scieee Science in your language
[en] (orig)

Reuse of safety certification artefacts across standards and domains: A systematic approach

Abstract

The research leading to this paper has received funding from the FP7 programme under grant agreement no 289011 (OPENCOSS).

Read accessible full text

Reuse of safety certification artefacts across standards and domains: A systematic approach

Author: Ruiz, Alejandra,Juez, Garazi,Espinoza, Huáscar,De la Vara, Jose Luis,Larrucea Uriarte, Xabier
Publisher: Elsevier
Year: 2017
DOI: 10.1016/j.ress.2016.08.017
Source: https://addi.ehu.eus/bitstream/10810/78562/5/RESS2016_100.pdf
1
Reuse o sa e y ce i ica ion a e ac s ac oss s anda ds and
domains: A sys ema ic app oach
Alejand a Ruiz, Ga azi Juez, Huásca Espinoza
ICT Di ision, TECNALIA
De io, Spain
{name.su name}@ ecnalia.com
Jose Luis de la Va a
Compu e Science Depa men , Ca los III Uni e si y o Mad id,
Leganés, Mad id, Spain
[email p o ec ed]m.es
Xabie La ucea
Escuela Uni e si a ia de Ingenie ía, Uni e sidad del País Vasco,
Vi o ia, Spain
xabie .la [email p o ec ed]us
Abs ac —P oduc euse is a common p ac ice in sa e y-c i ical sys ems enginee ing. Reuse can imp o e sys em
de elopmen and assu ance, and he e a e ecommenda ions on euse o some domains. C oss-domain euse, in
which a p e iously ce i ied p oduc ypically needs o be assessed agains di e en sa e y s anda ds, has howe e
ecei ed li le a en ion. No guidance exis s o his euse scena io despi e i s ele ance in indus y, hus p ac i ione s
need new means o ackle i . This pape aims o ill his gap by p esen ing a sys ema ic app oach o euse o sa e y
ce i ica ion a e ac s ac oss s anda ds and domains. The app oach is based on he analysis o he simila i ies and on
he speci ica ion o maps be ween s anda ds. These maps a e used o de e mine he sa e y ce i ica ion a e ac s ha
can be eused om one domain o ano he and euse consequences. The app oach has been alida ed wi h
p ac i ione s in a case s udy on he euse o an execu ion pla o m om ailway o a ionics. The esul s show ha he
app oach can be e ec i ely applied and ha i can educe he cos o sa e y ce i ica ion ac oss s anda ds and
domains. The e o e, he app oach is a p omising way o making c oss-domain euse mo e cos -e ec i e in indus y.
Keywo ds: sa e y-c i ical sys em, sa e y ce i ica ion, sa e y assu ance, euse, sa e y s anda d, c oss-domain.
1 INTRODUCTION
Sa e y-c i ical sys ems a e hose whose ailu e could end up in loss o inju ies o people o he en i onmen .
These sys ems a e usually equi ed o go h ough ce i ica ion p ocesses, o sa e y assu ance p ocesses in
gene al, acco ding o some sa e y (o sa e y- ela ed) s anda d [1]. The goal o ce i ica ion is o p o ide he
di e en s akeholde s, including he socie y, he assu ance ha a sys em does no in oduce unaccep able isks o
ca as ophic consequences [2]. P esc ip i e s anda ds such as DO-178C in a ionics [3] o he gene al IEC 61508
s anda d [4] de ine p ocesses and speci ic e idence o show compliance. O he s anda ds ollow a goal-based
s a egy whe e he achie emen o compliance and o a sys em’s sa e y objec i es is documen ed in an assu ance
case. This case oughly consis s o a gumen s ha jus i y compliance and sys em sa e y, a e suppo ed by
e idence, and a e e alua ed by a ce i ica ion body. An example o his second app oach is he De 00-56
mili a y s anda d [5]. I is also common ha speci ic domains ha e hei own applicable s anda ds, such as EN
50128 o ailway [6] and ISO 26262 o au omo i e [7].
Ce i ica ion p ocesses a e labo ious and expensi e, and end o inc ease he e o and cos o de elop sa e y-
c i ical sys ems. Fo example, Boeing’s 787 D eamline ai c a needed a ce i ica ion assessmen p ocess ha
las ed eigh yea s in o de o ob ain he ai wo hiness ce i ica e [8]. Du ing ha p ocess, he Fede al A ia ion
Agency epo ed abou 200,000 hou s o echnical wo k, which we e exceeded by he hou s equi ed by
echnicians om he company. Boeing needed o p esen mo e han 4,000 documen s o show e idence o
compliance.
Sa e y-c i ical sys ems a e ypically no buil up om sc a ch as a massi e and unique elemen , bu as a
composi ion o a sys ems and subsys ems wo king in collabo a ion. This decomposi ion in o subsys ems
in oduces a good chance o euse, whe e he e o needed o compliance can be assumed by he di e en
sys ems in which a subsys em o componen is used o eused. In o de o educe he ime needed o pu a new
sys em in o he ma ke , euse o e.g. so wa e imp o es p oduc i i y and eliabili y o de elopmen p ojec s and
This is he accep ed manusc ip o he a icle ha appea ed in inal o m in Reliabili y Enginee ing and Sys em Sa e y
158 : 153-171 (2017), which has been published in inal o m a h ps://doi.o g/10.1016/j. ess.2016.08.017. © 2016 Else ie
unde CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/)
2
lowe s hei o e all cos . These bene i s could inc ease up o 50% wi h a high le el o euse [9]. The e is
guidance on how o euse componen s and subsys ems in domains such as au omo i e [7] and a ionics [10].
No only he ce i ica ion in es men can bene i om euse bu also o he ad an ages ha e been ecognised
[11]:
- Dependabili y p ope y imp o emen . Sys ems and componen s a e ho oughly e i ied and p o ed
each ime hey a e eused. The his o ical da a o hei pe o mance and o eac ions o ulne abili ies
also inc eases, p o iding he sys em o componen de elope wi h exac da a o imp o e dependabili y.
- P ocess isk educ ion. When ollowing a euse s a egy, he isk is sha ed by each o he imes he
euse has been pe o med. E en mo e, euse suppo s he applica ion o p ocess bes p ac ices,
mi iga ing he p ocess isk associa ed.
- E ec i e use o specialis s. The knowledge is encapsula ed in he di e en componen s and associa ed
a e ac s ha a e eused, libe a ing his way a specialis om mono onous wo k. Specialis s can ocus
on challenging a eas whe e hei knowledge is specially needed.
- Accele a ed de elopmen . The ime o ma ke o a p oduc is educed by eusing p e iously de eloped
p ojec s o pa o p ojec s.
Di icul y in deciding i a componen can be eused and p o ision o sa e y e idence o sys ems ha euse
exis ing sys ems and componen s ha e been acknowledged as challenges bo h in he li e a u e [12] and by
p ac i ione s [13,14]. I is also conside ed ha indus y needs mo e sys ema ic app oaches o sa e y ce i ica ion
a e ac (aka e idence) euse [15]. A majo issue a ises when aiming o euse, in a gi en domain, a sys em ha
has been p e iously ce i ied and deployed in ano he domain. This is mainly due o he di e ences be ween he
applicable sa e y s anda ds and hus be ween he ce i ica ion equi emen s.
Al hough e ec i e c oss-domain euse is gaining a en ion in indus y [16] and seems o be a e-ce i ica ion
scena io in which a ound hal o he p ac i ione s dealing wi h sa e y e idence change impac analysis ha e been
in ol ed [14], he e a e no ecommenda ions o guidance o e ed by ce i ica ion au ho i ies in o de o acili a e
p oduc euse ac oss di e en domains. C oss-domain ce i ica ion is e en mo e challenging when audi o s and
assesso s aim o iden i y he equi ed e idence o ce i ying a sys em agains he a ge ce i ica ion scheme,
based on he e idence p o ided o he sou ce scheme. Wi hou a alid analysis and unde s anding o sa e y
e idence euse consequences, a sys em will need o go h ough he en i e ce i ica ion p ocess o he a ge
domain. This will equi e a conside able amoun o ime and esou ces, educing he bene i s o p oduc euse.
These issues gene ally apply o e-ce i ica ion o a sys em agains a di e en sa e y s anda d. This can occu
bo h in c oss-domain euse and in a single domain, whe e s anda ds and ce i ica ion schemes migh e.g. a y
among coun ies.
