Secu e- aul - ole an e icien indus ial in e ne o heal hca e hings
amewo k based on digi al win ede a ed og-cloud ne wo ks
Abdullah Lakhan
a, ,g
, Ali Azawii Abdul La ee
b,i
, Mohd Khanapi Abd Ghani
c
, Ka a Hameed Abdulka eem
d
,
Mazin Abed Mohammed
e, ,g,
⇑
, Jan Nedoma
, Radek Ma inek
g
, Begoña Ga cia-Zapi ain
h
a
Depa men o Compu e Science and Cybe secu i y, Dawood Uni e si y o Enginee ing and Technology, Ka achi, Pakis an
b
Human Resou ces Depa men , Uni e si y Headqua e , Uni e si y o Anba , Ramadi 31001, Anba , I aq
c
Biomedical Compu ing and Enginee ing Technologies (BIOCORE) Applied Resea ch G oup, Facul y o In o ma ion and Communica ion Technology, Uni e si i Teknikal Malaysia
Melaka, Du ian Tunggal 76100, Malaysia
d
College o Ag icul u e, Al-Mu hanna Uni e si y, Samawah 66001, I aq
e
College o Compu e Science and In o ma ion Technology, Uni e si y o Anba , Ramadi 31001, Anba , I aq
Depa men o Telecommunica ions, VSB-Technical Uni e si y o Os a a, Os a a, Czech Republic
g
Depa men o Cybe ne ics and Biomedical Enginee ing, VSB-Technical Uni e si y o Os a a, Os a a, Czech Republic
h
eVIDA Lab, Uni e si y o Deus o, 48007 Bilbao, Spain
i
Dep amen o adminis a i e and inancial a ai s, Uni e si y Headqua e , Uni e si y o Anba , Ramadi, 31001, Anba , I aq
a icle in o
A icle his o y:
Recei ed 29 Ma ch 2023
Re ised 4 Sep embe 2023
Accep ed 6 Sep embe 2023
A ailable online 14 Sep embe 2023
Keywo ds:
IoHT
Faul - ole an
Digi al win
Indus y 5.0
Blockchain
SFTS
Fog-cloud ne wo ks
CNN
abs ac
The Indus ial In e ne o Heal hca e Things (IIoHT) is he eme ging pa adigm in digi al heal hca e.
Con ex -awa e heal hca e senso s, local in elligen wa ches, heal hca e de ices, wi eless communica ion
echnologies, og, and cloud compu ing a e all pa s o he IIoHT used in heal hca e. The ubiqui ous
heal hca e se ices i p o ides o i s use s in p ac ice. Howe e , he cu en IIoHT heal hca e amewo ks
ha e secu i y and ailu e issues in mobile og and cloud ne wo ks whe e hey a e sp ead ou . This pape
p esen s he secu e, aul - ole an IIoHT F amewo k based on digi al win (DT) ede a ed lea ning-
enabled og-cloud models. The DT is an e ec i e echnology ha makes i ual copies o se e s a di -
e en loca ions. DT in eg a ed wi h ede a ed lea ning inside he og and cloud en i onmen s, whe e
he ailu e o asks and execu ion imp o ed o heal hca e senso da a. The s udy aims o educe p ocess-
ing ime and he isk o ask ailu e. The s udy p esen s he Secu e and Faul -Tole an S a egies (SFTS)-
enabled IIoHT amewo k ha op imizes wea able senso da a and execu es i wi h he minimum
o loading and p ocessing delays. Simula ion esul s show ha he p oposed wo k minimized he secu i y
isk by 40%, ailu e isk o asks isk by 50%, and he aining and es ing ime by 39% o senso da a du -
ing he execu ion o mobile og cloud ne wo ks.
Ó2023 The Au ho (s). Published by Else ie B.V. on behal o King Saud Uni e si y. This is an open access
a icle unde he CC BY license (h p://c ea i ecommons.o g/licenses/by/4.0/).
1. In oduc ion
The indus ial In e ne o Things (IIoT) is a e olu iona y pa a-
digm o imp o e digi al p oduc i i y in di e en businesses (e.g.,
heal hca e, manu ac u ing, and anspo ) (L , 2023). I s a ed
wi h Indus y 1.0, which op imized he s eam o engine machines.
Indus y 2.0 has a lo o op imiza ion in dis ibu ed elec ici y in
sma ci ies (Rashid e al., 2022). Indus y 3.0 de e mined he op i-
mal in o ma ion abou ages wi h dep h ans o ma ion in o esul s.
Indus y 4.0 is abou au oma ion and in elligence; many s a ic and
dynamic sys ems ha e been ans o med in o au oma ion. Ini i-
a ed by he Eu opean Union in 2021, he e olu iona y pa adigm
is known as Indus ial 5.0. The objec i e is o connec wea able
and indus ial de ices wi h dis ibu ed a i icial in elligence-
based se ices (Leng e al., 2022). Cloud compu ing is a emo e se -
h ps://doi.o g/10.1016/j.jksuci.2023.101747
1319-1578/Ó2023 The Au ho (s). Published by Else ie B.V. on behal o King Saud Uni e si y.
This is an open access a icle unde he CC BY license (h p://c ea i ecommons.o g/licenses/by/4.0/).
⇑
Co esponding au ho a : College o Compu e Science and In o ma ion
Technology, Uni e si y o Anba , Ramadi 31001, Anba , I aq.
E-mail add esses: [email p o ec ed] (A. Lakhan), aliazawii@uoanba .
edu.iq (A.A. Abdul La ee ), [email p o ec ed] (M.K. Abd Ghani), Khak9784@mu.
edu.iq (K.H. Abdulka eem), [email p o ec ed] (M.A. Mohammed),
[email p o ec ed] (J. Nedoma), [email p o ec ed] (R. Ma inek), mbga cia-
[email p o ec ed] (B. Ga cia-Zapi ain).
Pee e iew unde esponsibili y o King Saud Uni e si y.
P oduc ion and hos ing by Else ie
Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
Con en s lis s a ailable a ScienceDi ec
Jou nal o King Saud Uni e si y –
Compu e and In o ma ion Sciences
jou nal homepage: www.sciencedi ec .com
ice p o ide co po a ion whe e di e en p o ide s o e se ices
o he 5.0 indus y-enabled indus ies o execu ion (Leong e al.,
2021).
Many echnological de elopmen s in heal hca e sec o s based
on 5.0 wi h cloud compu ing ha e ecen ly been seen and imple-
men ed in di e en clinics. Cloud compu ing o e s i ual se ices
o he 5.0 indus ial In e ne o Heal hca e Things (IIoHT), he e -
sion o IIoT whe e emo e heal hca e moni o ing imp o es he
quali y o li e (Chi e al., 2022). Mode n A i icial In elligence (AI)
and machine lea ning algo i hms ha e a lo o impac in Indus y
5.0 o make he heal hca e applica ion (e.g., IoT Co id-19) wi h ull
au oma ion h ough di e en aining and es ing phases
(Konigsbu g, 2022; Salman and Geman, 2023). IIoHT is a collec ion
o heal hca e senso s equipped wi h human bodies ha moni o
a es in eal- ime. Cloud compu ing ou sou ces heal hca e se ices
a di e en laye s. Fo ins ance, og compu ing is he cloud pa a-
digm ha b ings cloud se ices o he heal hca e adio ne wo k
laye wi h minimum end- o-end la ency (Khoso e al., 2021). The
eal- ime moni o ing and huge amoun o da a gene a ed by sen-
so s wi h inc eased use s lead o challenges in he IIoHT sys em
o p ocessing. Fu he mo e, many challenges exis in indus ial
5.0-awa e IIoHT o heal hca e indus ies, such as esou ce con-
s ain s, ailu e o nodes, and secu i y, along wi h he quali y o
se ice equi emen s o applica ions (Younan e al., 2020).
The digi al win (DT) is an eme ging echnology o e ing di e -
en physical se e eplicas h ough i ualiza ion (Khan e al.,
2022; Elayan e al., 2021; Ghi a e al., 2020). DT echnology has
made many con ibu ions o indus y-5.0-enabled heal hca e ech-
nologies. Fo ins ance, DT o e s componen s, in as uc u e, and
esou ce eplicas o he same and di e en nodes. The goal is o
handle as heal hca e senso eques s on o he heal hca e se e s
(Volko e al., 2021; Haleem e al., 2023). The DT educes he
esou ce cons ain s issue o heal hca e senso s and se e s wi h
he i ual eplica a a ious loca ions. Howe e , secu i y is a c i -
ical issue in DT echnologies. Blockchain is decen alized, whe e all
au onomous physical and i ual en i ies can ans e da a wi h
alidi y and anspa ency (Azzaoui e al., 2021; Jimenez e al.,
2020; Akash and Fe dous, 2022; el Azzaoui e al., 2020; Zhang
e al., 2020). Blockchain schemes can alida e da a ansac ions
among physical and i ual se e s du ing hei execu ion wi hou
showing abs ac ion o he use s. Howe e , blockchain echnology
equi es a conside able amoun o esou ces o ansac ions.
The e o e, ansac ion and node ailu es a e common in
blockchain-based DT o heal hca e applica ions. The aul -
ole an , e icien DT echnologies p esen ed by hese s udies
(Nguyen e al., 2022; Da ishi e al., 2021; Alsha h i e al., 2023).
Howe e , hese echniques did no conside secu i y aspec s.
The e o e, he og cloud ne wo k has no digi al win-enabled IIoHT
sys em.(i) Due o a dis ibu ed sys em, he ailu e o esou ces
leads o he ailu e o heal hca e asks. I is a necessa y p ocess.
