a eek Sha ma e al., “Towa d de e minis ic communica ions in 6G
ne wo ks: S a e o he a , open challenges and he way o wa d,” IEEE
Access, ol. 11, pp. 106898–106923, 2023.
[46] D. Chand amouli, P. And es-Maldonado, and T. Kolding, “E olu ion
o iming se ices om 5G-A owa ds 6G,” IEEE Access, ol. 11, pp.
35150–35157, 2023.
[47] Z. Sa ka, M. Ashjaei, H. Fo ouhi, M. Danesh alab, M. Sjödin, and
S. Mubeen, “A comp ehensi e sys ema ic e iew o in eg a ion o
ime sensi i e ne wo king and 5G communica ion,” J. Sys . A chi ec .,
ol. 138, May 2023, A . no. 102852.
[48] R. Muza a , M. Ahmed, E. Sisinni, T. Sau e , and H.-P. Be nha d,
“5G deploymen models and con igu a ion choices o indus-
ial cybe -physical sys ems—A s a e o a o e iew,” IEEE
T ans. Ind. Cybe -Phys. Sys ., ol. 1, pp. 236–256, Sep. 2023,
doi: 10.1109/TICPS.2023.3311394.
[49] M. Maliosz, I. Moldo án, M. Má é, C. Simon, and J. Ha ma os,
“De e minis ic local cloud o indus ial applica ions,” in P oc. IEEE
19 h In . Con . Fac . Commun. Sys . (WFCS), 2023, pp. 1–8.
[50] E. Municio, G. Ga cia-A iles, A. Ga cia-Saa ed a, and X. Cos a-Pé ez,
“O-RAN: Analysis o la ency-c i ical in e aces and o e iew o ime
sensi i e ne wo king solu ions,” IEEE Commun. S and. Mag., ol.7,
no. 3, pp. 82–89, Sep. 2023.
[51] T. Miz ahi and Y. Moses, “Timing in so wa e-de ined and cen ally-
managed ne wo ks,” 2019, a Xi :1904.06676.
[52] V. S uhá , M. Behnam, M. Ashjaei, and A. V. Papadopoulos, “Real-
ime con aine s: A su ey,” in P oc. 2nd Wo kshop Fog Compu . IoT
(Fog-IoT), 2020, pp. 1–9.
[53] M. Alabadi, A. Habbal, and X. Wei, “Indus ial In e ne o Things:
Requi emen s, a chi ec u e, challenges, and u u e esea ch di ec ions,”
IEEE Access, ol. 10, pp. 66374–66400, 2022.
[54] J. Kaise , D. Mc a lane, G. Hawk idge, P. And é, and P. Lei ão,
“A e iew o e e ence a chi ec u es o digi al manu ac u ing:
Classi ica ion, applicabili y and open issues,” Compu . Ind., ol. 149,
Aug. 2023, A . no. 103923.
[55] A. Nas allah e al., “Pe o mance compa ison o IEEE 802.1 TSN
ime awa e shape (TAS) and asynch onous a ic shape (ATS),” IEEE
Access, ol. 7, pp. 44165–44181, 2019.
[56] M. A. Ojewale, P. M. Yomsi, and B. Nikoli´
c, “Mul i-le el p eemp ion
in TSN: Feasibili y and equi emen s analysis,” in P oc. IEEE 23 d In .
Symp. Real-Time Dis ib. Compu . (ISORC), 2020, pp. 47–55.
[57] D. E genç and M. Fische , “On he eliabili y o IEEE 802.1 CB
FRER,” in P oc. IEEE Con . Compu . Commun., 2021, pp. 1–10.
SASIAIN e al.: TOWARD THE INTEGRATION AND CONVERGENCE 315
[58] Y. Zhang, Q. Xu, L. Xu, C. Chen, and X. Guan, “E icien low
scheduling o indus ial ime-sensi i e ne wo king: A di isibili y
heo y-based me hod,” IEEE T ans. Ind. In o ma ., ol. 18, no. 12,
pp. 9312–9323, Dec. 2022.
[59] T. S übe , L. Osswald, S. Lindne , and M. Men h, “A su ey o
scheduling algo i hms o he ime-awa e shape in ime-sensi i e
ne wo king (TSN),” IEEE Access, ol. 11, pp. 61192–61233, 2023.