This pape aims o con ibu e o add essing he abo e issues by p oposing a no el app oach ha suppo s he
sys ema ic euse o sa e y ce i ica ion a e ac s ac oss s anda ds and domains (he ea e e e ed o as euse
app oach). The app oach has been de eloped in he scope o OPENCOSS (h p://www.opencoss-p ojec .eu),
which is a Eu opean indus y-academia p ojec on e olu iona y ce i ica ion o sa e y-c i ical sys ems o
au omo i e, a ionics, and ailway. OPENCOSS indus y s akeholde s ha e p o ided inpu o designing he
euse app oach and ha e con ibu ed o i s alida ion.
The euse app oach is based on he speci ica ion o how simila wo sa e y s anda ds a e by mapping hem. A
pai o sa e y s anda ds can ha e commonali ies and di e ences, hus hei sa e y c i e ia can be ully o pa ially
simila o no map migh exis . The esul ing maps can be la e exploi ed o de e mine he ex en o which a
p oduc complain wi h a sa e y s anda d complies wi h he o he and hus sa e y a e ac use consequences. The
app oach has been alida ed wi h p ac i ione s in a case s udy on he euse o an execu ion pla o m om
ailway o a ionics. To he bes o ou knowledge, no o he app oach has been de eloped in o de o
sys ema ically euse sa e y ce i ica ion a e ac s ac oss s anda ds and domains.
This pape ex ends he wo k p esen ed in [17], whe e we p esen ed an o e iew o he case s udy and epo ed
on he expe ience in mapping EN 50128 and DO-178. The ex ension is mainly based on: (1) a mo e de ailed
explana ion o he backg ound, he ounda ions, and he a ionale o he euse app oach; (2) a gene al and
gene ic desc ip ion o he p inciples o he app oach and o he p ocess o apply i , indica ing and explaining all
he necessa y ac i i ies, and; (3) a deepe p esen a ion and analysis o he alida ion o he app oach, including
he epo ing and discussion o u he esul s. This ex ension allows a eade o gain a wide unde s anding o
how he euse app oach has been de eloped, i s basis, how o apply he app oach, and he bene i s o i s
applica ion.
The es o he pape is o ganized as ollows. Sec ion 2 p esen s he backg ound o he pape . Sec ion 3 desc ibes
he euse app oach, and Sec ion 4 epo s on i s alida ion. Sec ion 5 summa ises ou conclusions and u u e
wo k.
3
2 BACKGROUND
This sec ion p esen s he main backg ound o he pape , which co esponds o (1) he cu en s a e o he
p ac ice, (2) a compa ison o sa e y s anda ds, and (3) he ela ed li e a u e.
2.1 S a e o he p ac ice
The pu pose o his sec ion is o desc ibe how sa e y-c i ical sys em ce i ica ion is handled in p ac ice. To his
end, he sec ion analyses sa e y ce i ica ion and euse in a ionics and ailway, he wo domains add essed in he
case s udy wi h which he euse app oach has been alida ed. The compa ison o sa e y s anda ds in he nex
sub-sec ion complemen s his analysis by p esen ing a b oade o e iew o how he s anda ds in u he domains
a e and o hei simila i ies and commonali ies.
Ce i ica ion is de ined in ci il a ia ion as a legal ecogni ion ha a p oduc , se ice, o ganiza ion, o pe son
complies wi h he equi emen s s a ed in a ce ain s anda d. This implies echnically checking he objec o
ce i ica ion o e i y o mally ha i complies wi h he applicable equi emen s. Fo ce i ying a p oduc , he
au ho i y should assess he design p ocess o he p oduc o ensu e an accep able le el o sa e y, check whe he
he p oduc ac ually con o ms o he expec ed design, and issuance a ce i ica e equi ed by he na ional laws o
show ha he p oduc has gone h ough he assessmen s p ocess [3].
Che el [18] desc ibes he di e en ac o s in ol ed in ai c a manu ac u ing. Fi s , he ai c a manu ac u e
ag ees wi h he a ionics au ho i y o he coun y upon he ype ce i ica e. This ce i ica e will include he i s
de ini ion o he p oduc wi h documen s de ining he ai c a cha ac e is ics. This is done a he e y beginning
o he design phase. The manu ac u e will hen make a con ac wi h he di e en a ionics sys em de elope s o
con ac he de elopmen o one o mo e sys ems o he ai c a and eques hei con ibu ion o he
ai wo hiness ce i ica ion p ocess. Sys em de elopmen supplie s should also con ac wi h he au ho i y in o de
o ge a Technical S anda d O de au ho iza ion o ensu e ha hei sys em is complian wi h he a ionics
s anda ds. Ge ing his au ho iza ion does no mean ha he sys em will be ce i ied. The ai c a manu ac u e
has o discuss wi h he au ho i y in o de o ge he au ho isa ion o ins alling he sys em on an ai c a . A e he
ins alla ion, he comple e ai c a goes h ough a sa e y assessmen and i is a e all he e alua ion p ocess ha
he ai c a is eady o ge he ai wo hiness ce i ica ion.
In he a ionics domain, he DO-297 s anda d [10] and he ad iso y ci cula AC 20-148 [19] deal wi h euse.
DO-297 appea ed as a consequence o he mo e om ede a ed a chi ec u es o IMA (In eg a ed Modula
A ionics) a chi ec u e, and AC 20-148 is he esul o c ea ing guidelines o so wa e componen euse.
IMA is he e m used o a dis ibu ed compu ing ne wo k aboa d ai c a , which suppo s a ionics applica ions
o many di e en assu ance le els, and i is designed o lexibili y in con igu a ions and modula i y. I suppo s
assu ance e idence euse o he educe e o equi ed when eusing componen s in di e en sys ems.
Compliance wi h DO-297 aims o educe he cos o main enance and ce i ica ion. IMA echnology has
in oduced he possibili y o agmen he ce i ica ion p ocess in o se e al asks: (a) module and/o pla o m
accep ance; (b) applica ion accep ance (so wa e and ha dwa e); (c) IMA sys em accep ance (in eg a ion o
mul iple applica ions); (d) ai c a in eg a ion; (e) change o modules o applica ions; and ( ) euse o modules o
applica ions. IMA aims o enable he euse o applica ions om di e en a ge sys ems wi hou inc easing he
ce i ica ion cos s. The IMA pla o m a chi ec ole es ablishes a ce i ica ion baseline abou sizing hypo hesis
(memo y, p ocesso h oughpu ), applicable ce i ica ion s anda ds (e.g., DO-254 and DO-178C), and
unc ionali y expec ed (e.g., API A653). The module supplie p o ides wha DO-297 calls he usage domain
(cha ac e is ics and usage cons ain s and quali ica ion ma e ial o ce i ica ion demons a ion.
Howe e , applying DO-297 is no an easy ask. E eleens [20] indica es ha one o he challenges o eusing an
IMA is he lack o su icien suppo o dealing wi h changes made in exis ing IMA sys ems o when eusing
design elemen s. The e is a need o jus i ica ion in o de o euse p e-quali ica ion documen s due o he numbe
o accep ance c i e ia, sa e y a gumen s, and e idence ha need o be conside ed in a new in eg a ion p ojec .
Ad iso y ci cula s in he a ionics domain a e documen s issued by au ho i ies (e.g., he Fede al A ionics
Agency om USA). These ci cula s a e no s anda ds bu a e in ended o p o ide guidance on accep ed
compliance means o speci ic opics. Fo ins ance, AC 20-148 p o ides ecommenda ions conce ning eusable
so wa e componen s. This ad iso y ci cula indica es ha , o euse componen s, s akeholde s mus iden i y any
ins alla ion, sa e y, ope a ional, unc ional, and pe o mance possible conce ns. Al hough AC 20-148 is no a
s anda d, i s applica ion is highly ecommended when using Reusable So wa e Componen s: “ he so wa e, i s
suppo ing RTCA/DO-178B so wa e li e cycle da a, and o he suppo ing documen a ion being conside ed o
euse. The componen designa ed o euse may be any collec ion o so wa e, such as lib a ies, ope a ing
sys ems, o speci ic sys em so wa e unc ions” [19]. AC 20-148 dis inguishes be ween wha he componen is
om he de elope pe spec i e and om he in eg a o one. This esembles p econdi ions and pos condi ions
ha should be accomplished o sui able euse.