The e o e, aul - ole an esou ces mus be pa o he IoT heal h-
ca e sys em o p ocess he senso y da a wi hou ailu e. (ii) Many
hings can go aw y wi h esou ce ailu e nodes, such as missing
deadlines, c i ical asks no doing hei jobs, and no ge ing he
bes esul s needed. (iii) The senso y da a is he challenging com-
ponen o he IoT-enabled heal hca e sys em, whe e p ocessing
cen alized se e nodes on big senso y da a akes much ime o
execu ion. (i ) Resou ce-e icien secu i y is mos impo an when
og nodes ha e limi ed esou ces a he adio ne wo k o p ocess-
ing wi h minimum end- o-end delays.
In his pape , we a e conside ing he ollowing esea ch ques-
ions: (i) Due o a dis ibu ed sys em, esou ce ailu e in hospi al
nodes leads o he comple e ailu e o c i ical heal hca e asks.
Exis ing DT ailu e s a egies (Nguyen e al., 2022; Da ishi e al.,
2021; Alsha h i e al., 2023) only ocused on eplica ailu e. How-
e e , i is no bene icial o la ge sys ems like dis ibu ed heal h-
ca e wi h many senso s. (ii) Exis ing DT-enabled heal hca e
sys ems did no combine secu i y and aul ole ance. The e o e,
he e mus be a balance be ween secu i y and ailu e o asks
and nodes in he DT-enabled heal hca e sys em.
This pape p esen s he secu e, aul - ole an Indus ial In e ne
o Heal hca e Things (IIoHT) sys em based on digi al win ede -
a ed og-cloud models. The s udy aims o educe he ime needed
o p ocess heal hca e senso da a o secu i y, ask execu ion, and
aul ole ance while using less esou ces. The pape has he ol-
lowing con ibu ions o he esea ch ques ions, including ede a ed
lea ning and digi al win echnology.
The s udy in eg a ed he og and cloud nodes based on digi al
win echnology, whe e all local nodes a di e en labo a o ies
ha e eplica copies o ained da a and p ocessing capabili y in
he same un ime en i onmen .
We in eg a ed he di e en kinds o heal hca e senso s in he
human body. We connec ed hem wi h mobile de ices, such
as ECG lead-1 and lead-2 senso s, w is wa ches ( empe a u e
and jogging senso s), and ankle magne ome e senso s. Each
senso can gene a e eal- ime da a and o load i o he p oxim-
i y labo a o ies o p ocessing.
The p oposed SFTS is mo e e icien ega ding secu i y, aul -
ole an , and esou ce scalabili y.
We in eg a ed he ede a ed lea ning scheme, whe e aining
and es ing a e de e mined based on a con olu ional neu al ne -
wo k (CNN). The agg ega ed node execu es all asks based on
hei gi en cons ain s.
The pape consis s o he ollowing pa s. The goals o he p e-
ious s udies o IoT heal hca e in og cloud models we e discussed
in he ela ed wo k. The p oblem a chi ec u e shows all compo-
nen s o he a chi ec u e. The p oposed algo i hm pa shows
how o sol e he p oblem in di e en s eps. The expe imen al pa
shows he simula ion con igu a ion and simula ion esul s. In con-
clusion, he esul s and u u e di ec ion o he wo k we e looked a
in ligh o he new limi s.
2. Rela ed wo k
Digi al win echnology in IIoHT sys ems has achie ed many
achie emen s in he heal hca e domain. Di e en heal hca e appli-
ca ions, such as disease p edic ion, secu e da a o loading, mobile
medicine, and IoT heal hca e, a e widely in eg a ed wi h DT ech-
nology. Fu he s udies sol ed he di e en heal hca e issues wi h
addi ional cons ain s, as shown in Table 1. These s udies (L ,
2023; Leng e al., 2022; Leong e al., 2021; Chi e al., 2022; Khoso
e al., 2021; Younan e al., 2020) discussed Indus y 5.0, a new
in o ma ion echnology e olu ion ha ans o ms adi ional
heal hca e applica ions, a chi ec u es, and sys ems in o digi al
and au oma ed o ms. In Indus y 5.0, many eme ging echnolo-
gies, such as DT, edge compu ing, machine lea ning, cloud compu -
ing, and blockchain echnologies, a e in eg a ed wi h heal hca e o
make i mo e obus and e icien . Howe e , hese s udies only dis-
cussed DT’s in o ma ion low and ad an ages wi h Indus y 5.0 o
heal hca e. The e o e, me hods and sys ems a e o be de eloped
based on he gi en p o o ypes in hese s udies.
This s udy (Khan e al., 2022) sugges ed DT-enabled og cloud
solu ions o di e en indus y applica ions. Fo ins ance, heal h-
ca e in in elligen ci ies, pha maceu ical medicine supply chains,
e c. In de ail, his pape discussed machine lea ning, edge compu -
ing, and IoT heal hca e senso -enabled DT indus y a chi ec u es
and esou ce eplicas. Di e en o loading and esou ce alloca ion
(RA) s a egies based on machine lea ning o edge and cloud com-
pu ing a e lis ed wi h hei cons ain s. Howe e , his wo k is mo e
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
2
gene al. So a , secu i y, aul ole ance, and o he issues will be
sol ed in he discussed solu ions.
A heal hca e DT-enabled con ex -awa e IoT og cloud solu ion
has been sugges ed by his s udy (Elayan e al., 2021). Fo p edic-
ion, he elec oca diog am (ECG) non-in asi e echnology da a
was o loaded o p oximi y clinical se e s. The heal hca e clinics
a e in eg a ed wi h he digi al wins, whe e homogeneous nodes
can sha e hei da a asse s and i ual se e copies. Howe e , his
wo k only ocused on limi ed IoT da a se ices wi h ixed nodes o
heal hca e con ex s. To imp o e he e iciency o IoT heal hca e
con ex -awa e DT, a dis ibu ed in elligen geospa ial sys em
based on cloud se ices is sugges ed in Ghi a e al. (2020). This sys-
em o e ed dis ibu ed con ex -awa e IoT heal hca e se ices
based on cloud compu ing. The e is no issue o esou ce scalabili y
in he wo k. Howe e , due o he many use s o his echnology,
he s o age and p ocessing cos s become highe o he se ice p o-
ide s. Secu i y sca ci y is also a challenging ask o his
echnology.
Howe e , p io s udies ocused on he con ex o IoT heal hca e
se ices, whe e da a is o loaded based on non-in asi e ECG sen-
so s o cloud compu ing. Fo he o loaded da a, he mobile medi-
cine sys em based on DT is in oduced in hese s udies (Volko
e al., 2021; Haleem e al., 2023). This wo k combined di e en
pha maceu ical companies and collec ed he se e da a asse s.
Howe e , he wo k could be mo e secu e and aul - ole an . Du ing
simula ion esul s, ailu e o asks du ing o loading and scheduling
was seen. Howe e , hese s udies (Azzaoui e al., 2021; Jimenez
e al., 2020; Akash and Fe dous, 2022; el Azzaoui e al., 2020;
Zhang e al., 2020) sugges ed blockchain and secu e algo i hm-
based solu ions sol e he secu i y limi a ions o p io s udies in
DT IoT heal hca e. Public blockchain echnology (Azzaoui e al.,
2021) implemen ed wi h he DT, whe e di e en blocks can sha e
i ual da a asse s. The main ad an age is ha he ansac ional
nodes do no need o p ocess and alida e p e ious ansac ions
o a oid delays and esou ce consump ion in og cloud ne wo ks.
Howe e , public blockchain echnology wi h DT sha es da a wi h
homogeneous nodes. The e o e, i canno be used wi h he e oge-
neous nodes in IoT heal hca e domains. The IoT heal hca e cybe -
space (Jimenez e al., 2020; Akash and Fe dous, 2022; el Azzaoui
e al., 2020; Zhang e al., 2020), such as he cybe -physical sys em,
is in eg a ed wi h digi al win echnology. DT in eg a ed wi h
mobile and og cloud ne wo ks, secu ely sha ing di e en da a
ypes.
These s udies (Nguyen e al., 2022; Da ishi e al., 2021;
Alsha h i e al., 2023) sugges ed DT-enabled aul - ole an ech-
niques such as p ima y backup and checkpoin ing on og and cloud
ne wo ks o IoT heal hca e applica ions. These s udies ocused on
he compile ime ailu e o se ices, asks, esou ces, and schedul-
ing o he assigned asks o he og and cloud ne wo ks. The p i-
ma y backup is in eg a ed in o he di e en og and cloud
ne wo ks as i ual se e s, whe e ask checkpoin ing echniques
a e implemen ed. These s udies (Rieke e al., 2020; Xu e al.,
2021) ede a ed lea ning enabled solu ions o dis ibu ed heal h-
ca e sys ems. Howe e , he p oposed amewo ks only suppo
ixed nodes and incu he esou ce ailu e o nodes du ing aining
and es ing in ne wo ks.
To he bes o ou knowledge, SFTS-enabled IIoHT is he new
solu ion. The main eason is ha he exis ing secu i y mechanisms
p o ided by hese s udies (Azzaoui e al., 2021; Jimenez e al.,
2020; Akash and Fe dous, 2022; el Azzaoui e al., 2020; Zhang
e al., 2020) in digi al win-enabled og cloud a e only suppo ed
on ich esou ce nodes. The e o e, mobile de ices can no in eg a e
hose models. The exis ing digi al win enabled IIoHT conside ed
he homogeneous en i onmen o da a eplica ion and execu ion.
Howe e , in ou case, we ha e di e en og and cloud nodes.
The e o e, ede a ed lea ning-enabled secu i y and p i acy a e
he new con ibu ions o DT-enabled IIoHT in mobile og cloud
ne wo ks.
