A framework for next generation cloud-native SDN cognitive resource orchestrator for IoTs (NG2CRO)

A F amewo k o Nex Gene a ion Cloud-Na i e SDN Cogni i e Resou ce O ches a o o IoTs (NG2CRO) ⋆ Ha iza Kanwal Janjua1[0000−0002−6655−6015], Ignacio de Miguel1[0000−0002−1084−1159], Ram´on J. Du ´an Ba oso1[0000−0003−1423−1646], Ma yam Masoumi1[0000−0002−3832−9106], Soheil Hosseini1[0000−0003−3006−5297], Juan Ca los Aguado1[0000−0002−2495−0313], Noem´ı Me ayo1[0000−0002−6920−0778], and E a is o J. Ab il1[0000−0003−4164−2467] Uni e sidad de Valladolid (UVa), Valladolid, Spain {ha izakanwal.janjua, ignacio.demiguel, amon.du an, ma yam.masoumi, soheil.hosseini, juanca los.aguado.manzano, noeme , ejab il}@u a.es Abs ac . SDN (So wa e De ine Ne wo king) and NFV (Ne wo k Func- ion Vi ualiza ion) a e he key enable s o 5G sys ems and also open many doo s in he cloud-na i e applica ion. Besides, i in i es new chal- lenges o he e iciency and scalabili y o esou ce managemen . This wo k aims o p o ide a cogni i e amewo k o 5G esou ce and se ice o - ches a ion in a cloud-na i e SDN en i onmen . The p oposed NG2CRO amewo k esou ce o ches a o is designed o adap he ne wo k’s sel - lea ning capabili ies and dynamici y while aken on o accoun he ne - wo k’s Ma ko ian p ope ies and di e se se ice equi emen s. We con- side inco po a ing AI (A i icial In elligence) echniques speci ically RL (Rein o cemen Lea ning) me hodologies because li e a u e has shown ha hese echniques can e icien ly add ess and comply wi h he cu en dynamic beha io s and he e ogenei y o 5G se ices and applica ions. In conclusion, bo h bene i s and liabili ies a e discussed o inco po a ing AI speci ically RL in o esou ce o ches a ion p ac ices ha p o ide us wi h u u e esea ch challenges. Keywo ds: Nex Gene a ion 5G Ne wo k ·A i icial In elligence ·Ne - wo k Au oma ion ·So wa e-De ine Ne wo king ·Cloud-Na i e SDN · Resou ce O ches a ion. 1 In oduc ion SDN o ches a ion acili a es ene gy and esou ce op imiza ion o he In e - ne o Things (IoT) indus y. SDN allows e icien , cos -e ec i e, and dynamic ⋆This wo k is pa o he IoTalen um p ojec , which has ecei ed unding om he EU H2020 esea ch and inno a ion p og amme unde he MSCA g an ag eemen No 953442. I is also suppo ed by Conseje ´ıa de Educaci´on de la Jun a de Cas illa y Le´on and he Eu opean Regional De elopmen Fund (G an VA231P20), and he Spanish Minis y o Science o Inno a ion and he S a e Resea ch Agency (G an PID2020-112675RB-C42 unded by MCIN/AEI/10.13039/501100011033). 2 H. K. Janjua e al. scalabili y o esou ces in cloud and edge compu ing se ices [1, 12]. The e o- lu ion o he IoT indus y has aised he demand o cloud da a cen e s whe e ens o housands o compu e nodes a e connec ed h ough housands o ne - wo ks, c ea ing a enues o manageabili y and pe o mance issues. Cloud da a cen e s need dynamic p o ision compu ing and ne wo k esou ces o adap o he luc ua ing demands o cus ome s [17]. Wi h ad en s in so wa e-de ined clouds (SDC) ha in eg a e SDN and cloud esou ce managemen whe e he p og ammable con olle p o ides dynamic and au onomic con igu a ion and managemen o he unde lying esou ces; cogni ion o such scena io can p edic he equi emen and inc ease he e iciency in such scena ios [2, 10]. Despi e he inc easing popula i y o s udies o join esou ce op imiza ion in he cloud en i onmen wi h SDN echnology, he ealiza ion is s ill limi