[60] R. A. Peña, M. Pascual, A. As a loa, D. U ibe, and J. Inchaus i, “Impac
o MACsec secu i y on TSN a ic,” in P oc. 37 h Con . Design
Ci cui s In eg . Ci cui s (DCIS), 2022, pp. 1–6.
[61] R. Se hi, A. Kadam, K. P abhu, and N. Ko a, “Secu i y conside a ions
o enable ime-sensi i e ne wo king o e 5G,” IEEE Open J. Veh.
Technol., ol. 3, pp. 399–407, Sep. 2022.
[62] T. Adame, M. Ca ascosa-Zamacois, and B. Bellal a, “Time-
sensi i e ne wo king in IEEE 802.11 BE: On he way o
low-la ency WiFi 7,” Senso s, ol. 21, no. 15, p. 4954, 2021,
doi: 10.1109/TICPS.2023.3311394.
[63] I. Val, Ó. Seijo, R. To ego, and A. As a loa, “IEEE 802.1 AS
clock synch oniza ion pe o mance e alua ion o an in eg a ed wi ed–
wi eless TSN a chi ec u e,” IEEE T ans. Ind. In o ma ., ol. 18, no. 5,
pp. 2986–2999, May 2022.
[64] D. Ca alcan i, C. Co dei o, M. Smi h, and A. Rege , “WiFi TSN:
Enabling de e minis ic wi eless connec i i y o e 802.11,” IEEE
Commun. S and. Mag., ol. 6, no. 4, pp. 22–29, Dec. 2022.
[65] M. Ashjaei, L. L. Bello, M. Danesh alab, G. Pa i, S. Sapona a,
and S. Mubeen, “Time-sensi i e ne wo king in au omo i e embedded
sys ems: S a e o he a and esea ch oppo uni ies,” J. Sys . A chi .,
ol. 117, Aug. 2021, A . no. 102137.
[66] Y. Peng, B. Shi, T. Jiang, X. Tu, D. Xu, and K. Hua, “A su ey on
in- ehicle ime-sensi i e ne wo king,” IEEE In e ne Things J., ol. 10,
no. 16, pp. 14375–14396, Aug. 2023.
[67] P. T akadas e al., “A cos -e icien 5G non-public ne wo k a chi ec u al
app oach: Key concep s and enable s, building blocks and po en ial use
cases,” Senso s, ol. 21, no. 16, p. 5578, 2021.
[68] X. Yang, D. Scholz, and M. Helm, “De e minis ic ne wo king (De Ne )
s ime sensi i e ne wo king (TSN),” in P oc. Semin. IITM, 2019,
pp. 79–84.
[69] N. Amb osy, T. Kampa, U. Juma , and D. G oßmann, “5G and De Ne :
Towa ds holis ic de e minism in indus ial ne wo ks,” in P oc. IEEE
In . Con . Ind. Technol. (ICIT), 2022, pp. 1–6.
[70] Time-Sensi i e Ne wo king (TSN) Task G oup, IEEE S anda d 802.1,
2019. [Online]. A ailable: h ps://1.ieee802.o g/ sn/
[71] D. B uckne e al., “An in oduc ion o OPC UA TSN o indus ial
communica ion sys ems,” P oc. IEEE, ol. 107, no. 6, pp. 1121–1131,
Jun. 2019.
[72] J. L. Messenge , “Time-sensi i e ne wo king: An in oduc ion,” IEEE
Commun. S and. Mag., ol. 2, no. 2, pp. 29–33, Jun. 2018.
[73] IEEE S anda d o Local and Me opoli an A ea Ne wo ks–B idges
and B idged Ne wo ks, IEEE S anda d 802.1Q-2022 (Re ision o IEEE
S anda d 802.1Q-2018), 2022.
[74] IEEE S anda d o Local and Me opoli an A ea Ne wo k–B idges and
B idged Ne wo ks, IEEE S anda d 802.1Q-2018 (Re ision o IEEE
S anda d 802.1Q-2014), 2018.
[75] IEEE S anda d o Local and Me opoli an A ea Ne wo ks–B idges
and B idged Ne wo ks, IEEE S anda d 802.1Q-2014 (Re ision o IEEE
S anda d 802.1Q-2011), 2014.