3. P oposed IIoHT amewo k
The s udy p esen s an IIoHT amewo k based on digi al win
and ede a ed og-cloud models, as Fig. 1 illus a es. As shown in
Fig. 1, we can call a chi ec u e o he p oposed amewo k. We
designed he amewo k based on wo co e echnologies: ede a ed
lea ning and DT. Fede a ed lea ning allows di e en hospi als o
ain, alida e, and secu ely sha e hei p i a e da a. The digi al
win is he backbone o he sys em. He e ogeneous copies o se -
e s o e he same se ices, like s o age, esou ces, and un ime
en i onmen s o execu ing p og ams. So, ou goal is o p ocess
heal hca e senso da a in he sho es amoun o ime while
emaining secu e and e icien . The p oposed SFTS consis ed o di -
e en schemes such as aul - ole an , secu i y, local p ocessing,
o loading con olu ional neu al ne wo ks (CNN), and agg ega ed
me hods. The heal hca e senso could be abno mal, so he sys em
could no be slow o ail du ing a pa ien ’s c i ical condi ion. The
s udy implemen ed di e en heal hca e senso s and moni o ed
hei heal hca e du ing daily ac i i ies. Wea able senso s such as
ECG (lead-I and lead-II), w is -wa ches, and ankle senso s gene a e
da a o mobile de ices. Fu he mo e, mobile de ices o load sen-
so y da a o nea by hospi als o p ocessing.
The heal hca e asks a e mobile heal hca e unc ions and mon-
i o ing and o loading he senso y da a o some pu pose. We mon-
i o ed ha each use o subjec pe o med di e en daily ac i i ies
a di e en in e als. We moni o he use s’ heal hca e based on
he gene a ed da a om senso s o he sys em. All hese senso s
a e connec ed o he hospi als ia di e en communica ion chan-
nels, such as wi eless and mobile ne wo ks. We implemen ed
wo main echnologies, digi al win and ede a ed lea ning, in he
dis ibu ed og cloud ne wo ks. All he local se e s o he hospi als
a e implemen ed a he adio ne wo k, and he cen alized cloud is
loca ed a he in as uc u e le el. We ained and alida ed o -
loaded da a models a he local hospi al ne wo ks based on
machine lea ning aining models and in eg a ed hei weigh s
Table 1
Exis ing IIoHT amewo ks based on digi al win.
S udy P oposed Indus.App. Gap Analysis
(Khan e al., 2022)2022 DT-Indus ies. IoT-Heal h Resou ce Replica
(Elayan e al., 2021; Ghi a e al., 2020)2021 DT-Con ex -Algo. IoT-Heal h Resou ce Scalabili y
(Volko e al., 2021)2021 DT-Mobile IoT Medicine Resou ce Sca ci y
(Haleem e al., 2023)2021 DT-Heal hca e IoT Heal h Fea u e and Se ice
(Azzaoui e al., 2021; Jimenez e al., 2020; Akash and Fe dous, 2022;
el Azzaoui e al., 2020; Zhang e al., 2020)2021
DT-Blockchain,Secu i y IoT Heal h PoW,Me hods
(Nguyen e al., 2022; Da ishi e al., 2021; Alsha h i e al., 2023)2020–2022 DT-Faul -De ec ion IoT Heal h Backup,Checkpoin ing
(Rieke e al., 2020; Xu e al., 2021)2020–2021 Fede a ed IoT Heal h T aining
P oposed Wo k DT-Fede a ed Cons ain s IoT Heal h Faul ,Secu i y,RA
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
3
based on ede a ed lea ning. All he local og se e s a e assumed
o be he main se e ’s digi al win. The goal is o con ol secu i y
issues, ailu e issues, and delay issues du ing he p ocessing o sen-
so y asks based on gi en cons ain s. Table 2 shows he no a ions
and abb e ia ions o di e en e ms and ma hema ical models.
We conside he Dnumbe o senso da a wi h he di e en
numbe o ea u es F. The ea u es o he da a da e 1 empe a u e,
2 ECG signal, 3 hea bea speed, 4 blood p essu e, 5 ankle
di ec ion, and o he s. The e o e, we o mula ed hem in he ollow-
ing way. The s udy conside s he Tnumbe o asks. Each ask
consis ed o da a dand ea u es F, deadline d
, p ocessing s a us
s
as g een anno a ion shown in Fig. 1 All he heal hca e asks
a e pe o med on he use o subjec de ices as well as og and
cloud nodes. I depends upon he a ailabili y o compu ing
esou ces. Be o e o loading, we conside he Mnumbe o mobile
de ices. Each mobile mhas p ocessing and compu ing capabili y
ep esen ed by
m
and
m
. The s udy conside ed he Snumbe o
heal hca e cloud se e s and DT numbe o digi al- win-enabled
og nodes implemen ed a he adio ne wo ks. All he use s o sub-
jec s can access any se e , whe he i is a cen alized ede a ed
cloud o digi al- win-enabled local og nodes o he se ices. All
he digi al- win-enabled og nodes and ede a ed cloud se e s
Fig. 1. Secu e- aul - ole an e icien indus ial in e ne o heal hca e hings amewo k based on digi al win ede a ed og-cloud ne wo ks.
Table 2
Abb e ia ions & no a ions and desc ip ion.
No a ions and Abb e ia ions Desc ip ion
IoT In e ne o Things
IIoT Indus ial In e ne o Things
IIoHT Indus ial In e ne o Heal hca e Things
IoHT In e ne o Heal hca e Things
RA Resou ce Alloca ion
DT Digi al Twin
ECG elec oca diog am
Mob-Heal hca e Mobile Heal hca e
FL Fede a ed Lea ning
DNumbe o senso da a
FNumbe o ea u es
d; Pa icula da a and ea u e
TTo al numbe o asks
MNumbe o mobile de ices
;mPa icula mobile de ice and ask
m
;
m
Mobile esou ce and speed
SNumbe o heal hca e cloud se e s
DT Digi al win numbe o og se e s
d ;sPa icula og node and cloud se e
d
;
d
og cloud esou ce and speed
s
;
s
Cloud se e esou ce and speed
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
4
a e homogeneous in execu ion un ime. Howe e , hey a e he e o-
geneous in esou ce capabili y. The e o e, each ede a ed esou ce
has
s
esou ce capabili y and
s
compu ing speed capabili y. Sim-
ila ly, all og nodes ha e
d
and
d
compu ing esou ce and speed,
espec i ely. We designed he ma hema ical model based on he
assignmen p oblem on mobile og cloud ne wo ks. We de e -
mined he local senso y p ocessing ime o asks as ollows.
L
e
¼X
M
m¼1
X
T
¼1
X
D
d¼1
d
m
þEnc yp ion:ð1Þ
Eq. (1) designed based on mobile og cloud assignmen p oblem
(Daigneaul and S -Hilai e, 2021). Eq. (1) analyzes and moni o s
senso y da a’s local p ocessing ime o speci ic asks on mobile
de ices. In his equa ion, L
e
ep esen s he execu ion ime o all
asks, whe e Lis he a iable ha s o es he execu ion ime o all
asks. Fu he mo e, he inclusion o 2Tindica es ha all asks,
om s a o end, should be execu ed a e assigning hem o hei
espec i e compu ing nodes. The a iable esigni ies he execu ion
o all asks on di e en compu ing nodes. All he asks a e
enc yp ed be o e o loading o any se e o p ocessing. The e o e,
he a iable Enc yp ion de e mines he enc yp ion and dec yp ion o
all asks among di e en compu ing nodes. The o loading ans-
mission ime o he senso y da a is calcula ed as below.
C
e
¼X
M
m¼1
X
T
¼1
X
D
d¼1
d
upload þd
download :ð2Þ
Eq. (2) de e mines communica ion o loading whe e asks a e o -
loaded om local de ices o compu ing nodes. We designed his
Eq. (2) based on communica ion o loading based on he same ne -
wo k ule (Kim, 2020). Eq. (2) de e mines he da a’s p e-de e mined
upload and download ansmission imes be o e and a e p ocess-
ing. On he o he hand, C
e
ep esen s he o loading and download-
ing o ask da a om local senso s o he compu ing se e s o
p ocessing. The a iable Cholds he communica ion ime o all
asks om local senso s o compu ing se e s du ing o loading
and downloading esul s. Fu he mo e, he inclusion o 2Tindi-
ca es ha all asks, om s a o end, o load hei da a and down-
load hei esul s om he se e s. The scheduling ime on he cloud
se e s is de e mined as ollows.
Cloud
e
¼X
S
s¼1
X
T
¼1
X
D
d¼1
d
s
þEnc yp ion ð3Þ
We designed Eq. (3) based on cloud scheduling on di e en com-
pu ing se e s (Panda e al., 2022). Eq. (3) analyzed and moni o ed
he cloud p ocessing ime o senso y da a o pa icula asks. The
a iable, e.g., Cloud
e
de e mines he execu ion ime o all asks on
cloud compu ing. The digi al- win-enabled nodes ha e he ollow-
ing p ocessing ime o all asks. Fu he mo e, he a iable o 2T
indica es ha all asks, om s a o end, a e execu ed on he cloud
se e s. All he asks mus be dec yp ed be o e s a ing any p ocess-
ing. The e o e, he a iable Enc yp ion de e mines he enc yp ion
and dec yp ion o all asks among di e en compu ing nodes. How-
e e , a e execu ion, all asks mus be enc yp ed o sha e ano he
cloud se e o s o age. We scheduled all o loaded asks on
digi al- win-enabled og cloud ne wo ks, designed based on digi al
win og cloud ules (Alaasam e al., 2020). We de e mined he p o-
cessing ime based on win digi al win og se e s in ou wo k and
de ined i in he ollowing way.