4
Rega ding ailway, he EN 50126 s anda d [21] co e s he speci ica ion and demons a ion o sa e y o all
ailway applica ions and a all le els o such applica ions, as app op ia e, om comple e ailway ou es o majo
sys ems wi hin a ailway ou e, and o indi idual and combined sub-sys ems and componen s wi hin hese majo
sys ems. This includes so wa e and ha dwa e. The s anda d also add esses eliabili y, a ailabili y, and
main ainabili y as essen ial aspec s o a ailway sys em ha con ibu e o sa e y. EN 50126 se es as he en y
poin o pa en s anda d o o he ailway s anda ds, such as EN 50128 o so wa e [6] and EN 50129 o
elec onic sys ems o signalling [22].
The p ocesses ha de ine he sa e y li ecycle o a ailway sys em can be ailo ed, p o ided ha he modi ica ions
do no ha e any consequence on he s anda d sa e y li ecycle and a e well mo i a ed. Fo each phase in he
design, he ac i i ies o be ca ied ou o sa e y assu ance will be execu ed in pa allel. The sa e y o a sys em is
mean as he p ope y ha ailu e a es o po en ially dange ous consequences a e low enough, o globally educe
he isk (i.e. he p obabili y o inju ies, a ali ies, damages) o a speci ied accep able alue. The p ocess equi es
he applica ion o EN 50129, which lis s ac o s ha in luence eliabili y, a ailabili y, main enance, and sa e y as
de ined in EN 50126. The CENELEC Applica ion Guideline (TR 50506-2 [23]) p o ides addi ional in o ma ion
on he applica ion o he s anda d o achie e he case o sa e y, and includes ma e ial conce ning Sa e y
Assessmen , Sa e y App o al, and C oss-Accep ance (i.e., euse).
EN 50129 de ines how he condi ions o sa e y accep ance and app o al shall be p esen ed. The condi ions shall
co e h ee majo hemes: (1) Quali y Managemen ; (2) Sa e y Managemen , and; (3) Func ional and Technical
Sa e y. The documen a y e idence ha hese condi ions ha e been sa is ied shall be included in a s uc u ed
sa e y jus i ica ion documen known as he Sa e y Case. Accep ance by quali ied o ganisa ions and na ional
egula o y bodies o he Sa e y Case, h ough ac i i ies o app o al, assessmen , and c oss accep ance, is he
ul ima e s ep o allow a ailway sys em o en e passenge se ice.
2.2 Compa ison o s anda ds
Se e al egula ions ha e been issued o each sa e y-c i ical domain. The s anda ds a e ela ed o he
de elopmen , implemen a ion, alida ion, and main enance o sa e y-c i ical sys ems. Some examples a e:
● IEC 61508 (gene ic), o unc ional sa e y o elec ical/elec onic/p og ammable elec onic sa e y-
ela ed sys ems
● ISO 26262 (au omo i e), o unc ional sa e y o oad ehicles
● EN 50126 ( ailway), o he speci ica ion and demons a ion o eliabili y, a ailabili y, main ainabili y
and sa e y
● EN 50128 ( ailway), o so wa e o ailway con ol and p o ec ion sys ems o communica ions,
signalling and p ocessing sys ems
● EN 50129 ( ailway), o sa e y- ela ed elec onic sys ems o signalling o communica ions, signalling
and p ocessing sys ems
● DO-178C (ae ospace), o so wa e conside a ions in ai bo ne sys ems and equipmen ce i ica ion
● DO-254 (ae ospace), o design assu ance o ai bo ne elec onic ha dwa e
● SAE-ARP 4754/4754A (ae ospace), o de elopmen o ci il ai c a and sys ems
● SAE-ARP 4761 (ae ospace), o conduc ing he sa e y assessmen p ocess on ci il ai bo ne sys ems and
equipmen .
O he s anda ds include IEC 62304, IEC 60601, IEC 14971 o medical equipmen , IEC 61513 o nuclea
ene gy, IEC 62061 o indus ial machine y, IEC 61511 o indus ial p ocesses, IEC 61800 o elec onic
con ol mo o s, and ISO 10218 o obo s. S anda ds a e also used in e.g. de ence and space.
Since a “common language” o sa e y is a e y long way o , a leas a clea unde s anding o simila i ies and
di e ences o in o m euse is needed. Di e en aspec s need o be analysed in he s anda ds in o de o add ess
he di e ences and simila i ies among hem. I has o be no ed ha a deep, comp ehensi e compa ison is ou o
he scope o his wo k. Howe e , i is impo an o unde s and he mos ele an issues.
1) Objec i es. Classi ica ion o s anda ds can be done based on di e en c i e ia.
● P esc ip i e, no ma i e, in o ma i e. Since no ma i e ones a e absolu ely manda o y, he
co esponding domain-speci ic p oduc needs o comply wi h ha s anda d. Con e sely, he in o ma i e
ones p o ide added in o ma ion and guidance on he use o he a o emen ioned ones, aiming o
acili a e hei applica ion.
● P ocess-o ien ed, objec i es-o ien ed, and p oduc -o ien ed. Whe eas au omo i e and a ionics ollow
in eg a ed sa e y, ailway p e e s he so-called ex e nal sa e y whe e his a ibu e is moni o ed and
gua an eed by a di e en speci ic sys em. The e a e basically wo app oaches o de ining he
implica ion o sa e y equi emen s: objec i es-o ien ed (p ocess-o ien ed) e sus p oduc -o ien ed. The
o me speci y equi emen s implica ions as objec i es, whe eas he la e de ines cons ain s on wha is
possible o obse e on an a e ac . P ojec managemen and independen assessmen a e p ocess-
o ien ed ac i i ies. Ha dwa e and so wa e design o coding ules a e usually p oduc -o ien ed.
5
2) Te minology/Vocabula y. Al hough he e is some common e minology used ac oss he di e en sa e y
s anda ds, some imes hei de ini ions do no absolu ely ma ch o e en a e di e en om gene al de ini ions. I
migh happen ha de ini ions abou common dependabili y e ms such as aul , e o , ailu e, sa e y, o some
o he e ms like andom ha dwa e and sys ema ic aul s, a e ei he di e en ly used o e en no conside ed
wi hin hem. An example o be highligh ed conce ns e o de ini ion. Fo example, Lap ie de ines i as he pa
o he sys em s a e ha may cause a subsequen ailu e [24], whe eas DO-178C conside s i a mis ake in design,
code o equi emen s and a aul i s mani es a ion. Di e ences also exis in common e ms like e i ica ion,
alida ion, sa e y, assessmen , and ce i ica ion.
3) Reuse. The a ionics domain uses comme cial-o - he-shel componen s, which we e o iginally designed o a
non-ae ospace ma ke . The ad iso y ci cula AC 20-148 p esen s he so wa e componen o euse as a
comme cial-o - he-shel componen , and DO-297 add esses module o applica ion euse o IMA pla o ms.
ISO 26262 includes in o ma ion ega ding he sa e y elemen ou o con ex , bu he guidelines p o ided a e e y
high-le el conside a ions. I is when dealing wi h he ha dwa e and so wa e quali ied componen concep s ha
we can go deepe in o he knowledge o he ac ual equi emen s o compliance. IEC 61508 uses ano he
ele an euse concep , he so-called sa e y manual o he quali ied i em.
4) Sa e y Li ecycle and Sa e y Managemen /Li ecycle. The e a e sa e y li ecycle simila i ies and di e ences
wi h ega d o ac i i ies ela ed o sa e y compliance and o he planning o such ac i i ies wi hin he sys em
de elopmen p ocesses. To s a wi h, some s anda ds de ine a p ecise sa e y managemen p ocess (e.g., ISO
26262 and IEC 61508), whe eas e.g. DO-178C does no and he numbe o equi ed p ocesses di e s. The
p o ision o a sa e y case (i.e., a documen ed body o e idence ha p o ides a con incing and alid a gumen
ha a sys em is adequa ely sa e o a gi en applica ion in a gi en en i onmen ) is compulso y in ailway while
highly ecommended in domains.
5) Ha dwa e De elopmen Li ecycle. Some s anda ds such as ISO 26262 conside ha dwa e p ocess and
p oduc in eg i y wi hin he same unique s anda d, whils a ionics es ablished a sepa a e speci ic s anda d (DO-
254). ISO 26262 and IEC 61508 include he de ini ion o speci ic ailu es a es pe in eg i y le el and di e en
ha dwa e me ics o be achie ed. Di e ences can be ound no only in he ypes o aul s bu also in he
ha dwa e me ics o calcula e. In au omo i e, speci ic maximum ailu e a e alue pe ASIL (Au omo i e Sa e y
In eg i y Le el) is es ablished, bu his is no de ined in all sa e y s anda ds. Conce ning ep og ammable
ha dwa e, i is no di ec ly conside ed in ISO 26262, whe eas DO-254 ackles his aspec .