[76] G. Alde isi, G. Pa i, and L. L. Bello, “In oducing suppo o
scheduled a ic o e IEEE audio ideo b idging ne wo ks,” in P oc.
IEEE 18 h Con . Eme g. Technol. Fac . Au om. (ETFA), 2013, pp. 1–9.
[77] IEEE S anda d o a P ecision Clock Synch oniza ion P o ocol o
Ne wo ked Measu emen and Con ol Sys ems, IEEE S anda d 1588-
2019 (Re ision o IEEE S anda d 1588-2008), 2020.
[78] IEEE S anda d o Local and Me opoli an A ea Ne wo ks–Timing and
Synch oniza ion o Time-Sensi i e Applica ions, IEEE S anda d 802.1
AS-2020 (Re ision o IEEE S anda d 802.1 AS-2011), 2020.
[79] IEEE S anda d o Local and Me opoli an A ea Ne wo ks–F ame
Replica ion and Elimina ion o Reliabili y, IEEE S anda d 802.1CB-
2017, 2017.
[80] IEEE S anda d o Local and Me opoli an A ea Ne wo ks–B idges
and B idged Ne wo ks–Amendmen 28: Pe -S eam Fil e ing and
Policing, IEEE S anda d 802.1Qci-2017 (Amendmen o IEEE S anda d
802.1Q-2014 as amended by IEEE S anda d 802.1Qca-2015, IEEE
S anda d 802.1Qcd-2015, IEEE S anda d 802.1Q-2014/Co 1-2015,
IEEE S anda d 802.1Qb -2015, IEEE S anda d 802.1Qbu-2016, IEEE
S anda d 802.1Qbz-2016), 2017.
[81] D. E genç, C. B ülha , J. Neumann, L. K üge , and M. Fische , “On
he secu i y o IEEE 802.1 ime-sensi i e ne wo king,” in P oc. IEEE
In . Con . Commun. Wo kshops (ICC Wo kshops), 2021, pp. 1–6.
[82] P. Meye , T. Häckel, S. Reide , F. Ko , and T. C. Schmid , “Ne wo k
anomaly de ec ion in ca s: A case o ime-sensi i e s eam il e ing
and policing,” 2021, a Xi :2112.11109.
[83] IEEE S anda d o Local and Me opoli an A ea Ne wo ks—Vi ual
B idged Local A ea Ne wo ks Amendmen 12: Fo wa ding and Queuing
Enhancemen s o Time-Sensi i e S eams,” IEEE S anda d 802.1Qa -
2009 (Amendmen o IEEE S anda d 802.1Q-2005), 2010.
[84] Time Sensi i e Ne wo ks o Flexible Manu ac u ing Tes bed
Cha ac e iza ion and Mapping o Con e ged T a ic Types, Ind. In e ne
Conso . (IIC), Bos on, MA, USA, 2019.
[85] R. Hummen, S. Keh e , and O. Kleinebe g, “TSN– ime sensi i e
ne wo king,” Hi schmann Au om., Roches e Hills, MI, USA, Whi e
Pape 00027, 2016.
[86] IEEE S anda d o Local and Me opoli an A ea Ne wo ks—B idges
and B idged Ne wo ks—Amendmen 25: Enhancemen s o Scheduled
T a ic, IEEE s anda d 802.1Qb -2015 (Amendmen o IEEE s anda d
802.1Q-2014 as amended by IEEE s anda d 802.1Qca-2015, IEEE
s anda d 802.1Qcd-2015, IEEE s anda d 802.1Q-2014/Co 1-2015),
2016.
[87] IEEE S anda d o Local and Me opoli an A ea Ne wo ks–B idges
and B idged Ne wo ks—Amendmen 34:Asynch onous T a ic Shaping,
IEEE S anda d 802.1Qc -2020 (Amendmen o IEEE S anda d 802.1Q-
2018 as amended by IEEE S anda d 802.1Qcp-2018, IEEE S anda d
802.1Qcc-2018, IEEE S anda d 802.1Qcy-2019, IEEE S anda d
802.1Qcx-2020), 2020.
[88] J. Spech and S. Samii, “U gency-based schedule o ime-sensi i e
swi ched e he ne ne wo ks,” in P oc. 28 h Eu omic o Con . Real-Time
Sys . (ECRTS), 2016, pp. 75–85.