F
e
¼X
DT
d ¼1
X
S
s¼1
X
T
¼1
X
D
d¼1
d
d
þEnc yp ion:ð4Þ
Eq. (4) is designed based on digi al win og cloud ules (Alaasam
e al., 2020). Eq. (4) aims o de e mine digi al win og nodes p o-
cessing ime o o loaded senso y da a o pa icula asks. We
implemen ed he digi al win mechanism o he og nodes, whe e
F
e
shows ha all asks a e execu ed on digi al- win-enabled og
nodes. All he og nodes a e ep esen ed by d2D, and he a iable
o 2Tindica es ha all asks, om s a o end, a e execu ed on
he og nodes om he schedule . To p esen a ede a ed lea ning
app oach, we conside he di e en he e ogeneous nodes o da a
sha ing and execu ion in ou wo k. All he asks mus be dec yp ed
be o e s a ing any p ocessing. The e o e, he a iable Enc yp ion
de e mines he enc yp ion and dec yp ion o all asks among di e -
en compu ing nodes. Howe e , a e execu ion, all asks mus be
enc yp ed o sha e ano he og se e o s o age. The e o e, we
main ain he da a secu i y o asks on di e en is de e mined in
he ollowing way.
Enc yp ion ¼X
M
m¼1
X
D
d¼1
Encðd;AES;publickeyÞþDecðEnc
d;p i
a ekeyÞ:ð5Þ
Eq. (5) designed based on ad anced enc yp ion s anda d secu i y
ule (Dha angan e al., 2022). Eq. (5) de e mines he enc yp ion
and dec yp ion o all ask da a on di e en nodes. Fo example, each
node enc yp s ask da a based on a public key using he Ad anced
Enc yp ion S anda d (AES) (Dha angan e al., 2022). Then, he nodes
dec yp he ask da a using a p i a e key. We implemen ed AES-256
wi h mul iple ounds and speci ic cha ac e is ics such as ound sub-
s i u ion and column eplacemen . This equa ion, o ins ance,
Enc yp ion ¼P
D
d¼1
Encðd;AES;publickeyÞþDecðEnc d;p i a ekeyÞ,
illus a es ha all nodes mus enc yp and dec yp he da a du ing
sha ing and execu ion in he ne wo k. O e all, we de e mined he
o al p ocessing ime o asks based on minimiza ion op imiza ion
enabled on mobile og cloud ule (Lakhan e al., 2022).
min To al ¼L
e
þC
e
þCloud
e
þF
e
:ð6Þ
Eq. (6) is designed based on he mobile og cloud scheduling
ule wi h he minimiza ion objec i e wi h he cons ain s
(Lakhan e al., 2022). Eq. (6) de e mines he o al ime o all asks
on di e en nodes wi h di e en ea u es. We deno ed he o al
ime as a iable To al and de e mined he execu ion ime wi h
he uni minu es. The a iable To al is an a ay ha s o es he exe-
cu ion ime on he cloud, he communica ion ime du ing o load-
ing and downloading, and he p ocessing ime on og nodes o all
asks. We calcula e he o al ime using he equa ion
To al ¼L
e
þC
e
þCloud
e
þF
e
. The e o e, indi idual imes impac
he o al p ocessing ime in ou a chi ec u e.
4. P oposed SFTS algo i hm me hodology
The s udy p esen s he SFTS algo i hm me hodology, which
consis s o di e en sub-schemes. The p ocess low o hese sub-
schemes has di e en connec ions, as shown in Fig. 2. The SFTS
algo i hm s a s wi h all pa ame e cons ain s such as
M;DT;D;S;T. The dis inc pa ame e s ha e al eady been explained
in Table 2. The local p ocessing scheme ini ia es he heal hca e
applica ion asks on local de ices. All he senso s a e connec ed
o mobile de ices, so he senso da a is gene a ed only on local
machines. Each local se e enc yp s he da a be o e o loading
and ecei es dec yp ion esul s om he digi al- win-enabled og
and cloud se e s. O loading is a communica ion scheme ha
ans e s gene a ed da a om local mobile de ices o a ailable
og nodes o u he p ocessing. We ain he gene a ed da a on
og nodes based on da a, esou ces, compu a ion ime, and dead-
line. We apply Con olu ional Neu al Ne wo ks (CNN) o he local
aining da a, and ede a ed lea ning (Rieke e al., 2020) in eg a es
he local aining da a in o he agg ega ed node o inal esul s.
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
5
The secu i y and aul - ole an schemes a e cen alized and con-
nec all nodes, enabling hem o acknowledge each o he abou
secu i y issues in he IIoHT amewo k.
Secu i y and aul ole ance a e he key challenges in he IIoHT
heal hca e a chi ec u e. The e o e, he s udy p oposed he SFTS
amewo k ha consis s o di e en sub-schemes. SFTS amewo k
consis ed o di e en schemes as shown in Algo i hm 1.
Algo i hm 1. SFTS Algo i hm F amewo k
Di e en le els exis in Algo i hm 1 o gene a ing and p ocess-
ing senso s based on hei cons ain s. Le el 1 is he scheme whe e
senso da a a e p ocessed based on a ailable esou ces and o -
loaded o he a ailable og and cloud nodes. Le el 2 is he commu-
nica ion channel ha ansmi s he da a wi hou secu i y and
ailu e issues. Le el 3 local ede a ed lea ning aining and es ing
models. Le el 4 is he agg ega ed node p ocessing. Le el 5 handles
he aul - ole an mechanism o he s udy. We de ined he all s eps
o Algo i hm 1, whe e he low s a s om local p ocessing o og
and cloud nodes in he ollowing way.
The local p ocessing scheme pe o ms based on Eq. (1). All he
local nodes a e sou ces o da a gene a ion. To ensu e secu i y
and p i acy, all he ask da a is enc yp ed be o e being sen o
he og and cloud se e s o p ocessing, as shown in Le el 1.
O loading is he mechanism whe e he algo i hm checks i he
communica ion ne wo k and compu ing esou ces a e a ail-
able; i allows local de ices o o load da a o he se e s. This
algo i hm’s main e iciency is p o iding a seamless en i on-
men when all he ne wo k’s communica ion channels and
compu ing node esou ces a e a ailable.
In Le el 3, all he asks a e o loaded o a ailable og nodes o
p ocessing. The og nodes a e in eg a ed wi h he digi al win,
whe e esou ces, ask upda es, and execu ion s a us aining
and es ing a e sha ed wi h he cloud nodes o agg ega ion.
The main e iciency o he digi al win is i s abili y o enable
og nodes o p o ide esou ce a ailabili y a he edge laye .
These og nodes ac as eplicas o cloud compu ing se e s, ha -
ing he same un ime en i onmen and in e ope abili y wi hin
he ne wo k.
The agg ega ion akes place in he cloud compu ing nodes, as
shown in le el 4, whe e all og nodes sha e hei ask execu ion
s a us and esou ces and s o e he inal esul s o he asks. This
sha ing is done o achie e esou ce scalabili y. We de ised an
agg ega ion mechanism based on e ical ede a ed lea ning,
whe e all he og nodes sha e me ics such as esou ces, ask
s a us, ailu e anno a ions, and ained and es ed execu ion
models wi h he cloud compu ing se e s. This sha ing aims
o u he op imize and imp o e he nodes’ e iciency o all
asks.
Fig. 2. P oposed algo i hm low diag am.
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
6
The aul - ole an scheme, as shown in Le el-5, is an impo an
componen implemen ed on all nodes. The main objec i e o
aul ole ance is o minimize he isk o ask ailu es. I in ol es
aining and escheduling all ailed asks on a ailable compu -
ing esou ces wi hin he gi en ask deadline.
4.1. Local p ocessing senso y da a
All he local heal hca e senso da a a e gene a ed by he di e -
en senso s and connec ed o mobile de ices. The local mobile
de ices analyzed he secu i y based on he ollowing ules.
Algo i hm 2. Local P ocessing Senso y Da a
We p esen he local p ocessing scheme Algo i hm 2 ha p o-
cessed he local p ocess as he IoT senso heal hca e mobile da a.
We discussed he IoT heal hca e da a based on mobile de ices in
he o m o a case s udy, as shown in Fig. 3. Fo unde s anding pu -
poses, we conside only h ee ypes o senso s. Fo ins ance, ECG
senso (lead-I and lead-II), w is wa ch senso s ( empe a u e and
hea -bea ange), and jogging speed and walk coun senso ha
is in eg a ed in o he subjec o use ’s ankle. We apply he wa ele
sca e ing echnique (Jean E il and Rajeswa i, 2022) o p e-p ocess
da a acco ding o ea u es. Howe e , i depends upon he a ailable
esou ces; i he equi ed p e-p ocessing and secu i y esou ces a e
su icien , hen he mobile de ices a e p ocessed locally. O he -
wise, his p ocess will o load o he a ailable digi al win og
nodes. The o loading schemes p oposed in Algo i hm 2 a e de ined
in he ollowing way.
Ini ially, all asks a e anno a ed as local asks, e.g., @ 2T, whe e
local compu ing nodes (e.g., m2M) execu e all asks o mee
he secu i y and p i acy equi emen s o da a locally. The e o e,
he inpu s equal D;M, and T.Tis he o al numbe o asks, Dis
he senso da a, and Mis he se o local compu ing nodes o
p ocessing he senso y da a a he local machines.
In S eps 1 and 2, all he asks a e anno a ed as local asks. The
main eason is ha each ask mus be ini ially execu ed locally.
We ead he asks one by one om he ask se , as shown in
S eps 1–2.
