TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 12, No. 10, Octobe
r 20
14, pp. 7471
~ 747
7
DOI: 10.115
9
1
/telkomni
ka.
v
12i8.558
2
7471
Re
cei
v
ed
Jan
uary 5, 2014;
Re
vised July
24, 2014; Accepted Augu
st
18, 2014
Perfect Forward Secure ID-based Key Agreement
Protocol in Group Communication
Pengshu
ai Qiao
Schoo
l of Envir
onme
n
tal a
nd
Munici
pa
l Engi
neer
ing
North Ch
ina U
n
iversit
y
of W
a
ter Resourc
e
s and El
ectric Po
w
e
r
email: p
engs
hu
aiqi
ao
@16
3
.co
m
A
b
st
r
a
ct
Severa
l id
entit
y-base
d
key a
g
ree
m
ent prot
ocols
us
in
g bi
l
i
ne
ar pa
irin
g h
a
ve b
e
e
n
pro
p
o
sed
i
n
recent ye
ars a
nd n
one
of th
em
has
achi
e
v
ed a
ll re
quir
e
d security
pro
perties. In this
pap
er, w
e
firstly
prop
ose
an
ID-
base
d
o
n
e
rou
nd
auth
enticat
ed
grou
p k
e
y a
g
ree
m
ent pr
otocol
w
i
th bi
lin
e
a
r p
a
iri
ngs, w
h
er
e
all p
a
rticip
ants can ge
ner
ate t
he gro
up sess
i
on key in o
ne r
oun
d. Based o
n
the intracta
bil
i
ty of elliptic cu
rve
discrete l
o
g
a
rit
h
m
prob
le
m, e
v
ery us
er
’
s
pr
iv
ate key can
be
proved to
be s
e
cure. Also
an
extend
ed vers
i
on
of one rou
nd a
u
thentic
ated gr
oup key a
g
re
e
m
e
n
t protoco
l
i
s
given, it prov
ide p
e
rfect forw
ard secrecy an
d
avoi
d key escrow
by the Key G
enerati
o
n
Center. F
i
nally, a compreh
ensiv
e securit
y
analysis a
n
d
a
compre
hens
ive
security
an
aly
s
is are
pr
ovid
e
d
. By co
m
pari
ng
w
i
th other protoco
l
s,
the
prop
osed
pr
otoc
o
l
requ
ires low
e
r
computati
on co
st.
Ke
y
w
ords
: ke
y agree
ment pr
otocol, perfect
forw
ard secrec
y
Co
p
y
rig
h
t
©
2014 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
In recent years,
colla
bor
ative and group-ori
ented
applic
ation
s
and protocols hav
e
gaine
d po
pul
arity. These a
pplication
s
typically
involv
e co
mmuni
ca
tion over
ope
n networks. One
of the important require
ments
is
se
curity. A key agreeme
n
t which p
r
ovi
des mutu
al key
authenti
c
atio
n bet
wee
n
p
a
rties i
s
calle
d an
aut
h
enti
c
ated
g
r
oup
key a
g
reeme
n
t (AGKA). K
e
y
establi
s
hm
en
t protocols a
r
e one of the
most impo
rta
n
t cryptog
r
ap
hic p
r
imitives that have be
en
use
d
in o
u
r
so
ciety. In 1
976, the first
unauth
enticated
key
ag
reement prot
ocol ba
sed on
asymmet
r
ic cryptographi
c tech
niqu
es was propo
se
d by Diffie and Hellma
n
[1]. It can assu
re th
e
se
curity of communi
catio
n
between th
e two u
s
e
r
s.
Ho
wever, it
dose not
a
u
thenticate u
s
e
r
s,
hen
ce suffe
rs the “man-in
-t
he-mi
ddle
”
attack. In
1984, Shamir [2] propo
sed the id
ea of ID-ba
s
e
d
crypto
system
where the identity in
formation of a use
r
functio
n
s
as hi
s pu
bl
ic key. A few key
agre
e
me
nt protocol
s h
a
ve
been
devel
o
ped b
a
sed
o
n
Diffie-Hell
man a
nd Sh
amir’
s
key setup
idea. In one
of brea
kthrou
ghs in
key a
g
ree
m
ent, Jo
ux [3] proposed a three p
a
rty single
ro
und
key ag
ree
m
e
n
t proto
c
ol u
s
ing pai
ring
s o
n
elliptic
cu
rve. This
wa
s t
he first
po
sitive appli
c
atio
n of
bilinea
r
p
a
iri
n
gs
i
n
crypto
g
r
aphy. Jo
ux et
al.