ed o de eloping in eg a ed managemen pla o ms p o iding simul- aneous con ollabili y and ene gy op imiza ion o compu ing and ne wo king in as uc u es. The es ing cos and ene gy e iciency in eal- ime a e equi ed o mig a ion o new echnologies [4,6]. Mo eo e , 5G ne wo ks a e complex, la ge, and di e se wi h he e ogeneous se ices which equi e high lexibili y and e iciency o mee he SLA and QoS ag eed upon by he se ices p o ide . Howe e , due o he lack o lexibili y in adi ional 4G ne wo ks, when comes o he dynamics o 5G sys ems due o he he e ogeneous se ices. The e o e, he cu en esou ce op imiza ion echniques a e no sui able o he cu en 5G in as uc u e, which, is aced wi h chal- lenges o SLA and QoS iola ion due o he incapabili y o handle he sys em dynamici y. Hence, he e is a need o ha e ad anced cogni i e con ol me hods ha can e icien ly maneu e and o ches a e he dynamici y o he ne wo k demands. Consequen ly, s udies ex apola ed ha RL-based cogni i e app oaches a e ben- e icial o lea n and explo ing he dynamics o he ne wo k au onomously which is capable o suppo se ice lexibili y and can p o ide an enhanced use expe- ience in e ms o SLA and QoS ul illmen [8]. 1.1 Aims & Objec i es The aim o his p ojec is an associa ion wi h he IoTalen um p ojec [7] speci ically o he opic ESR7: ”Cogni i e o ches a o o MEC and ne wo k esou ces”. Hence, he ision is o design a Cloud-Na i e SDN join esou ce o ches a ion sys em o he ne wo k and se ice esou ces. Also, i should op- imize he ene gy and cos p o ile ia pi o ing he abili y o sel -lea ning and sel -managemen o gua an ee pe o mance in he 5G sys ems. Mo eo e , he in eg a ion o Cloud and SDN pla o ms should le e age he amewo k unc ioning a he con olle ia in elligen ne wo k managemen , op- imal esou ce u iliza ion, load balancing, and se ice o ches a ion o p o ide IaaS (In as uc u e-as-a-Se ice) and SaaS (So wa e-as-a-Se ice) o use s ha will ope a e he MEC and he unde lying IoT ne wo k o IoTalen um a chi ec- u e. NG2CRO F amewo k 3 We aim o le e aged he objec i es o ESR7 by in eg a ing he e icien , op- imized, and p oac i e pe o mance-awa e Deep Rein o cemen Lea ning ech- niques o de elop he ad anced decision-making pa e n o ene gy moni o ing, p o ec ion, and con ol mechanisms o deli e op imal SLA and QoS in IaaS and SaaS en i onmen s. We designed some esea ch ques ions wi h he conside a ion o 5G ne wo k e- sou ces o ches a ion using a i icial in elligence speci ically ein o cemen lea n- ing and Machine lea ning along wi h he majo p oblems in cu en ne wo k esou ce managemen . Table 1 p o ides an o e iew o he esea ch ques ions in ela ion o he p oblems. Table 1. P oblem and Resea ch Ques ions P oblem (P) Resea ch Ques ions (RQ) P1: Dynamici y in a ic demands RQ1: How AI echniques has and use s’ mobili y. pe o med in li e a u e o ches a ion in 5G Ne wo k? P2: Dynamic sys ems ende RQ2: How RL-based me hods can be ma ko ian p ope ies due o use ul o esou ce o ches a ion o he e ogeneous ne wo k de ices. o add ess he dynamici y? P3: Need o sa e o ches a ion due o RQ3: Wha will be use o cons ain -awa e cons ain s o he physical sys em RL me hods ha ing an impac on SLA and e en in i ualized SDN NFV QoS o he 5G slicing ne wo k scena io? in as uc u e Sa e O ches a ion is Requi ed!!! which is e med as making he esou ce o ches a ion policies which should conside he physical sys em cons ain such as Bandwid