S ep 2 e i ies ha he local de ices ha e enough o execu e
senso y da a. The e o e, ini ially, we an icipa ed esou ce
checking be o e execu ion o all asks.
In S ep 4, he algo i hm de e mines ha i he local compu ing
nodes ha e enough esou ces, i anno a es all asks as local
asks and s a s hei execu ion based on secu i y equi emen s.
In S eps 5 o 7, we ensu e ha he asks a e execu ed wi hin
hei deadlines and mee he secu i y equi emen s. The algo-
i hm allows execu ion based on Eq. (1) o ask execu ion
and secu i y based on Eq. (5).
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
7
In S eps 8 o 11 de e mines ha he local p ocessing inished
hei execu ion and is looking o o loading o execu ion based
on a ailable esou ces and wi eless communica ion.
We de e mined he a ailable esou ces based on Eq. (8) be o e
s a ing any execu ion and o loading o all asks.
The local nodes o load hei da a o he og and cloud nodes.
A e he execu ion, all nodes send back hei esul s in
enc yp ed o m, dec yp ed by he local de ice o display.
4.2. O loading da a
O loading is a p ocess ha ini ia es om mobile de ices when
hey ha e no esou ces o asks ha need u he execu ion based
on hei gi en h esholds.
Algo i hm 3. O loading Da a Scheme
Fig. 3. Local p ocessing on senso da a.
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
8
The s udy designs he o loading scheme Algo i hm 3, whe e all
da a is o loaded om local mobile de ices o he a ailable og and
cloud nodes o p ocessing. Algo i hm 3 checks in ad ance; i he
communica ion channels ha e a highe capabili y o o loading, i
will o load da a o he a ailable og node o u he p ocessing.
O he wise, i he mobile o load inds weake and insu icien
bandwid h, he mobile wai s o he ne wo k a ailabili y and
esou ce a ailabili y inside he designed ne wo k. We de ine he
s eps o Algo i hm 3 in he ollowing way.
In S eps 1–2, he algo i hm checks communica ion a ailabili y
o o loading da a o he connec ed og and cloud nodes.
In S eps 3–4, he algo i hm de e mines he communica ion ime
and a ailabili y o o loading and downloading da a om com-
pu ing nodes o he local machines.
In S eps 5–6, de e mines i he communica ion channel is es ab-
lished, hen he algo i hm allows he local p ocessing algo i hm
o o load all asks o he a ailable compu ing nodes o
execu ion.
4.3. Fede a ed lea ning enabled local aining and es ing models
The s udy designed he ede a ed lea ning and i s eplica i ual
local og nodes a di e en adio ne wo ks. In ou case, we imple-
men ed ede a ed lea ning a di e en nodes. The main agg ega ed
node is powe ul cloud compu ing, whe e digi al- win-enabled
same i ual copies a e in eg a ed a di e en heal hca e clinics
as shown in Fig. 4. Each node has weigh s W¼ w¼1;do s;:Wg
ha consis s o di e en ea u es (e.g., esou ce, deadline, aining,
and es ing da ase s). All he cloud and og nodes in ede a ed
lea ning-enabled digi al wins coope a e and communica e wi h-
ou o e head issues. Da a secu i y and p i acy a e main ained
among nodes based on enc yp ion, dec yp ion ules, and aul -
ole an echniques. We p esen he ede a ed lea ning
echnology-enabled amewo k o IoT wea able heal hca e
de ices. In ou sys em, ede a ed lea ning is he comple e sys em
based on cloud da a se e s and i s eplica og se e s based on
digi al win echnology. The ede a ed is di ided in o local aining
and es ing pa and agg ega ed decision pa s o IoT heal hca e
de ices. The s udy conside s he FLM numbe o ede a ed lea ning
aining se e s such as FLM lm d s1g. We ained bo h
eplica and agg ega ed nodes based on esou ces capabili y and
IoT da ase s and ep esen ed by Z¼ z¼1;...;Zg. In ou s udy,
we conside ed ho izon al ede a ed lea ning whe e all og and
agg ega ed sha e hei da ase s bu in di e en sample modes.
We de e mined he ede a ed lea ning mechanism based on he
ollowing equa ion.
min
w
ð ÞX
F
lm1
LMX
DT
d ¼1
X
S
s¼1
lmðwjzÞ:ð7Þ
We designed he Eq. (7) based on ede a ed lea ning ules o og
and cloud ne wo ks (Lakhan e al., 2021). In Eq. (7) wis he lea ning
weigh o ede a ed lea ning on gi en inpu 2T asks and ained
da ase z. I is he same o all da ase s on di e en nodes. We
ained and es he model based on a neu al ne wo k. We di ide
he objec i e unc ion pe o mance in o local, og, and cloud com-
ponen s as shown in Eq. (7).
Fig. 4. Fede a ed lea ning p ocessing on digi al win.
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
9
ailu e o connec ion o a ailabili y o communica ion ne wo ks
and emo e nodes. Fo ins ance, we a e o loading o he emo e
compu ing nodes such as og and cloud, incu ing 14-min delays.
All he nodes a e execu ing asks in pa allel ways. The e o e, we
need o minimize he ailu e delays o all easons o mee he
equi emen s o asks in highe wi h he deadline cons ain s.
Due o he isk o he heal hca e o humans, we se deadlines o
all asks and ensu e all asks mus be execu ed unde hei dead-
line. We deno ed he b own line as he secu i y ask alida ion in
he o m o delay on di e en compu ing nodes. We ep esen ed
he secu i y delays on di e en compu ing nodes, such as mobile
de ices, communica ions, og, and cloud nodes. The To al is he
objec i e unc ion ha combines a ious delays. Fig. 8 shows ha
all he asks, e.g., 26 asks as shown in he x-axis, incu ed eigh een
minu es (18) delays a di e en nodes as shown in he Y-axis.
These delays include enc yp ion dec yp ion and alida ion o asks
a nodes. The e o e, i is necessa y o p ocess all asks on di e en
nodes wi h ewe delays acco ding o gi en asks.
In ou scena io, we conside he andom numbe o heal hca e
asks and inc ease hei sizes in he expe imen . We implemen ed
da a p e-p ocessing s a egies, such as wa ele sca e ing and CNN,
o ea u e ex ac ion in he IIoHT amewo k. We e alua ed he
pe o mances o schemes in he expe imen . Fig. 9 analyzes he
pe o mance o asks wi h he di e en delays as a o al delay.
The a iable To al is he sum o p ocessing delays de e mined in
minu es. The o al delay becomes highe and highe du ing he
eco e y o he secu i y and ailu e isks o all asks wi h he
highe numbe o asks in di e en mobile og and cloud compu -
ing nodes. The y-axis ep esen s he delays o asks as anno a ed as
To al, and he y-axis shows he ange o asks wi h andom num-
be s, e.g., 30. We eco e he secu i y ailu e issues on di e en
compu ing nodes. Fo ins ance, mobile de ices pe o med on he
local de ices and communica ion ne wo ks iden i ied he o iginal
da a in enc yp ed o m. Howe e , i he e a e some issues in he
wi eless communica ion and equi e a eco e y ime, he o loaded
asks wai un il connec ion eco e y om secu i y ailu e. I is sim-
ila o he ailu e delay eco e y, which akes delays o all asks on
di e en compu ing nodes. We deno ed he numbe o asks om
0 o 5 on mobile de ices wi h he 6-min delay wi h he secu i y
eco e y. A he same ime, 9 min o delay wi h he ailu e o asks
du ing eco e y on mobile de ices. We o loaded asks h ough
wi eless communica ions, and all asks om 0 o 20 incu ed 10
o 12 min wi h bo h ailu e and secu i y delays and impac on o al
delay. The 0 o 30 asks a e execu ed on og and cloud nodes and 16
and 18 min and de e mined o al delays o all asks. The e o e, all
asks ha e highe delays i we manage hem e icien ly on di e en
compu ing nodes.
We implemen ed he p oposed scheme along wi h baseline
app oaches in he expe imen . Fo ins ance, SFTS, IoT Wi hou Fed-
e a ed, IoT Wi h Fede a ed, and IoT Fog Cloud schemes e alua e
he pe o mance o heal hca e asks on he e ogeneous nodes. We
e alua ed he pe o mance o all asks in e ms o o al delays as
de e mined in objec i e unc ion To al as shown in Eq. (6). In he
simula ion esul , as shown in Figu e Fig. 10, he y-axis indica es
ha he di e en numbe o asks de e mines To al ime in min-
u es o all asks. The SFTS scheme execu ed all asks on di e en
compu ing nodes wi h he To al delays in 2 min. The IoT Wi h Fed-
e a ed scheme pe o med all asks on di e en compu ing nodes
wi h he To al delays in 4 min. The IoT Fog Cloud Wi h Fede a ed
scheme execu ed all asks on di e en compu ing nodes wi h he
To al delays in 8 min. Howe e , wi hou ede a ed lea ning, i has
10 min on di e en compu ing nodes o all asks. We analyzed
he pe o mance o di e en a chi ec u es and schemes e alua ed
based on a ious asks du ing he expe imen . We analyzed and
moni o ed he pe o mance o all me hods: SFTS, IoT Wi hou Fed-
e a ed, IoT Wi h Fede a ed, and IoT Fog Cloud, and no ed he To al
o al delays o all asks in he en i onmen . Fig. 10 shows he pe -
o mance o di e en schemes, bu SFTS ou pe o med all exis ing
me hods.