a
pplie
s the
pairi
ng t
e
ch
niqu
e a
n
d
a
c
hieve
s
ke
y
agre
e
me
nt a
m
ong th
ree
p
a
rties in a
n
a
s
toni
shin
gly simple
way. He nam
es his
proto
c
ol
“trip
a
r
tite
Diffie-Hellma
n
”. Again, Jo
ux’s ori
g
inal
proto
c
ol
works in th
e Weil pairin
g
an
d hen
ce i
s
less
conve
n
ient fo
r a real a
ppli
c
ation u
s
e. Here
we
intro
d
u
ce a
simplifi
ed version u
s
ing the modifi
ed
Weil pai
ring.
Since
Bone
h
and
F
r
an
kli
n
’s
pion
eeri
n
g work
[4]
on the ID-bas
ed encr
yption (
I
BE)
system i
n
2
001,
several
pap
ers h
a
ve attemp
ted
to e
s
tablish
ID-b
ased
a
u
thenticated
key
agre
e
me
nt protocol
(ID-AG
KA). Choi et
al. [5] and
Du
et al. [6] prop
ose
d
two I
D
-AGKA proto
c
ols
from bilin
ea
r pairi
ng
s an
d BD [7]
scheme
s
. However, Zh
ang
and
Ch
en
[8] sho
w
e
d
an
imperso
natio
n attack on
th
ese
two
p
r
oto
c
ol
s. To
pr
event such a
n
a
ttack, they
su
gge
st ad
ding
a
time param
eter to the messag
e being
si
gned. However,
SHIM [9] sho
w
e
d
that the proto
c
ol is
still inse
cu
re
again
s
t insi
d
e
r colludin
g
attacks. In
2006, Lin et al
. [10] propo
sed a multip
arty
key a
g
reeme
n
t protocol,
but thei
r p
r
ot
ocol
ha
s
di
sadvantag
es i
n
nu
mbe
r
of ro
und
s, p
a
iring-
comp
utation and com
m
un
ication
ban
d
w
idth.
Zh
ou
et al. propo
sed a
one
-to
-
many ma
ppi
ng
sha
r
ed
key a
g
ree
m
ent, which i
s
ba
se
d
on one
-t
o-m
any encryptio
n mechani
sm
model, but t
h
e
roun
d num
be
r of their sche
me is two [11
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 10, Octobe
r 2014: 747
1
– 7477
7472
To reali
z
e g
r
oup key agre
e
ment and e
x
tend Joux
e
t
al.’s proto
c
ol, Barua et al. [12]’s
first pro
p
o
s
ed
a three-gro
u
p
and a two
-
grou
p Diffi
e-Hellma
n
key
agre
e
me
nt protocol. After that,
many protoco
l
s were propo
sed i
n
[13
-
17]
. Abdel
Alim
Kamal p
r
opo
sed
an atta
ck on Piao
et al.’s
scheme
whi
c
h de
scribe
s
a poly
nomi
a
l-based key
m
anag
ement schem
e
for se
cure
intra-gro
u
p
and i
n
ter-gro
up
comm
uni
cation
[18].
Marimuth
u
Rajara
m a
nd
Thilagavathy
Do
rai
r
aj Su
resh
prop
osed a
n
interval-ba
s
e
d
key
agree
ment ap
pr
o
a
c
h
whi
c
h a
d
p
o
ts re-keying
[19]. To de
crease
the numbe
r o
f
round
s and
make AGKA more efficie
n
t
,
Shi et al. propo
sed on
e round ID-ba
s
e
d
AGKA protocol with biline
a
r pari
n
g
s
[20], which ca
n gene
rate the se
cret se
ssi
on key in
one
roun
d. Shi et
al.’s p
r
oto
c
ol
just requi
re
s one
ro
u
nd a
nd le
ss t
r
an
smitted data,
so it ha
s a
g
ood
efficient. However, a
s
illu
st
rated i
n
thei
r l
i
terature,
in t
heir
protocol i
f
two o
r
m
o
re
than t
w
o
use
r
s’
long-te
rm p
r
i
v
ate keys a
r
e
comp
romi
se
d, the adv
ersary ca
n comp
ute the
previous se
ssion key,
so thei
r p
r
ot
ocol
ca
nnot
provide
pe
rfe
c
t forward
se
cre
c
y. Simila
rly, their p
r
ot
ocol
also
ca
nnot
prevent KG
C from escrowi
ng the esta
bli
s
he
d se
ss
ion k
e
ys
. Based on Shi et al.’s
work
, we firs
t
prop
ose a
n
I
D
-b
ased
one
roun
d a
u
then
ticated
gro
u
p
key a
g
re
eme
n
t protocol
which
satisfie
s
the
requi
re
d se
curity attributes and provid
e lowe
r
com
p
utation co
st. The pro
p
o
s
e
d
pape
r’s section
stru
cture is o
r
gani
zed a
s
:
Introdu
ction
- Security
propertie
s - Pr
oposed I
D
-AGKA - Secu
r
i
t
y
analy
s
is –Efficienc
y
analy
s
is– Concl
u
sion.