h, CPU, RAM and Memo y 2 Backg ound S udies In he esea ch a ea o 5G esou ce and se ice o ches a ion, we will speci - ically discuss he cu en s a e-o - he-a p ojec s ela ed o Cloud and SDN managemen and o ches a ion likewise IoTalen um. 2.1 P ojec s Rela ed o 5G Resou ce O ches a ion Vi ual Ne wo k Func ions as a Se ice o e i ualized in as uc u es (T- NOVA) ocuses on ne wo k unc ions as a se ice (NFaaS) [9]. This p ojec p o- ides a solu ion o deploying and managing NFaaS, which allows ope a o s o c ea e and deli e new se ices e icien ly on i ualized in as uc u e. 5GEx ocuses on c ea ing an ecosys em ha allows he exchange o esou ces and se ices o and om di e en 5G expe imen al in as uc u es [3]. The aim o 5GEx is o acili a e c oss-domain expe imen a ion and alida e 5G echnolo- gies. Sona a [13] ocuses on de eloping a se ice-o ien ed 5G a chi ec u e ha 4 H. K. Janjua e al. suppo s he dynamic c ea ion and managemen o ne wo k se ices, including sma ci ies, Indus y 4.0, and i ual/augmen ed eali y. Vi al [16] is a 5G- ela ed p ojec ha ocuses on he de elopmen o an in eg a ed amewo k o he design, deploymen , and managemen o 5G ne wo ks and se ices. I ad- d esses he challenges associa ed wi h he deploymen o 5G ne wo ks, including ne wo k slicing, o ches a ion, and au oma ion. 5G T ans o me [11] aims o de elop a lexible and scalable 5G a chi ec u e ha can suppo he dynamic c ea ion and managemen o ne wo k slices. The p ojec ’s ocus is on de eloping an SDN-based a chi ec u e ha allows he e icien and lexible o ches a ion o 5G ne wo k esou ces. 5G G ow h p esen s an ecosys em o he co-c ea ion o 5G se ices and applica ions [14]. I p o ides a pla o m o allow collabo a ion be ween di e en s akeholde s o de elop and deploy new 5G se ices and appli- ca ions. Inspi e 5G Plus [15] p ojec aims o de elop an in eg a ed pla o m o c ea e, deploy, and manage 5G se ices and applica ions o enable he seamless in eg a ion o se e al 5G componen s and se ices, such as ne wo k slicing, edge compu ing, and AI. 5G Zo o aims o de elop a secu i y amewo k o 5G ne - wo ks [5]. The p ojec aims o add ess he secu i y challenges associa ed wi h he deploymen o 5G ne wo ks by p oposing a secu i y a chi ec u e ha can ensu e he con iden iali y, in eg i y, and a ailabili y o 5G ne wo k esou ces. Table 2. P ojec s ela ed o 5G esou ce O ches a ion P ojec Domain Resou ce O ches a ion Goals T-NOVA [9] Cloud, NFV, Enable end- o-end se ice p o isioning ia SDN NFVaaS, VNFaaS, NFVIaaS 5GEx [3] Cloud, NFV, Enable c oss-domain o ches a ion o 5G se ices SDN Sona a [13] Cloud, NFV De elop an in eg a ed se ice pla o m SDN Vi al [16] Cloud, NFV, De elop a amewo k o IoT and sma ci y SDN, IoT 5G T ans o me [11] Cloud, NFV, De elop SDN se ice-o ien ed 5G ne wo k SDN a chi ec u e 5G G ow h [14] NFV De elop 5G in as uc u e and se ices o u al a eas Inspi e 5G Plus [15] NFV, SDN, De elop 5G in as uc u e and se ices o DLTs, IoT IIoT 5G Zo o [5] DLTs De elop an in eg a ed pla o m o 5G se ices 3 P oposed NG2CRO F amewo k We designed a cloud-na i e SDN-based amewo k add essing cogni i e capa- bili ies using RL me hods. Also, he p oposed a chi ec u e is designed o achie e NG2CRO F amewo k 5 an in elligen , au onomous, and sel -managed ne wo k and esou ce o ches a- ion sys em. The design o he p oposed sys em is sepa a ed in o ou logical laye s ha a e; he use plane (UP), he da a plane (DP), and he