In ou simula ion c i e ia, To al a iable o all asks mus be
less gi en he deadlines o all asks du ing execu ion on he e oge-
neous nodes. We de e mined he local p ocessing delay du ing
mobile o loading wi h he p oposed scheme SFTS. The y-axis in
Fig. 11 shows ha all he mobile o loading asks wi h he collec ed
da a ake 5 o 7 min om senso collec ion o mobile p ocessing
and du ing o loading in SFTS. The secu i y and ailu e delay is also
con olled unde he gi en deadlines. We de e mined he o al
delays as To al incu ed wi hin 10 min du ing mobile o loading
wi h 20 asks. The baseline IoT Wi h a Fede a ed scheme execu ed
all mobile p ocessing asks, including ailu e and secu i y, wi hin
Fig. 8. Secu e and aul - ole an o al p ocessing ime o all asks on di e en compu ing nodes.
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
16
9 min wi h a andom numbe o 20 asks. Fu he mo e, he IoT Fog
Cloud scheme incu ed 9 min o mobile o loading, 10 min o
communica ion, and 12 min du ing og and cloud scheduling o
all asks. We analyzed he pe o mance o di e en a chi ec u es,
and schemes we e e alua ed based on a ious asks du ing he
expe imen . We esea ched and moni o ed he pe o mance o all
s a egies: SFTS, IoT Wi hou Fede a ed, IoT Wi h Fede a ed, and
IoT Fog Cloud, and no ed he To al o al delays o all asks in he
en i onmen . Fig. 11 shows he pe o mance o di e en schemes,
bu SFTS ou pe o med all exis ing me hods.
Ou simula ion conside s he ade-o cons ain s be ween he
esou ce consump ion o di e en compu ing nodes and objec i e
unc ion To al o all asks. The ade-o is always con lic ing
because he e is less delay in he aim unc ion To al, which con-
sumes a highe a io o compu ing esou ces. The e o e, we sched-
ule all asks based on gi en deadlines o handle he esou ce
cons ain issues o mobile de ices and less esou ce consump ion
o og and cloud nodes. These cons ain s, such as secu i y, ailu e,
and deadline, consume he esou ces o all compu ing nodes. We
de e mined he ailu e and secu i y-enabled scheduling wi h he
checkpoin ing mechanism, whe e all nodes can esume ailu e o
asks om he poin o ailu e. We analyzed he pe o mance o di -
e en a chi ec u es and schemes e alua ed based on di e se asks
du ing he expe imen . We examined and moni o ed he pe o -
mance o all s a egies: SFTS, IoT Wi hou Fede a ed, IoT Wi h Fed-
e a ed, and IoT Fog Cloud. We no ed he To al o al delays o all
asks in he en i onmen . Fig. 12 shows he pe o mance o di e -
en schemes, bu SFTS ou pe o med all exis ing me hods. Fig. 12
Fig. 9. Reco e y o al delays o asks du ing secu i y and ailu e issues in di e en compu ing nodes.
Fig. 10. O loading and scheduling pe o mance o heal hca e asks wi h di e en schemes on di e en compu ing nodes.
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
17
shows he p oposed scheme SFTS execu ed all asks wi h he min-
imum esou ce consump ion unde deadlines wi h 10-min delays.
These app oaches, IoT Fede a ed and IoT Fog Cloud, consume many
esou ces and a e incu ed wi h a o al o 12 o 14 min o delays.
The baseline app oaches also consume much mo e esou ces due
o scheduling all asks wi hou deadlines on a ailable nodes.
Resou ce consump ion is he key challenge in he IIoT heal h-
ca e en i onmen . The e a e di e en kinds o esou ce consump-
ion o he IIoT heal hca e en i onmen . The isk o secu i y and
esou ce a ailabili y ailu e is he key challenge o IIoT a chi ec-
u es. The o loading and scheduling a e he schemes in which
we main ained he esou ce consump ion in he ne wo k. We ana-
lyzed he pe o mance o di e en a chi ec u es and schemes
based on di e en asks du ing an expe imen . We analyzed and
moni o ed he pe o mance o all schemes SFTS, IoT Wi hou Fed-
e a ed, IoT Wi h Fede a ed, and IoT Fog Cloud and no ed he To al
o al delays o all asks in he en i onmen . Fig. 13 shows he pe -
o mance o di e en schemes, bu SFTS ou pe o med all exis ing
schemes. The di e en aspec s a e e alua ed, such as ailu e a io,
secu i y alida ion and esou ce leakage, and a ailabili y o og and
cloud nodes du ing o loading o p ocessing. I has been obse ed
ha he baseline schemes wi h di e en IoT heal hca e asks ha e
a ailu e a io highe han SFTS. The secu i y alida ion is mo e
app op ia e wi h he SFTS han baseline schemes.
We alida ed he wo k in di e en aspec s, such as local p o-
cessing delay, o loading communica ion delay (DT), og p ocessing
Fig. 11. Secu e and ailu e enabled pe o mances o schemes.
Fig. 12. Failu e and secu i y eco e y pe o mances o asks wi h di e en schemes.
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
18
delay o aining and alida ion, and cloud delay. We analyzed he
ailu e and secu i y aspec s wi h he di e ences as shown in
Table 7. The di e en scena ios such as Me hod,
T;L
e
;C
e
;Cloud
e
;F
e
;To al, Failu e, and Secu i y a e he cons ain s
whe e pe o mance e alua ion can be compa ed and analyzed wi h
he di e en me hods. We can analyze he pe o mance o he p o-
posed scheme wi h di e en me hods. Fo ins ance, SFTS ob ained
he esul s o asks in di e en scena ios and di e en nodes, asks
= 1–10, local p ocessing = 6, o loading ime = 8, og delay = 6,
cloud delay = 10, and o al delay = 30, whe e only h ee asks 3
a e ailed, 10 shows ha , all asks a e success ully me he secu i y
be ween nodes wi hou any ailu e o asks. We analyzed he all
esul s o all me hods wi h di e en asks om = 1 o 30 wi h di -
e en me hods as shown in Table 7.
Secu i y alida ion in di e en nodes is a c ucial p ocess in di -
e en heal hca e en i onmen s. Table 8 shows he da a’s enc yp-
ion and dec yp ion alida ion on di e en nodes. The
pa ame e s a e conside ed in his phase, as shown in he ollowing
way. Fo ins ance, Task, Da a, Enc yp ion, Dec yp ion, Valida ion,
and S a us a e he enc yp ion and dec yp ion alida ions a he
ede a ed lea ning-enabled DT mobile og cloud ne wo k. Fo
ins ance, a ask = 1 has da a 99, has enc yp ion eO2aTgb6g2VL
+83 oW2jCw==, and again dec yp s a 99. A he same ime, hese
alida ions exis be ween m1d 1 wi h he s a us yes. Fu he -
mo e, his da a is o loaded om he DT og node o cloud compu -
ing. Fo ins ance, a ask = 1 wi h da a 120, enc yp ion
KWuVYMKzEW0oPeb4ydQd7w==, and dec yp ion 120, be ween
d 1s1 wi h s a us yes. In his way, each piece o da a om
mobile de ices o og nodes and og nodes o cloud compu ing is
alida ed based on enc yp ion and dec yp ion wi h he sha ed keys
ede a ed by mobile og cloud ne wo ks.
We discussed he di e en me ics ha we e iden i ied du ing
he simula ion. We discussed hese me ics, such as esou ce con-
sump ion, Scalabili y, and memo y% du ing he execu ion o asks
in he simula o as shown in Table 9. We analyzed hese me ics
wi h he di e en baseline app oaches and p oposed me hods o
he di e en numbe o asks, e.g., 30. We e alua ed hese me ics
on he mobile node, og nodes, and cloud compu ing. In ou simu-
la ion, we keep he scalabili y ixed, which means in ou cu en
p oblem, we a e no conside ing he esou ce p o isioning wi h
he cos cons ain s. The e o e, we keep he scalabili y ixed wi h-
ou scaling up and scaling down du ing he execu ion o asks in
he simula o . We de e mine he esou ce consump ion in mega-
by es (MB) as shown in Table 9 o all me hods. We de e mined
he memo y usage a io o he execu ion o all asks on di e en
nodes wi h he di e en me hods, as shown in Table 9. We can
Fig. 13. Resou ce consump ion o IIoHT senso y asks wi h di e en ailu e and secu i y schemes.
Table 7
Wo k alida ion in di e en aspec s and scena ios wi h di e en .
Me hod T L
e
C
e
Cloud
e
F
e
To al Failu e Secu i y
SFTS = 1–10 6 8 6 10 30 3 10
Wi hou Fede a ed = 1–10 16 18 16 10 60 3 10
IoT Wi h Fede a ed = 1–10 12 10 7 13 42 5 7
IoT Fog Cloud 30 = 1–10 15 15 18 48 7 9
SFTS = 11–20 7 6 8 10 31 2 10
Wi hou Fede a ed = 11–20 17 17 17 10 61 3 8
IoT Wi h Fede a ed = 11–20 10 10 10 16 46 3 9
IoT Fog Cloud 30 = 11–20 20 20 20 60 6 9
SFTS = 21–30 6 8 6 11 31 2 10
Wi hou Fede a ed = 21–30 16 20 16 10 62 4 8
IoT Wi h Fede a ed = 21–30 12 12 7 13 44 1 5
IoT Fog Cloud = 21–30 30 15 15 20 60 1 6
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
19
obse e om Table 9 ha SFTS consumes less esou ces and mem-
o y han exis ing me hods on mobile og and cloud nodes du ing
he execu ion o asks. The main eason is ha we implemen ed
he eplica o he da a p ocessing based on digi al wins. The e o e,
he esul s and da a eco e y om he cloud a e mig a ed o down-
loaded o he og nodes o scheduling. I is a obus and e icien
way o use he digi al win-enabled og and cloud nodes o he
dis ibu ed Io applica ions and minimize he esou ce consump-
ion and memo y usage o simila asks du ing secu i y and ailu e
si ua ions in ou a chi ec u e.