6) So wa e De elopmen Li ecycle. No all he sa e y s anda ds p esc ibe a speci ic so wa e de elopmen
li ecycle. ISO 26262 is one o he s anda ds ha explici ly de ine a V-Model o de elop so wa e.
7) Sa e y Ca ego ies o Le els o In eg i y. Unde a ious names bu add essing he same aim, hey cons i u e
a undamen al basis o sa e y s anda ds. They a e called Sa e y In eg i y Le els (SILs) in IEC 61508 and
ailway, ASIL in he au omo i e domain, and De elopmen Assu ance Le el (DAL) in a ionics. All o hem
depic he isk and he e ec s o he po en ial ailu es o he conside ed sys em, making possible o quan i y he
sa e y le el o a sys em and consequen ly o e alua e c i icali y. Thus i associa es a alue ha cha ac e izes how
much he sa e y depends on he absence o ailu es om he sys em unde conside a ion. The highe he in eg i y
le el is, he mo e exhaus i e he de elopmen and V&V p ocesses need o be so ha de elopmen aul s a e
a oided as much as possible. This implies ha he equi emen s o comply wi h inc ease as he sa e y le el does.
This is he case when dealing wi h speci ic echniques o me hods o a ce ain design phase. Depending on he
c i icali y le el o add ess, ei he complemen a y echniques o mo e exhaus i e ones a e needed o comply wi h
he s anda d (see echniques). Railway and au omo i e unc ional sa e y s anda ds associa e a speci ic SIL/ASIL
wi h a maximum mean ime o ailu e o minimum (dange ous) ailu e a e. In a ionics, he e is no a emp o
nume ically e alua e he p obabili y o ailu e due o such aul s, bu o conside ha ul illing s anda d’s
equi emen s p o ides a le el o con idence compa ible wi h he se e i y o he isk.
8) Haza d Analysis and Risk Assessmen (HARA). The haza ds ela ed o he sa e y- ela ed unc ions a e
add essed in all he s anda ds ollowing a sys ema ic analysis. Acco ding o ISO 26262, HARA is he me hod o
iden i y and ca ego ize haza dous e en s o i ems and o speci y sa e y goals and ASILs ela ed o he p e en ion
o mi iga ion o he associa ed haza ds in o de o a oid un easonable isk. Fu he mo e, his s anda d speci ies
me hods such as FMEA (Failu e Modes and E ec s Analysis) o b ains o ming in o de o de i e he possible
haza ds. The impac ac o s such as se e i y o exposu e used o de e mine he co esponding in eg i y le els can
di e o some ex en among domains. Fo example, au omo i e akes con ollabili y in o accoun , and in
a ionics se e i y is implici ly conside ed as high exposu e.
9) Ve i ica ion and Valida ion. Ve i ica ion usually s ands o he de e mina ion o comple eness and o
comple eness and co ec speci ica ion o implemen a ion o equi emen s om a phase o sub-phase [7]. E en
hough he de ini ions migh be a bi di e en among s anda ds, he aim is he same in all o hem. Conce ning
he applied me hods, hey can di e . Common ones a e analysis and e iew.
The same applies o alida ion. To be mo e p ecise, ISO 26262 e e s o his e m as “sa e y alida ion”, whe e
me hods such as e iews, ep oducible es s wi h pass/ ail c i e ia, o analyses can be applied o assu e ha he
sa e y goals a e su icien and ha e been achie ed. Valida ion in IEC 61508 s ands o sa e y alida ion as well

6
and simila me hods a e implemen ed (e.g., es ing and s a ic/dynamic analysis). Howe e , DO-178C does no
speci y any me hod o be applied and i s de ini ion is based on he comple eness o he speci ied equi emen s.
10) Techniques. The applied echniques o me hods do no only depend on he ype o s anda d bu also on he
c i icali y le el and he p oduc de elopmen phase. Fu he mo e, some echniques can be ecommended o
highly ecommended whe eas o he s a e s ic ly manda o y. Se e al echniques a e usually lis ed o each
de elopmen phase. A ull mapping o all he equi ed echniques is ou o he scope o his wo k. Howe e ,
some o he mos ema kable a e aul injec ion, o mal me hods, and FMEA (Failu e Modes and E ec s
Analysis), FMEDA (Failu e Modes E ec s and Diagnos ics Analysis), FTA (Faul T ee Analysis), DFA
(Dependen Failu e Analysis), es ing, walk- h ough, and simula ion. Mos o hese echniques appea in
di e en sa e y s anda ds. None heless, he e a e simila i ies and di e gences in e ms o li ecycle phase,
objec i e and scope o ecommenda ion le el.
11) Tool quali ica ion. Failu e o au oma e complex e i ica ion and de elopmen ac i i ies can comp omise
sys em sa e y. In o de o mi iga e his isk, in eg i y equi emen s in e ms o ool quali ica ion can be p o ided.
Mos domains ha e in oduced ool quali ica ion equi emen s conce ning planning, documen a ion,
classi ica ion analysis, quali ica ion epo ing, and con idence, and ca ego ize ools based on he po en ial impac
o i on he applica ion. Rega ding equi emen s on ool de elopmen , DO-330 is cu en ly he s anda d wi h
mo e elabo a ed equi emen s.
12) Secu i y Aspec s. Secu i y- ela ed sa e y issues a e s a ing o be s a e o he a and some hing i is qui e
clea : he e is no sa e y wi hou secu i y. Howe e , no all he s anda ds conside secu i y aspec s wi hin hei
equi emen s, and he end seems o be he gene a ion o complemen a y secu i y s anda ds.
2.3 Rela ed li e a u e
Rela ed li e a u e on he euse o ce i ica ion a e ac s ac oss s anda ds and ac oss domains can be ega ded as
limi ed. None heless, some au ho s ha e analysed euse needs and p oposed app oaches o suppo a e ac
euse. We e iew his kind o publica ions in his sec ion.
Reusing a e ac s om one domain in o o he s is no a s aigh o wa d ac i i y, and i equi es a nego ia ion o
i s euse [25]. The e a e sca ce indus ial epo s desc ibing componen euse such as he one desc ibed in [26].
In he so wa e enginee ing domain, se e al esea ch wo ks ha e been ocused on eusing componen s [27,28],
a chi ec u es [29], and echniques [28,30,31]. Some app oaches a ge ing euse a e also applied in sec o s such
as manu ac u ing [32], whe e cos -e ec i eness o euse is conside ed o decide upon euse. The au omo i e
indus y also euses some pa s o hei componen s [33,34].
Reusing an a e ac ha has been p e iously ce i ied in one domain implies a wide and deep analysis o i s use
in a di e en domain o p ojec . I is especially ele an when human li es a e in ol ed o hey migh be
a ec ed by a misbeha iou o ailu e o a sys em. Reusing a p ojec is no s aigh o wa d and is e en mo e
di icul when he con ex changes, as o example in c oss-domain euse. This can be a eason o why e y ew
a emp s ha e been made. Zelle e al. [35] p opose a c oss-domain assu ance p ocess in conjunc ion wi h a
de elopmen me hodology o sa e y- ele an so wa e. The objec i e was o educe he e o equi ed o
pe o m a sa e y assessmen by eusing sa e y analysis echniques and ools as well as a e ac s p oduced du ing
he sa e y assu ance p ocess. The p ocess consis ed o gene ic and domain-speci ic s eps ha mus be execu ed
in each o he conside ed domains as well as s eps ha a e only necessa y in speci ic domains. The au ho s we e
able o euse echniques and ools o sa e y analysis on di e en domains. Howe e , no all o he phases o
hei p oposed p ocess we e domain-independen and sa e y ce i ica ion a e ac euse was no conside ed in
hei esea ch. Papadopoulos and McDe mid de eloped a simila app oach [36].