con ol plane (CP) ( u he CP is subdi ided in o he ollowing: managemen , o ches a ion, and p o isioning), managemen plane (MP) and applica ion plane (AP). On he op o igu e 1 we desc ibed he use cases o NG2CRO. Also, we aimed o sol e he p oblems ela ed o ne wo k o ches a ion add essed in sec ion ?? by sepa a ing each pa o he ne wo k (NVF-I, VNF-O, and slice o ches a ion & managemen ) by sepa a ing each module in mas e -sla e a chi ec u e. He e in mas e -sla e a chi ec u e, each pa will do i s esou ce managemen on i s own bu ge he upda es om he main con olle , which can lowe he delays and la ency when ope a ions a e di ec ly pe o med by he main con olle . The main aim o he ou logical laye s is o adap he con e gence o DP in o CP and MP ha leads owa ds he AP. The da a plane pe o ms da a ans e o uppe abs ac laye s in e ms o in as uc u e and con igu a ion in o ma ion. The con ol plane pe o ms he ollowing asks ha a e, moni o ing o ne wo k unc ion i ualiza ion managemen (NFVI-M), Vi ual ne wo k unc ion o ches- a ion (VNF-O), VNF li e cycle managemen , and Slice managemen . Resou ce p o isioning and o ches a ion a e he capabili ies o CP, bu in con as , he con e gence o CP ope a ions mus be adap able o DP as he MEC en i onmen is dynamic and he e ogeneous. Mo eo e , o main ain se ice con inui y he CP con e gence o DP mus be add essed in e ms o in as uc u e adap abili y, se ice lexibili y, dynamic upda es, and esou ce p o isioning. Following, he p ocess low o NG2CRO is depic ed in igu e1 whe e he low is labeled wi h numbe s. A 1. The compa ible applica ion scena ios a e discussed and also he applica ion se ices ela ed o he scena ios a e gi en. 2. The enabling on haul and backhaul echnologies a e gi en o 5G applicabili y. 3. Nex he in o ma ion lows goes owa d and con olle in he con ol plane. He e he in o ma ion low will be in back ou h manne be ween he con olle and applica ions. 4. He e we ha e he cloud-na i e SDN con olle o suppo he cloud-na i e 5G applica ions. Fu he on, he esou ce o ches a ion ope a ions a e ca ied ou by he con- olle , he e in s ep numbe 10 he policy op imiza ion, and model agg ega ion o AI me hods is pe o med. The da a ed in o he main b ain is acqui ed om he modules which a e VIM (5) and NFVO (7). A le el 10 NG2CRO will make use o RL echniques o build he esou ce o ches a ion policies and send hese policies o NFV and VN -O o ches a o componen s. Nex , i inco po a es he TL ( ans e lea ning) echniques also o build u u e p edic ions, bu he e we can ace he p oblems wi h da a p i acy issues which is an a ea o add ess. A le el (9) bandwid h managemen is ca ied ou among 5G slices, we ha e ca ied ou he expe imen a ion o his module he e [8]. Ou u u e wo k includes u he expe imen a ion o modules 7 and 6. Las ly, he main b ain is a main con olle ha manages all o he ope a ions based on he gi en s a es as a ic in o ma ion and sends back he equi ed ac ion based on he cu en s a e and op imal policies. 6 H. K. Janjua e al. The P oposed NG2CRO is he subpa o he IoTalen um [7] p ojec and i aimed o sol e he issue in cu en s a e-o - he-a esou ce o ches a ion p ojec s by adhe ing o he s anda ds used by exis ing p ojec s. Fu he mo e, we plan o pe o m he expe imen a ion o each module o NG2CRO wi h RL echniques and in eg a e all o hem wi h he main con olle wi h he aim o op imizing he RL me hods complian wi h ne wo k managemen . 