2. Securit
y
p
r
operti
e
s
An authenticated gro
up key agreeme
n
t protoc
ol is de
sire
d to have the followi
n
g
comm
on security prop
ertie
s
[21, 22]:
1) Impli
c
it Key Authenticati
on: An n-p
a
rt
y ke
y agree
ment protocol
provide
s
imp
licit key
authenti
c
atio
n if ea
ch me
mber i
n
the
set of proto
c
ol
partie
s
i
s
a
s
sured
that no
party out
side
the
set ca
n learn the grou
p se
cret key.
2) Pe
rfect F
o
rward Se
crecy
:
We
say that
a p
r
oto
c
ol h
a
s p
a
rtial fo
rward
se
crecy if
one
or
more
but not
all the entitie
s' lo
ng-te
rm
keys
can
b
e
corrupted with
out
co
mpromi
sing previously
establi
s
h
ed session keys, and we
say that a prot
oco
l
has perfe
ct forwa
r
d secrecy if the long-
term key
s
of all the entities involved m
a
y be co
rru
pted witho
u
t co
mpromi
sin
g
a
n
y sessio
n key
previou
s
ly est
ablished
by these entities.
3)
Know
n S
e
ss
ion Key
Sec
u
r
i
ty: R
e
s
i
s
t
ance to known s
e
ss
ion k
e
y secur
i
ty is
the
prop
erty that each run
pro
duces a diffe
rent se
ssio
n key and
com
p
romi
se of p
a
st se
ssion
keys
doe
s not allo
w com
p
romise of future se
ssi
on keys.
4) Key-Comp
romi
se Impe
rson
ation: Wh
en A’s privat
e key is
com
p
romi
se
d, it
may be
desi
r
abl
e that this event do
es not en
able
an adv
ersa
ry to impersona
te other entities to A.
5) Un
kno
w
n Key-Share:
In
an un
kno
w
n
ke
y
-
share attack, an
a
d
v
ersa
ry convi
n
ce
s a
grou
p of entities that they sha
r
e a
key with t
he adv
ersary, wh
ereas in fa
ct the key is
sha
r
ed
betwe
en the
grou
p and a
n
o
ther pa
rty.
6) No Key Co
ntrol: It should not be possible
for any of the participa
nts or an adv
ersary
to force the
session
key to a pre
-
sele
cte
d
va
lue or p
r
e
d
ict the value
of the sessio
n key.
3. Proposed
ID-AGKA
3.1. Sy
stem
Setup
We ta
ke
G
1
t
o
be a cycli
c
elliptic cu
rve group with
large prim
e order q and the
bilinear
map e : G
1
×G
1
→
G
2
. The
key ge
neratio
n ce
nter
(KG
C
) g
ene
rat
e
s
t
he sy
st
e
m
p
a
ram
e
t
e
r
s
{q,
G
1
,
G
2
, e, P,
H
1
, H
2
}. s is rand
omly cho
s
en
from Z
q
*
as the KGC’
s pri
v
ate key. P
pu
b
(
=
sP
) i
s
t
h
e
KGC’s p
ubli
c
key. Each
user
U
i
has
an identity ID
i
∈
{0, 1}
*
and long
-term pu
blic
key
Q
i
=
H
1
(ID
i
)P
+
P
pub
. U
i
submits his ID
i
to the KGC and KGC
se
nds b
a
ck the
long-te
rm private
key
S
i
=
(H
1
(I
D
i
)+
s)
-1
P to user
U
i
se
cu
rel
y
.
3.2. The ID-AGKA
The protocol one ro
und a
u
t
henticate
d
group key
agre
e
ment incl
ud
es three pha
ses: data
transmission phase, verification pha
se and key computat
ion phase. It is illu
strated as follows:
1)
To two u
s
ers
U
i
, U
j
(1
≤
i, j
≤
m, i
≠
j), U
i
picks
a ran
d
o
m
integer r
i
∈
Z
q
*
as his
ep
heme
r
al
private key.