Sa eCe (h p://sa ece .eu/) is a Eu opean esea ch p ojec s ela ed o euse o sa e y ce i ica ion a e ac s ac oss
s anda ds and domains. The Sa eCe p ojec add essed an indus ial use case on he euse o ool quali ica ion
ac oss domains [37]. The p oposal is based on h ee pilla s: (1) c oss-domain equi emen s spanning di e en
s anda ds, (2) c oss-domain de elopmen p ocess acco ding o he associa ed s anda ds and hei in eg i y le els,
and (3) c oss-domain ools, ins an ia ed acco ding o he associa ed s anda d. They p oposed a ool quali ica ion
p ocess line o enable he euse ac oss domain o p ocess elemen s. Gallina e al. [38] s a ed ha his app oach
also suppo s he euse o ce i ica ion a e ac s by ela ing he p ocess line wi h he co esponding amily o
p ocess-based a gumen s ela ed o p ocess compliance. Howe e , his has no been shown and a sys ema ic
p ocess o sa e y ce i ica ion a e ac euse has no been speci ied. Gallina and Sza má i [39] ha e also
p oposed he use o on ologies o iden i ica ion o commonali ies and di e ences among sa e y p ocesses.
Model-d i en app oaches can also applied o ce i ica ion pu poses [40]. New sys ems can be composed o
subsys ems s emming om di e en domains [41], bu i is no clea how an a e ac can be eused in his
con ex . Some app oaches ely on sa e y cases ( hus on a gumen s) (e.g., [42]). Howe e , wi h his app oaches
he enginee needs o in e p e he equi emen s and objec i es o he s anda ds ha will apply o he speci ic
si ua ion and some imes his is open o in e p e a ions.
7
3 APPROACH FOR REUSE OF SAFETY CERTIFICATION ARTEFACTS
As explained in he p e ious sec ion, p ac ices o sa e y-c i ical sys em enginee ing and assu ance a y among
domains. The e a e no only simila i ies bu also di e ences be ween he applicable sa e y s anda ds. The e o e,
any app oach a ge ing sa e y ce i ica ion a e ac euse mus p o ide a sys ema ic way o check commonali ies
and iden i y he di e ences, and o ackle hem. Al hough ecommenda ions on euse can be ound o some
domains, no guidance has ye been p o ided on how o sys ema ically euse sa e y ce i ica ion a e ac s ac oss
s anda ds and domains. The ela ed li e a u e has also p o ided e y ew insigh s in o how o euse sa e y
ce i ica ion a e ac s ac oss s anda d and domains. A new, sys ema ic app oach is necessa y o eusing sa e y
ce i ica ion a e ac s ac oss s anda ds and domains.
This sec ion p esen s he app oach de eloped in he OPENCOSS p ojec o euse sa e y ce i ica ion a e ac s
ac oss s anda ds and domains. We i s in oduce he o e all app oach o e olu iona y ce i ica ion o sa e y-
c i ical sys ems and hen he p inciples and he p ocess o euse o sa e y ce i ica ion a e ac s. Bo h he
p inciples and he p ocess a e suppo ed by he ool pla o m de eloped in OPENCOSS [43]. This shows ha
hey can be implemen ed. We use in o ma ion om he DO-178C a ionics s anda d [3] and he EN 50128
ailway s anda d [6] as unning examples. These s anda ds ha e been used in he alida ion o he euse
app oach (Sec ion 4). Finally, we discuss p ac ical conside a ions o he euse app oach.
3.1 OPENCOSS app oach o e olu iona y ce i ica ion o sa e y-c i ical sys ems
OPENCOSS is a la ge-scale Eu opean esea ch p ojec on sa e y assu ance and ce i ica ion o embedded
sys ems. The OPENCOSS conso ium comp ises ou academic pa ne s and 13 companies, including sa e y-
c i ical sys em manu ac u e s, componen supplie s, ce i ica ion au ho i ies, sa e y assesso s, and ool endo s.
The p ojec is also suppo ed by a la ge ad iso y boa d wi h ep esen a i es om mo e han 20 in e na ional
o ganiza ions.
The p ojec has (1) de ised a common ce i ica ion amewo k ha spans di e en e ical ma ke s o ailway,
a ionics, and au omo i e, and (2) de eloped an open-sou ce sa e y ce i ica ion in as uc u e. The ul ima e goal
o he p ojec is o b ing abou subs an ial educ ions in ecu ing sa e y ce i ica ion cos s and a he same ime
educe ce i ica ion isks h ough he in oduc ion o mo e sys ema ic sa e y assu ance p ac ices. The p ojec
deals wi h: (1) c ea ion o a common ce i ica ion concep ual amewo k; (2) composi ional ce i ica ion; (3)
e olu iona y chain o e idence; (4) anspa en ce i ica ion p ocess; and (5) compliance-awa e de elopmen
p ocess.
The euse o sa e y ce i ica ion a e ac s ac oss s anda ds and domains is mos ly enabled by he common
ce i ica ion concep ual amewo k. The main objec i e o he amewo k is o c ea e a language ha can be used
in di e en domains o desc ibe sa e y- ela ed in o ma ion, s anda ds, and p ojec s. Such a language acili a es
he analysis and he compa ison o sa e y s anda ds, and he euse o sa e y- ela ed in o ma ion ac oss p ojec s.
This includes p ojec s unde di e en sa e y s anda ds o in di e en domain.
Fig. 1 ske ches he app oach de ined in OPENCOSS o e olu iona y ce i ica ion o sa e y-c i ical sys ems. The
app oach is model-based and is suppo ed by se e al me amodels a ge ed a di e en sa e y assu ance and
ce i ica ion needs. The se o me amodels co esponds o he common ce i ica ion concep ual amewo k.
The Re e ence Assu ance F amewo k Me amodel suppo s he speci ica ion o he sa e y compliance needs ha
ha e o migh ha e o be conside ed in an assu ance p ojec . The needs can be speci ied by means o e e ence
assu able elemen s in he o m o e e ence equi emen s o ul il, e e ence a e ac o manage, and e e ence
ac i i ies o execu e. Sa e y compliance needs can be om speci ic s anda ds, ecommended p ac ices, o
company-speci ic p ac ices, and ypically ha e o be ailo ed o p ojec -speci ic cha ac e is ics. The la e is done
by means o baselines, which co espond o he speci ic sa e y c i e ia o a s anda d wi h which a gi en
assu ance p ojec has o show compliance. A baseline is usually a subse o all he sa e y c i e ia p esen in a
s anda d and a ies among p ojec s. Fo example, he sa e y c i e ia will a y i a sys em is de eloped using
model-based echniques.
Ano he sou ce o in o ma ion o sa e y compliance is he da a abou he p oduc o which compliance is
sough . The e o e, he me amodels also include he concep s and ela ionships necessa y o modelling and
managing p ojec -speci ic in o ma ion. This in o ma ion needs o be eco ded ega dless o which sa e y
s anda d is being ollowed. OPENCOSS has de ined me amodels o modelling he p ocess execu ed o c ea e a
p oduc (P ocess Me amodel), he e idence o sa e y and o compliance (E idence Me amodel), and he
a gumen s ha will be used o jus i y key sa e y- ela ed decisions aken du ing he p ojec (A gumen a ion
Me amodel). The co esponding model ep esen s he assu ance asse s o a p ojec .
Two o he me amodels a e p oposed. The Vocabula y Me amodel is a means o de ine and eco d he e ms and
concep s used o cha ac e ize eusable assu ance asse s such as e idence, a gumen a ion, and p ocess da a.
Finally, he e is a Mappings Me amodel. Maps can be speci ied be ween ocabula y e ms (e.g. om di e en
domains), be ween he assu ance in o ma ion ga he ed du ing a p ojec and i s baseline o indica ing
8
compliance, and be ween sa e y s anda ds (i.e. e e ence assu ance amewo ks) o indica ing how he s anda ds
ela e and i equi alences exis be ween hem. The la e is a key o euse o sa e y ce i ica ion a e ac s ac oss
di e en s anda ds and domains. In gene al, he mappings aim o allow enginee s and manage s o make
in o med decisions abou he app op ia eness and implica ions o eusing assu ance in o ma ion ac oss p ojec s,
sa e y s anda ds, and domains.
Fu he in o ma ion abou he OPENCOSS app oach o e olu iona y ce i ica ion o sa e y-c i ical sys ems can
be ound in [44].