[89] IEEE S anda d o Local and Me opoli an A ea Ne wo ks—B idges
and B idged Ne wo ks—Amendmen 26: F ame P eemp ion, IEEE
S anda d 802.1Qbu-2016 (Amendmen o IEEE S anda d 802.1Q-
2014), 2016.
[90] IEEE S anda d o E he ne Amendmen 5: Speci ica ion and
Managemen Pa ame e s o In e spe sing Exp ess T a ic, IEEE
S anda d 802.3b -2016 (Amendmen o IEEE S anda d 802.3-2015 as
amended by IEEE S anda d 802.3bw-2015, IEEE S anda d 802.3by-
2016, IEEE S anda d 802.3bq-2016, IEEE S anda d 802.3bp-2016),
2016.
[91] A. A es o a, K.-S. J. Hielsche , and R. Ge man, “Simula i e e alua ion
o he TSN mechanisms ime-awa e shape and ame p eemp ion and
hei sui abili y o indus ial use cases,” in P oc. IFIP Ne w. Con .
(IFIP Ne w.), 2021, pp. 1–6.
[92] IEEE S anda d o Local and Me opoli an A ea Ne wo ks–B idges and
B idged Ne wo ks–Amendmen 29: Cyclic Queuing and Fo wa ding,
IEEE 802.1Qch-2017 (Amendmen o IEEE S anda d 802.1Q-2014
as amended by IEEE S anda d 802.1Qca-2015, IEEE S anda d
802.1Qcd(TM)-2015, IEEE S anda d 802.1Q-2014/Co 1-2015, IEEE
S anda d 802.1Qb -2015, IEEE S anda d 802.1Qbu-2016, IEEE
S anda d 802.1Qbz-2016, IEEE S anda d 802.1Qci-2017), 2017.
[93] W. Quan, J. Yan, X. Jiang, and Z. Sun, “On-line a ic scheduling
op imiza ion in IEEE 802.1Qch based ime-sensi i e ne wo ks,” in
P oc. IEEE 22nd In . Con . High Pe o m. Compu . Commun., IEEE
18 h In . Con . Sma Ci y, IEEE 6 h In . Con . Da a Sci. Sys .
(HPCC/Sma Ci y/DSS), 2020, pp. 369–376.
[94] B. Wang, F. Luo, and Z. Fang, “Pe o mance analysis o IEEE 802.1
Qch o au omo i e ne wo ks: Compa ed wi h IEEE 802.1 Qb ,” in
P oc. IEEE 4 h In . Con . Compu . Commun. Eng. Technol. (CCET),
2021, pp. 355–359.
[95] IEEE S anda d o Local and Me opoli an A ea Ne wo ks–Vi ual
B idged Local A ea Ne wo ks Amendmen 14: S eam Rese a ion
P o ocol (SRP), IEEE S anda d 802.1Qa -2010 (Re ision o IEEE
S anda d 802.1Q-2005), 2010.
[96] M. Gu ié ez, A. Ademaj, W. S eine , R. Dob in, and S. Punnekka ,
“Sel -con igu a ion o IEEE 802.1 TSN ne wo ks,” in P oc. 22nd IEEE
In . Con . Eme g. Technol. Fac . Au om. (ETFA), 2017, pp. 1–8.
[97] IEEE S anda d o Local and Me opoli an A ea Ne wo ks–B idges
and B idged Ne wo ks—Amendmen 31: S eam Rese a ion P o ocol
(SRP) Enhancemen s and Pe o mance Imp o emen s, IEEE S anda d
802.1Qcc-2018 (Amendmen o IEEE S anda d 802.1Q-2018 as
amended by IEEE S anda d 802.1Qcp-2018), 2018.
[98] M. Bjo klund, “YANG-a da a modeling language o he ne wo k
con igu a ion p o ocol (NETCONF),” In e ne Eng. Task Fo ce, RFC
6020, 2010.
[99] R. Enns, M. Bjo klund, J. Schoenwaelde , and A. Bie man, “Ne wo k
con igu a ion p o ocol (NETCONF),” In e ne Eng. Task Fo ce, RFC
6241, Jun. 2011.
[100] A. Bie man, M. Bjo klund, and K. Wa sen, “RESTCONF p o ocol,”
In e ne Eng. Task Fo ce, RFC 8040, Jan. 2020.