3.1 Discussion The pla o ms gi en in he able 2 ha e dis inc goals and se s o capa- bili ies, bu hey all seek o make i possible o manage and o ches a e SDN esou ces e ec i ely in a scalable and lexible way. Howe e , ce ain pla o ms could ha e es ic ions o gaps in hei SDN-speci ic unc ionali y, o hey migh no suppo se ups wi h many endo s. Fo ins ance, ce ain sys ems could lack sophis ica ed SDN-speci ic capabili ies like e ec i e low managemen , ne wo k slicing, o ne wo k unc ion chaining. A pla o m’s capaci y o se e complex, mul i- endo se ups may also be cons ained, which can make in eg a ion and adminis a ion mo e di icul . I ’s c ucial o assess he speci ic equi emen s o he ne wo k and conside he ad- an ages and disad an ages o each pla o m be o e selec ing he bes SDN man- agemen and o ches a ion pla o m. we aimed o design a pla o m ha ied o sui he needs o cu en and help o e icien ly manage and o ches a e he SDN esou ces by ca e ully conside ing hese c i e ia. The in eg a ion o RL in o SDN and NFV esou ce o ches a ion has he po en ial o ha e bene icial as well as de imen al impac s on he in e ope abili y, lexibili y, and scalabili y o he in as uc u e. Also, we s i e o in eg a e RL algo i hms in o ou wo k. he ini ial pa o in eg a e RL in o his wo k is been published in [8]. On he plus side, adding RL o SDN and NFV esou ce o ches a ion can inc ease he in as uc u e’s au oma ion and in elligence, which can u he boos in e ope abili y and adap abili y. Fo ins ance, RL algo i hms may be applied o o ecas a ic pa e ns, op imize esou ce alloca ion, and de ec abno mali ies in eal ime, allowing o mo e e ec i e use o ne wo k esou ces. Addi ionally, RL can p o ide a dynamic esponse o shi ing ne wo k se ings, enhancing he in as uc u e’s adap abili y. RL algo i hms may make SDN and NFV esou ce o ches a ion mo e sensi i e o changes in a ic pa e ns, ne - wo k opology, and use beha io by con inually lea ning and adjus ing o new scena ios. The in eg a ion o RL in o SDN and NFV esou ce o ches a ion migh hinde scalabili y and in e ope abili y, which is a se ious d awback. Fo ins ance, o RL algo i hms o be e ec i e, huge olumes o da a mus be collec ed and p ocessed ins an ly. This can make scaling di icul , especially in la ge-scale ne wo ks whe e he e is a lo o da a o manage. The lack o a gene ally adop ed s anda d o RL in eg a ion in o SDN and NFV esou ce o ches a ion migh also cause compa ibili y and in e ope abili y p oblems. This can make i challenging o ne wo k ope a o s o combine a ious NG2CRO F amewo k 7 HEALTH CARE SMART HOME GAMING IIoT Use Equipmen (UE) Radio Access Ne wo k (RAN) Mul i-Access edge compu ing (MEC) So wa e De ine Cloud compu ing (SDC) Backhaul F on haul Op imiza ion In as uc u e SDN con olle Logical i ual In as uc u e Managemen (VIM) OpenS ack Vi ual Compu e Vi ual S o age Vi ual Ne wo k Ne wo k Func ion Vi ualiza ion In as uc u e (NFVI) (CapEX/OpEx, Slice managemen , Resou ce O ches a ion by communica ion o VIM) o Se ice o ches a ion o Kube ne es o VNFM closed loop li ecycle o VNF ini ia ion o VNF e mina ion o VNF mig a ion o Resou ce o ches a ion and managemen o RAN managemen o Flow inspec ion o Ne wo k Deli e y Se ice o RAN in e ence managemen o Se ice Func ion Chain MAIN BRAIN (Fede a ion Lea ning & Global Model agg ega ion) Policy op imiza ion o bandwid h and la ency Resou ce alloca ion policy managemen Vi ual Ne wo k Func