Fig. 1. O e iew o he OPENCOSS app oach o sa e y assu ance and ce i ica ion [45]
3.2 P inciples o euse o sa e y ce i ica ion a e ac s ac oss s anda ds and domains
The applica ion o he euse app oach is based on p inciples elici ed om cu en p ac ices and needs (see
Sec ion 2), and discussed wi h OPENCOSS indus y pa ne s. The e a e ou main p inciples: he in en o a
sa e y ce i ica ion a e ac mus be aken in o accoun when aiming o euse i , maps mus be es ablished
be ween he sou ce s anda d ( euse om) and he a ge one ( euse o), p ojec compliance mus be de e mined
(by means o maps), and needs and gaps esul ing om sa e y ce i ica ion a e ac euse mus be de e mined.
1) Sa e y ce i ica ion a e ac in en
Sa e y ce i ica ion a e ac euse is no a challenge pe se. In heo y, any a e ac is eusable. The main euse
need s ems om he ac ha each sa e y s anda d has i s own equi emen s o ul il, and such equi emen s can
a y among s anda ds. Fo example, DO-178C lis s objec i es ( equi emen s) o he di e en so wa e
de elopmen p ocesses, and some objec i es a e no ully add essed in EN 50128. When eusing a sa e y
ce i ica ion a e ac , i mus be de e mined wha equi emen s o he a ge s anda d a e ul illed when he
sou ce s anda d is complied wi h. Sa e y ce i ica ion a e ac euse can be ega ded as he p ocess a ge ed a
de e mining wha equi emen s o a gi en sa e y s anda d a e ul illed when compliance wi h ano he s anda d
has been achie ed.
9
The abo e need can only be me i he s anda d’s equi emen s o whose ul ilmen a sa e y ce i ica ion a e ac
con ibu es a e eco ded. These equi emen s co espond o he a e ac in en : wha p ope ies a e assu ed in he
a e ac and hus why he a e ac is necessa y. Fo example, he DO-178C So wa e Requi emen s Da a mus
include he pe o mance c i e ia, iming equi emen s and cons ain s, and memo y size cons ain s so ha he
a e ac ul ils i s in en (i.e., o show ha such cha ac e is ics ha e been conside ed and speci ied).
Sa e y ce i ica ion a e ac in en is also based on he ac i i ies ha use o p oduce he a e ac . In gene al, and i
we hink o ac i i ies ha use and p oduce se e al sa e y ce i ica ion a e ac s, he o e all aim o an ac i i y
co esponds o a highe -le el in en o he indi idual in en o he ou pu a e ac s o he ac i i y. The
achie emen o his highe -le el in en is also enabled by he indi idual in en o he inpu a e ac s o he
ac i i y. Fo example, EN 50128 In eg a ion P ocess (ac i i y) aims o demons a e ha so wa e and ha dwa e
in e ac co ec ly o pe o m hei in ended unc ions. To his end, he ac i i y uses he So wa e In eg a ion Tes
Speci ica ion and he So wa e/Ha dwa e In eg a ion Tes Speci ica ion as inpu , and p oduces he So wa e
In eg a ion Tes Repo and he So wa e/Ha dwa e In eg a ion Tes Repo as ou pu .
2) Equi alence mapping be ween s anda ds
In addi ion o eco ding he in en o he sa e y ce i ica ion a e ac s, i is also necessa y o de e mine he
equi alence be ween s anda ds o sa e y ce i ica ion a e ac euse. This can be done by means o maps ha
indica e he ex en o which he c i e ia o he s anda ds a e equal (e.g., be ween EN 50128 So wa e
Requi emen s Speci ica ion and DO-178C So wa e Requi emen s Da a). Based on hese mappings, he
simila i y and di e ences o he s anda ds can be assessed, and hus how complian a sa e y ce i ica ion a e ac
is wi h a gi en s anda d acco ding o i s compliance wi h ano he s anda d.
Th ee gene al ypes o maps can exis be ween he elemen s o wo s anda ds:
● Full map: he elemen s in he mapping a e iden ical; he cha ac e is ics o he elemen in i s o iginal
con ex (i s o m, i s equi ed con en , i s p econdi ions, i s objec i es, i s pos -condi ions on i s use...)
ully sa is y he equi emen s o he con ex in which i is o be eused.
● Pa ial map: he elemen s a e simila , bu hey a e no iden ical; depending on he con ex and he
objec i es, he di e ences be ween hem migh be signi ican ; in his case, a clea eco d o he
simila i ies and di e ences is equi ed.
● No map: he e is insu icien simila i y be ween he elemen s o enable us o asse a map; in his case, i
may be impo an o eco d he di e ences, and he easons why he mapping is disallowed, in o de o
in o m u he gap analysis and p e en inad e en euse.
Full maps a e usually a e in he assu ance domain and he majo i y o maps a e pa ial.
Th ee elemen s play a ole in equi alence mapping: a e ac s, ac i i ies, and equi emen s. P agma ically, any o
hese e e ence assu able elemen s can be eused. The accep abili y o he euse needs o be a gued in e ms o
he o e all assu ance objec i es indica ed by a s anda d: i.e. wha needs o be demons a ed o assu ance and
compliance in he a ge con ex .
Equi alence maps a e also necessa y be ween he baseline o an assu ance p ojec and he e e ence amewo k
(o amewo ks) o he s anda d acco ding o which a sys em has o be assu ed. Some di e ences migh exis as
esul o e.g. ha ing o ailo how o ollow a s anda d acco ding o he speci ic cha ac e is ics o a sys em.
3) Compliance mapping
Ano he necessa y ype o maps o c oss-s anda d and c oss-domain euse is compliance maps. These mappings
speci y how he in o ma ion o an assu ance p ojec (i.e., i s body o assu ance asse s) complies wi h i s baseline.
As equi alence maps, compliance maps can be ull, pa ial, o no map. By mapping an a e ac o a e e ence
a e ac selec ed o a baseline, he in en o he a e ac is indica ed.
The compliance maps o he sou ce assu ance p ojec will ypically be ull and 1:1. I s baseline will co espond
o a empla e acco ding o which he p ojec is execu ed. Fo example, a baseline om a e e ence amewo k o
DO-178C will ha e So wa e Requi emen s Da a as an a e ac o p o ide, and an assu ance p ojec can ha e a
single a e ac ha maps o So wa e Requi emen s Da a. None heless, an assu ance p ojec can also manage,
s uc u e, o g oup i s a e ac s in a di e en way o wha a s anda d indica es, bu s ill being complian . Fo
example, an assu ance p ojec could ha e mo e han one a e ac o i s So wa e Requi emen s Da a, such as
high-le el equi emen s speci ica ion and low-le el equi emen s speci ica ion. Each o hese a e ac s would
pa ially map o So wa e Requi emen s Da a.
Compliance maps o he a ge assu ance p ojec can be de i ed om he compliance maps o he sou ce
p ojec . In his case, he likelihood o de i ed ull maps is low because o he di e ences ha usually exis
be ween s anda ds. The assu ance in o ma ion o he sou ce p ojec ul ils he equi emen s o i s baseline and i
u n o some sou ce e e ence amewo k. The a ge assu ance p ojec will ha e a di e en baseline and
e e ence amewo k, hus di e en equi emen s o ul il. None heless, some e e ence equi emen s can be
simila o equal. Fo example, an a e ac ha complies wi h EN 50128 So wa e Requi emen s Speci ica ion
will pa ially map o DO-178C So wa e Requi emen s Da a.
16
co e age. The es ima es o euse om AP1 o AP2 wi h he euse app oach is based on AP1 wi h he euse
app oach, he co e age o AP1 wi hou he app oach, and on euse om AP1 o AP2 wi hou he app oach.