5.3. Findings and sho comings o SFTS and baseline algo i hms
In his s udy, we p esen ed he SFTS me hod, which consis s o
di e en schemes such as local p ocessing, ede a ed lea ning-
enabled og and cloud nodes, and digi al win. The main inding
o he SFTS is o execu e all IoT con en asks wi h minimum
delays. The o al delays as we de e mined in Eq. (6) combina ions
o di e en delays. The e o e, we scheduled all IoT asks o he di -
e en compu ing nodes wi h minimum delays. In he esul dis-
cussion, we showed he indings and limi a ions o SFTS o all
asks on mobile, og, and cloud nodes. We ha e implemen ed he
ou baseline s a egies, IoT Wi hou Fede a ed lea ning, IoT Wi h
Fede a ed lea ning, and IoT Fog Cloud schemes o IoT con en
asks in he simula ion en i onmen . We conside ed he di e en
cons ain s such as secu i y, p ocessing delay, deadline, esou ce
consump ion, and ailu e o asks. We analyzed he pe o mances
o all algo i hms as shown in he esul analysis and discussion
wi h he IoT andom numbe o asks in og and cloud ne wo ks.
We in es iga ed IoT asks’ secu i y and ailu e delay cons ain s
on mobile, og, and cloud nodes wi h all algo i hms designed du -
ing execu ion in he a chi ec u e. This baseline IoT Wi hou a Fed-
e a ed lea ning s a egy, scheduled all mobile, og, and cloud
ne wo k asks. The s a egy IoT Wi hou Fede a ed lea ning sched-
uled all asks on di e en compu ing nodes in a secu e and delay-
e icien o m. Howe e , as all simula ion esul s show, his s a -
egy has su e ed highe delays. The main eason is ha , in his
s a egy, he main node made all he decisions du ing he secu i y
eco e y and ailu e o asks and acknowledged all nodes. Tha
means all nodes a e clus e ed in he ne wo k, bu decisions made
by cen alized nodes su e highe delays o all asks. In he p e i-
ous highe delays, all he asks missed hei deadlines and
deg aded he pe o mances o applica ions. This IoT Fog Cloud
s a egy di ided he scheduling asks decision among di e en
compu ing nodes based on he schedule and go less delay han
IoT wi hou he ede a ed s a egy. The main eason is ha all
he schedule s can communica e wi h each o he and eschedule
he ailu e o asks om he poin o ailu e. Howe e , due o many
cons ain s, such as local p ocessing, communica ion, ailu e, secu-
i y, and emo e p ocessing, his s a egy su e ed om delays ana-
lyzed one cons ain a a ime. The ede a ed lea ning di ided he
compu ing analyzing o di e en a di e en og and cloud and
agg ega ed o he cen alized nodes. Howe e , one node’s ailu e
in ede a ed lea ning s ill su e s om highe delays in he ne -
wo k. SFTS is he op imal s a egy ha in eg a es he digi al win
on og nodes, whe e cloud nodes a e eplica ed and execu ed on
og nodes. The digi al win is adap i ely in eg a ed wi h he cloud
nodes, whe e, based on ede a ed lea ning, we can exchange hei
upda es o all connec ed nodes. The e o e, he SFTS s a egy o IoT
asks in he ne wo k can easily manage all he cons ain s on di -
e en compu ing nodes. Howe e , he e a e s ill limi a ions in
he SFTS scheme o b oad-le el in as uc u e. All he p oposed
and baseline s a egies did no conside he wai ime o asks
be o e scheduling in he mobile og cloud ne wo ks. The baseline
s a egies and p oposed wo k widely miss ask powe consump-
ion, cos , and sus ainabili y. These limi a ions s ill need o
imp o e he o al delays and ime complexi y o become e icien
o all ime-sensi i e asks wi h he gi en p io i y. The e o e, all
baseline and p oposed app oaches mus add ess hese limi a ions
in u u e wo ks. In he cu en e sion o me hods, we did no con-
side esou ce p o isioning o a oid esou ce scalabili y. In his
wo k, we only exploi ed he ixed ype o esou ces. We keep he
ixed scalabili y in he cu en a chi ec u e e sion and do no con-
side esou ce p o isioning. Howe e , he di e si y ea u es can
inc ease he a io o asks and ypes. The e o e, powe consump-
ion on local de ices and wi eless ne wo ks could be inc eased.
In u u e wo k, we conside he powe consump ion, cos , esou ce
p o isioning, scalabili y, and exis ing cons ain s wi h he mo e
obus and adap i e schemes.
6. Conclusion
Based on pe o mance e alua ion, he s udy analyzed and p o-
cessed he di e en heal hca e senso y da a o moni o and p edic
Table 8
Enc yp ion and dec yp ion de ec ion schemes.
Task Da a Enc yp ion Dec yp ion Valida ion S a us
= 1 99 eO2aTgb6g2VL + 83 oW2jCw== 99 m1d 1 yes
= 1 120 KWuVYMKzEW0oPeb4ydQd7w== 120 d 1s1 yes
= 2 155 K7 + 93/Wi0ecCKP14ySmoeg== 155 m1d 2 yes
= 2 1000 Q h bmiJ o19Y9/JAyRiQ== 1000 d 2s2 yes
Table 9
Enc yp ion and dec yp ion de ec ion schemes.
Me hod Task Node Resou ce Consump ion Scalabili y Memo y%
SFTS 30 Mobile 380 MB Fixed 0.3
SFTS 30 Fog Nodes 1000 MB Fixed 0.7
SFTS 30 Cloud 1500 MB Fixed 0.9
IoT Wi hou Fede a ed 30 Mobile 500 MB Fixed 0.8
IoT Wi hou Fede a ed 30 Fog Nodes 1500 MB Fixed 0.9
IoT Wi hou Fede a ed 30 Cloud 2000 MB Fixed 0.12
IoT Wi h Fede a ed 30 Mobile 500 MB Fixed 0.7
IoT Wi h Fede a ed 30 Fog Nodes 1600 MB Fixed 0.14
IoT Wi h Fede a ed 30 Cloud 2200 MB Fixed 0.16
IoT Fog Cloud 30 Mobile 700 MB Fixed 0.9
IoT Fog Cloud 30 Fog Nodes 2000 MB Fixed 0.19
IoT Fog Cloud 30 Cloud 3000 MB Fixed 0.21
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
20
he heal hca e da a in dis ibu ed mobile og and cloud ne wo ks.
This pape p esen ed he secu e, aul - ole an , empowe ed wea -
able heal hca e senso s awa e Indus ial In e ne o Things (IIoT)
F amewo k based on digi al win ede a ed og-cloud models.
The s udy aims o educe he ime and esou ces needed o p ocess
heal hca e senso da a o secu i y, ask execu ion, and aul ole -
ance. The s udy p esen s he Secu e and Faul -Tole an
Scheme (SFTS) algo i hm amewo k ha op imizes he IoT senso
da a and execu es he heal hca e da a wi h he minimum o load-
ing and p ocessing delays. Simula ion esul s show ha he p o-
posed wo k minimized he secu i y isk by 40%, ailu e isk o
asks isk by 50%, and he aining and es ing ime by 39% o all
IIoT heal hca e asks in mobile og cloud ne wo ks.
The s udy will implemen blockchain echnology in u u e wo k
and conside he di e en heal hca e clinics o senso y da a in
he e ogeneous compu ing nodes. The cu en wo k e sion has
powe consump ion, elec ici y cos , and dis ibu ed esou ce sha -
ing ha ha e ye o be conside ed in he p esen wo k. In u u e
wo k, we will add mo e cons ain s in he conside ed amewo k
wi h he disc e e and polynomial ime o all asks. The e exis con-
s ain s in he SFTS ha ela e o he ime complexi y associa ed
wi h di e se limi a ions. To illus a e, mobile de ice wai ing ime
du ing o loading has ye o be es ablished. As a esul , asks wi hin
he scope o IoT migh encoun e challenges wi h mee ing deadli-
nes and main aining pe o mance due o p olonged wai ing in e -
als. The di icul y o achie ing load balance in ede a ed lea ning
be ween og and cloud nodes endu es. Consequen ly, hese consid-
e a ions ied o he equilib ium o og and cloud nodes lead o elon-
ga ed wai ing du a ion. This ci cums ance signi ican ly impac s he
ime complexi y o IoT og cloud asks wi hin he amewo k o ou
a chi ec u e. The e o e, in u u e wo k, we will conside he delays
o IoT og cloud asks in ou ex ended a chi ec u e.
Au ho Con ibu ions
All au ho s con ibu ed equally o he inal dissemina ion o he
esea ch in es iga ion as a ull a icle. All au ho s ha e ead and
ag eed o he published e sion o he manusc ip .
E hical App o al
The manusc ip does no epo on o in ol e he use o any ani-
mal o issue and is ‘‘No applicable” o his manusc ip .
Funding
This a icle was co- unded by he Eu opean Union unde he
REFRESH - Resea ch Excellence Fo REgion Sus ainabili y and
High- ech Indus ies p ojec numbe
CZ.10.03.01/00/22_003/0000048 ia he Ope a ional P og amme
Jus T ansi ion. Also, his wo k was suppo ed by he Minis y o
Educa ion, You h and Spo s o he Chezk Republic conduc ed by
VSB - Technical Uni e si y o Os a a, Czechia unde G an s
SP2023/039 and SP2023/042.
Da a A ailabili y S a emen s
The s udy exploi ed he public heal hca e senso s, and da ase s
a e publically a ailable on he ollowing URL. h ps://gi hub.com/
ABDULLAH-RAZA/IIoTheal hca e-Senso s-Da a. The da ase s ha e
di e en ea u es such as ECG Lead-1, Lead-2, Ankle, BP (blood
p essu e), Temp (Tempe a u e), HB (Hea bea s), CH (Choles e ol)
speed, loca ion, eal- ime, s ayed Ac i i y, and Use s.