ion manage (VNFM) Ne wo k Func ion Vi ualiza ion O ches a o (NFVO) End- o-End Slice Manage o ches a o & MEC Applica ion O ches a o VNF VNF VNF VNF VNF/MEC app VNF/MEC app VNF/MEC app Compu ing esou ces Radio access moni o o on haul op imiza ion T a ic classi ica ion T a ic p edic ion RAN managemen alloca ion Radio ne wo k in o ma ion and agg ega ion ia ML VNFM li e cycle policy managemen ia esou ce p o isioning, p edic ion and op imiza ion Applica ion awa e pe o mance MEC Pla o m MEC Pla o m/MEC app manage MEC sys em de elopmen in 5G Op imiza ion a edge Local con en caching P i acy conce ns VNF li e cycle policy managemen NFVO policy op imiza ion and managemen Video analy ics o came as and UVAs in IIoT, S adium and connec ed ca s Applica ion plane AR, VR o loading and p ocessing o sma homes, heal h ca e, and digi al games Au onomous ehicle managemen and in e ac ion wi h each o he and MEC o Ac i e de ice loca ion and acking o unin e up ed gaming expe ience while use is mo ing om on UE o o he 1 2 3 4 5 6 10 7 8 9 SFC Deploymen Fig. 1. P oposed NG2CRO F amewo k 8 H. K. Janjua e al. sys ems and pa s, which can a ec he in as uc u e’s o e all in e ope abili y and adap abili y. O e all, in eg a ing RL in o SDN and NFV esou ce o ches a ion has he po en ial o ha e bene icial as well as ad e se impac s on scalabili y, lexibil- i y, and in e ope abili y. I is c ucial o ca e ully weigh he possible ad an ages and challenges o in eg a ing RL in o SDN and NFV esou ce o ches a ion as well as o build s anda ds and bes p ac ices o gua an ee in e ope abili y and scalabili y. 3.2 Conclusions In his wo k, we ad ess he he cogni i e SDN esou ce o ches a o o Io- Talen um a chi ec u e and p oposed a amewo k named NG2CRO. We discuss he exis ing s a e-o - he-a p ojec s ela ed o 5G ne wo k esou ce manage- men and o ches a ion. We discuss and analyze he exis ing lis ed p ojec s ha add ess ne wo k managemen using cogni i e echniques ac oss he cloud and SDN domains. Mo eo e , a b ie discussion is gi en p o iding a cu a ed lis o bene i s and limi a ions o an AI-enabled cloud-na i e SDN con olle o he esou ce o ches a o . Such as RL echniques will be e y use ul in e ms o im- p o ed e iciency and au oma ion, eal- ime analy ics and insigh s, and p edic i e main enance. Bu also, his in i es c ucial poin s o conside which a e p i acy issues, AI explainabili y, Ad e sa ial AI, and lack o s anda diza ion ega ding RL esou ce o ches a ion amewo ks. We aimed o add ess he complexi y and cos -incu ing p oblem o implemen ing RL echniques o esou ce o ches a ion and he p oposed NG2CRO amewo k is he ini ial s ep o his pa h. Re e ences 1. Bega, D., G amaglia, M., Pe ez, R., Fio e, M., Banchs, A., Cos a- P´e ez, X.: Ai-based au onomous con ol, managemen , and o ches a ion in 5g: F om s anda ds o algo i hms. IEEE Ne wo k 34(6), 14–20 (2020). h ps://doi.o g/10.1109/MNET.001.2000047 2. Benzaid, C., Taleb, T.: 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 wo k 34(2), 186–194 (2020). h ps://doi.o g/10.1109/MNET.001.1900252 3. Be na dos, C.J., Ge ¨o, B.P., Di Gi olamo, M., Ke n, A., Ma ini, B., Vaishna i, I.: 5gex: ealising a eu ope-wide mul i-domain amewo k o so wa e-de ined in- as uc u es. T ansac ions on Eme ging Telecommunica ions Technologies 27(9), 1271–1280 (2016) 4. Camps-Mu , D., Ga as, A., Gho aishi, M., H asnica, H., Kaloxylos, A., Anas a- sopoulos, M., Tzanakaki, A., S