Based on he es ima es by he p ac i ione s, c oss-domain euse (i.e., Reuse om AP1 o AP2) wi h he cu en
p ac ices esul s in only 4.5% o e o and cos educ ion. The applica ion o he euse app oach could lead o
e o educ ion by 53.8% and cos educ ion by 26.6%,
Table. 2. Me ic measu emen
Me ic Value Commen s
Ra io o assu ance asse s ha a e eused 0,74 Focused on ( e e ence) baseline equi emen s
Reused assu ance asse s 155 Reused compliance equi emen s
To al assu ance asse s 210 To al se o a ionics baseline equi emen s
Ra io o baseline elemen s ha do no need a
new compliance map
0,19 Focused on ( e e ence) baseline equi emen s
Baseline elemen s ha do no need a new
compliance map 40 Numbe o asse s eused whose compliance map is ull
To al baseline elemen s 210 To al se o e e ence equi emen s
Ra io o e e ence assu able elemen s wi h
applicable equi alence maps
0,67 Focused on e e ence equi emen s
Re e ence assu able elemen s wi h applicable
equi alence maps 210 Numbe o e e ence equi emen s wi h equi alence
maps
To al e e ence assu able elemen s 315 To al se o e e ence equi emen s
Ra io o compliance maps au oma ically
c ea ed
0,74 Focused on au oma ed compliance maps o baseline
equi emen s
Compliance maps au oma ically c ea ed 155
To al compliance maps 210
Ra io o e e ence assu able elemen s wi h
some equi alence map
0,67 Focused on e e ence equi emen s
Re e ence assu able elemen s wi h equi alence
maps 210 Numbe o e e ence equi emen s wi h equi alence
maps
To al e e ence assu able elemen s 315 To al se o a ionics e e ence equi emen s
Ra io o assu ance asse euse ha a e eused
ac oss domains
0,74 Focused on ( e e ence) baseline equi emen s
Assu ance asse s eused ac oss domains 155 Reused compliance equi emen s
To al assu ance asse s 210 To al se o a ionics baseline equi emen s
Ra io o baseline elemen s whose compliance
wi h has o be shown
0,81 Focused on ( e e ence) baseline equi emen s
Baseline elemen s ha need a new compliance
map on he new domain 170 Numbe o asse s ha need a compliance map, ull o
pa ial, in he new domain
To al baseline elemen s 210 To al se o a ionics compliance equi emen s

17
Table. 3. E o and cos es ima ion wi hou he euse app oach
Table. 4. E o and cos es ima ion wi h he euse app oach
4.4 Discussion
We discuss in his sec ion he answe s o he RQs, he p ac ical conside a ions o applying he euse app oach,
and he alidi y o he esul s ob ained and he conclusions d awn.
We conside ha he answe o RQ1 (Can he app oach be e ec i ely applied o euse o ce i ica ion
a e ac s?) is posi i e. The esul s o he case s udy, which we e alida ed by p ac i ione s, show ha he euse
app oach was success ully applied o analysing he euse o an execu ion pla o m om ailway o a ionics. All
he ac i i ies o he p ocess o enac he app oach (Sec ion 3.3) could be execu ed, and in acco dance o he euse
p inciples p esen ed (Sec ion 3.2). The whole se o a e ac s om ailway could be eused and he euse
consequences could be de e mined.
When analysing he impac o applying he p oposed app oach (RQ2), he esul s s ongly sugges ha he euse
app oach can educe o ecu ing cos s o p oduc sa e y ce i ica ion ac oss s anda ds and domains. Fi s ,
me ic measu emen shows gains abo e 65%, wi h almos a 20% o baseline elemen s ha do no new
compliance maps (i.e., al eady complian elemen s a e euse). Second, p ac i ione s conside ha he
applica ion o he euse app oach can lead o e o educ ion by abo e 50% and o cos educ ion by abo e 25%.
E en hough he es ima es we e oo op imis ic, we conside ha he es ima es p o ide e idence o he po en ial
e o and cos educ ions ha he euse app oach can enable.
The e a e aspec s ela ed o he alidi y o he case s udy ha a e inhe en o his esea ch me hod, such as he
applica ion o he app oach in a single case and in a gi en con ex ( euse o an execu ion pla o m om ailway
o a ionics). This a ec s ex e nal alidi y. O he aspec s o which a eade mus be awa e a e as ollows:
● Fully accu a e esul s can only be ob ained i he euse app oach is applied a eal ull p ojec . The
cu en alida ion has been he ini ial s ep owa ds demons a ing he po en ial o he app oach in
p ac ice.
● Some esul s a e based on es ima es. This has been mi iga ed by aking measu es o p o ide sound
es ima es (e.g., in ol emen o se e al p ac i ione s).
● The case s udy has ocused on he euse be ween wo so wa e s anda ds. The esul s migh hus di e
o e.g. sys em-le el s anda ds. The same applies o euse si ua ions in which goal-based s anda ds a e
in ol ed and assu ance p ojec s ha ha e o show compliance wi h se e al s anda ds.
None heless, we a e con iden in he o e all alidi y o he case s udy, hus o he euse app oach. Fi s ,
p ac i ione s ha e been s ongly in ol ed h oughou he case s udy. They also p o ided egula eedback du ing
OPENCOSS on how he euse app oach should be o i indus y needs. Second, mos p ac i ione s in ol ed in
sa e y-c i ical sys em enginee ing and ce i ica ion deal wi h DO-178, EN 50128, o simila s anda ds (e.g.,
18
ARP4754 o IEC 61508-based s anda ds, espec i ely), acco ding o he backg ound in o ma ion o ecen la ge
su eys on sa e y e idence managemen [13,14].
5 Conclusion
P oduc euse is a common ac i i y in he de elopmen o sa e y-c i ical sys ems. I can imp o e sa e y-c i ical
sys em enginee ing and ce i ica ion, and he e is guidance o p oduc euse ac oss sys ems o a same domain.
Howe e , euse ac oss s anda ds and domains has ecei ed li le a en ion and no ecommenda ions exis o
such euse scena ios Al hough simila i ies migh exis be ween domain, eusing sa e y ce i ica ion a e ac s
om one domain o ano he is no a s aigh o wa d because each domain has i s enginee ing p ac ices and
s anda ds.
This pape has p esen ed a sys ema ic app oach o e ec i ely euse sa e y ce i ica ion a e ac s ac oss s anda ds
and domains. The euse app oach is based on he mapping o he sa e y c i e ia o wo s anda ds and he
mapping o he a e ac s o an assu ance p ojec o he s anda d wi h which he p ojec has o comply. The
esul ing chain o maps can be la e used o iden i y he consequences o eusing sa e y ce i ica ion a e ac s
om a sou ce assu ance p ojec o a a ge p ojec in ano he domain and wi h di e en applicable s anda ds.
The euse app oach has been alida ed in a case s udy on euse om ailway o a ionics. In collabo a ion wi h
p ac i ione s, we we e able o apply he app oach o euse assu ance in o ma ion complian wi h EN 50128 in a
DO-178 p ojec . The applica ion esul ed in he euse o all he ailway a e ac s, ull compliance demons a ion
o almos 20% o he elemen s o he a ge assu ance p ojec , and almos a 75% o compliance needs co e age.
P ac i ione s u he es ima ed ha he use o he euse app oach could lead o e o educ ion by abo e 50%
and cos educ ion by abo e 25%. The e o e, we a gue ha he app oach can be e ec i ely applied and ha i
can educe he cos o sa e y ce i ica ion ac oss s anda ds and domains. We u he conclude ha i is a
p omising way o making c oss-domain euse mo e cos -e ec i e in indus y.
As u u e wo k, seman ically en iching he maps, u he aking e minological aspec s in o accoun , and
inc eased, mo e au oma ed ool assis ance a e aspec s om which he euse app oach migh bene i . We a e also
in e es ed in conduc ing a case s udy wi h some goal-based s anda d and a di e en one o gain insigh s in o
possible u he needs o he euse app oach, such as hose ha migh a ise om ha ing o p o ide an assu ance
case.
Acknowledgemen . The esea ch leading o his pape has ecei ed unding om he FP7 p og amme unde
g an ag eemen n° 289011 (OPENCOSS). The au ho s also hank he OPENCOSS pa ne s who p o ided inpu
o and eedback on he app oach p esen ed in he pape and i s alida ion, especially F anck Aimé, Ka ina
A wood, Céd ic Che el, Tim Kelly, and Cy il Ma chand.
Re e ences
1. Knigh JC. Sa e y c i ical sys ems: challenges and di ec ions. ICSE 2002, pp. 547–550.
2. Rushby J. Jus -in- ime ce i ica ion. ICECCS 2007, pp. 15–24.
3. RTCA DO-178C/EUROCAE ED-12c. So wa e Conside a ions in Ai bo ne Sys em and Equipmen
Ce i ica ion, 2011
4. IEC 61508. Func ional sa e y o elec ical/elec onic/p og ammable elec onic sa e y ela ed sys ems, 2011.
5. UK Minis y o De ence. In e im De ence S anda d 00-56,Issue 3: Sa e y Managemen Requi emen s o
De ence Sys ems. Pa 2: Guidance on Es ablishing a Means o Complying wi h Pa 1, 2004.
6. CENELEC. EN 50128 - Railway applica ions — Communica ion, signalling and p ocessing sys ems —
So wa e o ailway con ol and p o ec ion sys ems, 2011.