Decla a ion o Compe ing In e es
The au ho s decla e ha hey ha e no known compe ing inan-
cial in e es s o pe sonal ela ionships ha could ha e appea ed
o in luence he wo k epo ed in his pape .
Re e ences
Akash, S.S., Fe dous, M.S., 2022. A blockchain based sys em o heal hca e digi al
win. IEEE Access 10, 50523–50547.
Alaasam, A.B., Radchenko, G., Tche nykh, A., González Compeán, J., 2020. Analy ic
s udy o con aine izing s a e ul s eam p ocessing as mic ose ice o suppo
digi al wins in og compu ing. P og amm. Compu . So w. 46, 511–525.
Alsha h i, S., Hemdan, E.E.-D., El-Sha ai, W., Sayed, A., 2023. Digi al win-based
au oma ed aul diagnosis in indus ial io applica ions. CMC-Compu . Ma e .
Con inua 75 (1), 183–196.
Azzaoui, A.E., Kim, T.W., Loia, V., Pa k, J.H., 2021. Blockchain-based secu e digi al
win amewo k o sma heal hy ci y. Ad . Mul imedia Ubiqui . Eng. 716, 107.
Chi, H.R., Wu, C.K., Huang, N.-F., Tsang, K.F., Radwan, A., 2022. A su ey o ne wo k
au oma ion o indus ial in e ne -o - hings owa ds indus y 5.0. IEEE T ans.
Indus . In .
Daigneaul , J., S -Hilai e, M., 2021. P o i maximiza ion model o he ask
assignmen p oblem in 2- ie og/cloud ne wo k en i onmen s. IEEE Ne wo k.
Le . 3 (1), 19–22.
Da ishi, H., Ciuonzo, D., Rossi, P.S., 2021. Real- ime senso aul de ec ion, isola ion
and accommoda ion o indus ial digi al wins. In: 2021 IEEE In e na ional
Con e ence on Ne wo king, Sensing and Con ol (ICNSC), ol. 1. IEEE, pp. 1–6.
Dha angan, B., P a een, J., Rajagopal, S., Jegajo hi, B. e al., 2022. Secu e cloud-based
e-heal h sys em using ad anced enc yp ion s anda d. In: 2022 3 d
In e na ional Con e ence on Elec onics and Sus ainable Communica ion
Sys ems (ICESC). IEEE, pp. 642–646.
Elayan, H., Aloqaily, M., Guizani, M., 2021. Digi al win o in elligen con ex -awa e
io heal hca e sys ems. IEEE In e ne Things J. 8 (23), 16 749–16 757.
el Azzaoui, A., Kim, T.W., Loia, V., Pa k, J.H., 2020. Blockchain-based secu e digi al
win amewo k o sma heal hy ci y. In: Ad anced Mul imedia and
Ubiqui ous Enginee ing: MUE-Fu u eTech 2020. Sp inge , pp. 107–113.
Ghi a, M., Siham, B., Hicham, M., Abdelha id, A.E.M., Lau en , D., 2020. Geospa ial
business in elligence and cloud se ices o con ex awa e digi al wins
de elopmen . In: 2020 IEEE In e na ional con e ence o Mo occan Geoma ics
(Mo geo). IEEE, pp. 1–6.
Haleem, A., Ja aid, M., Singh, R.P., Suman, R., 2023. Explo ing he e olu ion in
heal hca e sys ems h ough he applica ions o digi al win echnology. Biomed.
Technol. 4, 28–38.
Jean E il, N., Rajeswa i, R., 2022. Wa ele sca e ing ans o m and long sho - e m
memo y ne wo k-based nonin asi e blood p essu e es ima ion om
pho ople hysmog aph signals. SIViP 16 (1), 1–9.
Jimenez, J.I., Jahankhani, H., Kendzie skyj, S., 2020. Heal h ca e in he cybe space:
Medical cybe -physical sys em and digi al win challenges. Digi al Twin
Technol. Sma Ci ies, 79–92.
Khan, S., A slan, T., Ra na ajah, T., 2022. Digi al win pe spec i e o ou h indus ial
and heal hca e e olu ion. IEEE Access 10, 25732–25754.
Khoso, F.H., Lakhan, A., A ain, A.A., Soom o, M.A., Nizamani, S.Z., Kanwa , K., 2021. A
mic ose ice-based sys em o indus ial in e ne o hings in og-cloud assis ed
ne wo k. Eng. Technol. Appl. Sci. Res. 11 (2), 7029–7032.
Kim, S., 2020. New applica ion ask o loading algo i hms o edge, og, and cloud
compu ing pa adigms. Wi eless Commun. Mobile Compu . 2020, 1–14.
Konigsbu g, J.A., 2022. Mode n wa a e, spi i ual heal h, and he ole o a i icial
in elligence. Religions 13 (4), 343.
Lakhan, A., Mohammed, M.A., Kad y, S., Abdulka eem, K.H., Al-Dhie , F.T., Hsu, C.-H.,
2021. Fede a ed lea ning enables in elligen e lec ing su ace in og-cloud
enabled cellula ne wo k. Pee J Compu . Sci. 7, e758.
Lakhan, A., Mohammed, M.A., Abdulka eem, K.H., Jabe , M.M., Nedoma, J., Ma inek,
R., Zmij, P., 2022. Delay op imal schemes o in e ne o hings applica ions in
he e ogeneous edge cloud compu ing ne wo ks. Senso s 22 (16), 5937.
Leng, J., Sha, W., Wang, B., Zheng, P., Zhuang, C., Liu, Q., Wues , T., Mou zis, D.,
Wang, L., 2022. Indus y 5.0: P ospec and e ospec . J. Manu . Sys . 65, 279–
295.
Leong, Y.K., Tan, J.H., Chew, K.W., Show, P.L., 2021. Signi icance o indus y 5.0. In:
The P ospec o Indus y 5.0 in Biomanu ac u ing. CRC P ess, pp. 95–114.
L , Z., 2023. Digi al wins in indus y 5.0. Resea ch 6, 0071.
Naha, R.K., Ga g, S., Chan, A., Ba ula, S.K., 2020. Deadline-based dynamic esou ce
alloca ion and p o isioning algo i hms in og-cloud en i onmen . Fu u e
Gene a . Compu . Sys . 104, 131–141.
Nguyen, T.N., Ponci oli, R., B uck, P., Esselman, T.C., Riga i, J.A., Vilim, R.B., 2022. A
digi al win app oach o sys em-le el aul de ec ion and diagnosis o
imp o ed equipmen heal h moni o ing. Ann. Nucl. Ene gy 170, 109002.
Panda, S.K., Nanda, S.S., Bhoi, S.K., 2022. A pai -based ask scheduling algo i hm o
cloud compu ing en i onmen . J. King Saud Uni .-Compu . In o . Sci. 34 (1),
1434–1445.
Rashid, L., Rubab, S., Alhaisoni, M., Alqah ani, A., Alsubai, S., Binbusayyis, A., Bukha i,
S.A.C., 2022. Analysis o dimensionali y educ ion echniques on in e ne o
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
21
hings da a using machine lea ning. Sus ain. Ene gy Technol. Assessm. 52,
102304.
Rieke, N., Hancox, J., Li, W., Mille a i, F., Ro h, H.R., Alba qouni, S., Bakas, S., Gal ie ,
M.N., Landman, B.A., Maie -Hein, K., e al., 2020. The u u e o digi al heal h
wi h ede a ed lea ning. NPJ Digi al Med. 3 (1), 119.
Salman, A.O., Geman, O., 2023. E alua ing h ee machine lea ning classi ica ion
me hods o e ec i e co id-19 diagnosis. In . J. Ma he. S a . Compu . Sci. 1, 1–
14. h ps://doi.o g/10.59543/ijmscs. 1i.7693.
Unnisa, S., Vijayalakshmi, A., Jagun, Z.T., 2023. Deep neu al ne wo k a chi ec u e
and applica ions in heal hca e. Deep Lea n. Heal hca e Decis. Mak., 25
Volko , I., Radchenko, G., Tche nykh, A., 2021. Digi al wins, in e ne o hings and
mobile medicine: a e iew o cu en pla o ms o suppo sma heal hca e.
P og amm. Compu . So w. 47, 578–590.
Xu, J., Glicksbe g, B.S., Su, C., Walke , P., Bian, J., Wang, F., 2021. Fede a ed lea ning
o heal hca e in o ma ics. J. Heal hca e In o ma . Res. 5, 1–19.
Yang, Z., Yang, Y., Xu, C., 2022. Demons a ion on unblocking checkpoin o aul -
ole ance in p egel-like sys ems. In: Web and Big Da a: 6 h In e na ional Join
Con e ence, APWeb-WAIM 2022, Nanjing, China, No embe 25–27, 2022,
P oceedings, Pa III. Sp inge , pp. 456–460.
Younan, M., Houssein, E.H., Elhoseny, M., Ali, A.A., 2020. Challenges and
ecommended echnologies o he indus ial in e ne o hings: A
comp ehensi e e iew. Measu emen 151, 107198.
Zhang, J., Li, L., Lin, G., Fang, D., Tai, Y., Huang, J., 2020. Cybe esilience in heal hca e
digi al win on lung cance . IEEE Access 8, 201900–201913.
A. Lakhan, A.A. Abdul La ee , M.K. Abd Ghani e al. Jou nal o King Saud Uni e si y – Compu e and In o ma ion Sciences 35 (2023) 101747
22