ini asan, G., An e ski, K., Ba anda, J., Scheppe , K., Case i, C., Chiasse ini, C., Ga cia-Saa ed a, A., Guima es, C., Kondepu, K., Li, X., Magoula, L., Malin e no, M., Cogalan, T.: Ai and ml – enable s o beyond 5g ne wo ks (05 2021). h ps://doi.o g/10.5281/zenodo.4299895 5. Chainho, P., Fe nandez-Fe n´andez, A., Fe n´andez Hidalgo, J., Gil P´e ez, M., Jo - que a Vale o, J.M., Ma ´ınez Pe ez, G., Mi sud, J.M., Hue as Celd an, A., Lekidis, A., Sanchez S´anchez, P.M., e al.: O e iew o he secu i y and us mechanisms in he 5gzo o p ojec . EUCNC 2021 (2021) NG2CRO F amewo k 9 6. Duan, Q.: In elligen and au onomous managemen in cloud-na i e u u e ne - wo ks—a su ey on ela ed s anda ds om an a chi ec u al pe spec i e. Fu u e In e ne 13, 42 (02 2021). h ps://doi.o g/10.3390/ i13020042 7. IoTalen um: Cogni i e o ches a o o mec and ne wo k esou ces. io alen- um.eu/es s (Las accessed 8 May 2023) 8. Janjua, H.K., de Miguel, I., Du ´an Ba oso, R.J., de Dios, G., Aguado, J.C., ´ Al a ez, N.M., Fe n´andez, P., Lo enzo, R.M.: E icien op imiza ion o ac o -c i ic lea ning o cons ained esou ce o ches a ion in an wi h ne wo k slicing. In: 2023 26 h Con e ence on Inno a ion in Clouds, In e ne and Ne wo ks and Wo kshops (ICIN). pp. 100–104 (2023). h ps://doi.o g/10.1109/ICIN56760.2023.10073489 9. Kou is, M.A., McG a h, M.J., Ga dikis, G., Xilou is, G., Riccobene, V., Papadim- i iou, P., T ou a, E., Libe a i, F., T ubian, M., Ba all´e, J., e al.: T-no a: An open-sou ce mano s ack o n in as uc u es. IEEE T ansac ions on Ne wo k and Se ice Managemen 14(3), 586–602 (2017) 10. La ah, M., Toke , L.: A i icial in elligence enabled so wa e de ined ne wo king: A comp ehensi e o e iew. IET Ne wo ks 8, 79–99 (2018) 11. Mangues-Ba alluy, J., Ba anda, J., Pascual, I., Ma ´ınez, R., Ve o i, L., Landi, G., Zu i a, A., Salama, D., An e ski, K., Ma ´ın-P´e ez, J., e al.: 5g- ans o me se ice o ches a o : design, implemen a ion, and e alua ion. In: 2019 Eu opean Con e ence on Ne wo ks and Communica ions (EuCNC). pp. 31–36. IEEE (2019) 12. Manias, D.M., Shami, A.: The need o ad anced in elligence in n managemen and o ches a ion. IEEE Ne wo k 35(1), 365–371 (2021). h ps://doi.o g/10.1109/MNET.011.2000373 13. Ma a , J.: Sona a app oach owa ds de ops in 5g ne wo ks. In: SDN Wo ld Cong ess (2017) 14. Nguyen, T.H., Yoo, M.: A n desc ip o gene a o o acke -based n manage- men and o ches a ion. In: 2018 In e na ional Con e ence on In o ma ion and Communica ion Technology Con e gence (ICTC). pp. 260–262. IEEE (2018) 15. O iz, J., Sanchez-Ibo a, R., Be nabe, J.B., Ska me a, A., Benzaid, C., Taleb, T., Alemany, P., Mu˜noz, R., Vilal a, R., Gabe , C., e al.: Inspi e-5gplus: In elligen secu i y and pe asi e us o 5g and beyond ne wo ks. In: P oceedings o he 15 h In e na ional Con e ence on A ailabili y, Reliabili y and Secu i y. pp. 1–10 (2020) 16. T ichias, K., Landi, G., Sede , E., Ma quez-Ba ja, J., F izzell, R., Io dache, M., Demes ichas, P.: Vi al-5g: inno a i e ne wo k applica ions (ne apps) suppo o e 5g connec i i y o he anspo & logis ics e ical. In: 2021 Join Eu opean Con- e ence on Ne wo ks and Communica ions & 6G Summi (EuCNC/6G Summi ). pp. 437–442. IEEE (2021) 17. Zhang, S., Zhu, D.: Towa ds a i icial in elligence enabled 6g: S a e o he a , challenges, and oppo uni ies. Compu e Ne wo ks 183, 107556 (12 2020). h ps://doi.o g/10.1016/j.comne .2020.107556