7. ISO 26262. Road ehicles — Func ional sa e y, 2011.
8. Boeing. Ce i ying Boeing's Ai planes. Online, h p://787upda es.newai plane.com/Ce i ica ion-P ocess
(accessed 24-Jan-16)
9. Gill NS. Reusabili y issues in componen -based de elopmen . ACM SIGSOFT So wa e Enginee ing No es,
28(4): 4–4, 2003.
10. RTCA DO-297/EUROCAE ED-124 In eg a ed Modula A ionics (IMA) De elopmen Guidance and
Ce i ica ion Conside a ions, 2005.
11. Somme ille I. So wa e Enginee ing (10 h ed.). Pea son, 2015.
12. Nai S, de la Va a JL, Sabe zadeh M, B iand L. An Ex ended Sys ema ic Li e a u e Re iew on P o ision o
E idence o Sa e y Ce i ica ion. In o ma ion and So wa e Technology 56(7): 689-717, 2014
13. Nai S, de la Va a JL, Sabe zadeh M, Falessi D. E idence Managemen o Compliance o C i ical Sys ems
wi h Sa e y S anda ds: A Su ey on he S a e o P ac ice. In o ma ion and So wa e Technology 60: 1-15,
2015
14. de la Va a JL, Bo g M, Wnuk K, Moonen L. Su ey on Sa e y E idence Change Impac Analysis in
P ac ice: De ailed Desc ip ion and Analysis. Simula Resea ch Labo a o y, Technical Repo 2014-18, 2014.
19
15. Ma in H, Baumga S, Lei ne A, Wa zenig D. Challenges o euse in a sa e y-c i ical con ex : A s a e-o -
p ac ice s udy. SAE Technical Pape 2014-01-0218, 2014
16. Mach ouh J, Blanqua JP, Bau e on P, Boulange JL, Delseny H, Gassino J, Ladie G, Ledino E, Leeman
M, As uc JM. C oss domain compa ison o Sys em Assu ance. ERTS-2012, pp. 1–3.
17. Ruiz A, La ucea X, Espinoza H, Aime F, Ma chand C. An Indus ial Expe ience in C oss Domain
Assu ance P ojec s. Eu oSPI 2015, pp. 29–38.
18. Che el C. A ionics Sys em Ce i ica ion, Ce i ica ion Toge he Con e ence, 2011.
19. FAA Ad iso y Ci cula : AC 20-148 Reusable So wa e Componen s, 2004.
20. E eleens RLC. RTO-EN-SCI-176 In eg a ed Modula A ionics De elopmen Guidance and Ce i ica ion
Conside a ions, 2006.
21. CENELEC. EN 50126 - Railway applica ions — The speci ica ion and demons a ion o eliabili y,
a ailabili y, main ainabili y and sa e y (RAMS), 1999.
22. CENELEC. EN 50129 - Railway applica ions — Communica ion, signalling and p ocessing sys ems —
Sa e y ela ed elec onic sys ems o signalling, 2003.
23. CENELEC. TR 50506-2: Railway applica ions - Communica ion, signalling and p ocessing sys ems -
Applica ion Guide o EN 50129, Pa 2: Sa e y Assu ance, 2009.
24. Lap ie JC. Dependabili y: Basic Concep s and Te minology. Sp inge , 1992.
25. Li J, Zhang HC, Lin Z. Asymme ic nego ia ion based collabo a i e p oduc design o componen euse in
dispa a e p oduc s. Compu . Ind. Eng. 57(1): 80–90, 2009.
26. K ohn CA. Space Shu le Componen Reuse S udy. IEEE T ans. Reliab. R-25(4): pp. 234–238, 1976.
27. Jha M, O’B ien L. A compa ison o so wa e euse in so wa e de elopmen communi ies. MySEC 2011,
pp. 313–318.
28. Salinesi C, Mazo R, Djebbi O, Diaz D, Lo a-Michiels A. Cons ain s: The co e o p oduc line enginee ing.
RCIS 2011, pp. 1–10.
29. Ha n H, Be zins V, Luqi. So wa e e olu ion ia eusable a chi ec u e. ECBS 1999, pp. 11–17.
30. Obbink H, Pohl K (Eds.). So wa e P oduc Lines. Sp inge , 2005.
31. an de Linden F (Ed.). So wa e A chi ec u es o P oduc Families. Sp inge , 2000.
32. Mangun D, Thu s on DL. Inco po a ing componen euse, emanu ac u e, and ecycle in o p oduc po olio
design. IEEE T ans. Eng. Manag. 49(4): 479–490, 2002.
33. Amelia L, Wahab DA, Che Ha on CH, Muhamad N, Azha i CH. Ini ia ing au omo i e componen euse in
Malaysia. J. Clean. P od. 17(17): 1572–1579, 2009.
34. Go TF, Wahab DA, Rahman MNA, Ramli R, Hussain A. Gene ically op imised disassembly sequence o
au omo i e componen euse. Expe Sys . Appl. 39(5): 5409–5417, 2012.
35. Zelle M, Hö ig K, Ro h elde M. Towa ds a C oss-Domain So wa e Sa e y Assu ance P ocess o
Embedded Sys ems. SASSUR 2014, pp. 396–400.
36. Papadopoulos Y, McDe mid JA. The po en ial o a gene ic app oach o ce i ica ion o sa e y c i ical
sys ems in he anspo a ion sec o . Reliab. Eng. Sys . Sa . 63(1):47-66, 1999.
37. pSAFECER p ojec . Deli e able “De ini ion o c oss-domain use case desc ip ion, use case-speci ic
equi emen s and assessmen c i e ia”, 2012.
38. Gallina B, Kashiya andi S, Zugsb a l K, Ge en A. Enabling C oss-Domain Reuse o Tool Quali ica ion
Ce i ica ion A e ac s. SAFECOMP Wo kshops 2014, pp. 255-266
39. Gallina B, Sza má i Z. On ology-Based Iden i ica ion o Commonali ies and Va iabili ies Among Sa e y
P ocesses. PROFES 2015, pp. 182-189
40.
Gallina B. A Model-D i en Sa e y Ce i ica ion Me hod o P ocess Compliance. ISSREW 2014, pp. 204–
209.
41.
Rod iguez-Dapena P. So wa e sa e y ce i ica ion: a mul idomain p oblem. IEEE So w. 16(4): 31–38,
1999.
42. Zeng F, Lu M, Zhong D. So wa e Sa e y Ce i ica ion F amewo k Based on Sa e y Case. CSSS 2012, pp.
566–569.
43. OPENCOSS p ojec . OPENCOSS Pla o m - Final P o o ype Use Manual. Online, h p://www.opencoss-
p ojec .eu/node/7, 2014 (accessed 24-Jan-16)
44. OPENCOSS p ojec . Deli e able D4.4 - Common Ce i ica ion Language: Concep ual Model, e sion 1.4.
Online, h p://www.opencoss-p ojec .eu/node/7, 2015 (accessed 24-Jan-16)
45. de la Va a JL, Ruiz A, A wood K, Espinoza H, Panesa -Walawege RK, Lopez A, del Rio I, Kelly T.
Model-Based Speci ica ion o Sa e y Compliance Needs: A Holis ic Gene ic Me amodel. In o ma ion and
So wa e Technology 72: 16-30, 2016.
46. OPENCOSS p ojec . Deli e able D5.2 - De ailed equi emen s o he OPENCOSS composi ional
ce i ica ion app oach. Online, h p://www.opencoss-p ojec .eu/node/7, 2012 (accessed 24-Jan-16)
47. OPENCOSS p ojec . Deli e able D1.2 - Use cases desc ip ion and business impac . Online,
h p://www.opencoss-p ojec .eu/node/7, 2012 (accessed 24-Jan-16)
20
48. OPENCOSS p ojec . Deli e able D1.4 - Implemen a ion o use cases on op o OPENCOSS pla o m.
Online, h p://www.opencoss-p ojec .eu/node/7, 2015 (accessed 24-Jan-16)
49. OPENCOSS p ojec . Deli e able D1.3 - E alua ion amewo k and quali y me ics, e sion 1.2. Online,
h p://www.opencoss-p ojec .eu/node/7, 2013 (accessed 24-Jan-16)
50. OPENCOSS p ojec . Deli e able D1.5 - OPENCOSS Benchma king. Online, h p://www.opencoss-
p ojec .eu/node/7, 2015 (accessed 24-Jan-16)