T
hen he co
mp
utes T
i
= r
i
-1
Q
i
and
send
s T
i
to
U
j
.
Upon
the re
ceipt of T
i
, U
j
also
pi
cks
a ran
dom int
eger
r
j
∈
Z
q
*
as hi
s eph
e
m
eral p
r
ivate
key, comp
utes T
j, i
= r
j
r
i
-1
Q
i
and T
j
= r
j
-1
Q
j
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Perfect Fo
rward Se
cure ID-b
ased Ke
y Agreem
ent
Protocol in Gro
up… (Pe
n
g
s
h
uai Qiao
)
7473
respe
c
tively, and then
se
n
d
s the d
a
ta {
T
j,i
, T
j
} to U
i
. Finally, U
i
co
mputes
T
i,j
= r
i
r
j
-1
Q
j
and
retu
rns
it to U
j
.
2) To ve
rify the validity of re
ceived
dat
a, U
i
comp
utes e
(
T
j,i
, r
i
T
j
) and comp
are
s
it
with
e(Q
i
, Q
j
). If they are not t
he same val
ue, U
i
stop
s the
session. Otherwise,
U
i
is
su
re that
the
received m
e
ssag
es
are f
r
o
m
U
j
. Similarl
y, U
j
comp
ut
e
s
e(
T
i,j
, r
j
T
i
) and comp
are
s
with e(Q
i
, Q
j
).If
they are not same, U
j
sto
p
s the se
ssio
n. Otherwi
se,
U
j
is su
re that the receiv
ed messa
g
e
s
are
valid.
3)
Up
on th
e recei
p
t of T
1,i
, T
2,i
, …, T
i-1,i
, T
i+1,i
, …, T
m,i
from
othe
r u
s
ers,
u
s
er U
i
compute
s
t
he se
cret
se
ssi
on k
e
y
:
K
i
= H
2
(e
(Q
i
+
)
S
r
,
T
m
i
j
,
1
j
i
i
i
,
j
)=
H
2
(e
)
r
...
r
r
(
m
2
1
)
P
,
P
(
)
(1)
Each u
s
e
r
pe
rform
s
the p
r
oce
dure ab
ove, t
hus all u
s
ers i
n
the g
r
o
up can get th
e sam
e
se
ssi
on k
e
y
as f
o
llow
s
:
K =
H
2
(e
)
r
...
r
r
(
m
2
1
)
P
,
P
(
)
(2)
At round
1, we assu
me tha
t
n users U
1
(1)
, U
2
(1)
, … , U
n
(1)
(n
≥
2) want to
sha
r
e a
co
mmon
se
ssi
on
se
c
r
et
key
.
E
a
ch
U
i
(1)
choo
se
s
a rand
om n
u
m
ber r
i
(1)
as his eph
eme
r
a
l
private
key. We
take a
n
integ
e
r nu
mbe
r
m
as the
ba
sed
numbe
r for group
s divisi
on
and p
a
rtition t
he n u
s
e
r
s int
o
m
n
sub
g
ro
up
s,
f
o
r n
≥
m
≥
2. T
h
e
sub
g
ro
up j, for j=1, 2, …,
m
n
-
1, ha
s m u
s
ers a
nd
co
mpute
s
the commo
n se
ssi
on su
b-key K
1j
= H
2
(e
m
1
i
)
1
(
ji
r
)
P
,
P
(
)
by the propo
sed proto
c
ol. T
he last su
bgroup
m
n
ha
s n (mod
m) u
s
ers. If the value of n (m
od m) is not equal to one, use
r
s in the last
sub
g
ro
up al
so use protocol to gene
rat
e
the commo
n se
ssion
su
b-key K
1j
, wh
ere j
=
m
n
. If
the
value of
n
(m
od m
)
i
s
equ
al to o
ne, it
mean
s th
e la
st subg
ro
up
only contain
s
one
u
s
e
r
, an
d we
take this u
s
e
r
’s eph
eme
r
al
private key a
s
the last sub
-
key.
At the next r
ound, ea
ch subgroup
U
j
(2)
, for j=
1, 2, …,
m
n
, tak
e
s
K
1j
as hi
s ephem
eral
private keys
r
j
(2)
re
sp
ectiv
e
ly and
broa
dca
s
ts U
a
(1)
’s publi
c
key a
s
the
subg
ro
up p
ublic val
ue,
here
U
a
(1)
is
a mem
b
e
r
of
su
bg
rou
p
U
j
(2)
and
a
≡
1(m
od
m
)
. We p
a
rtition
th
ese
m
n
sub
g
ro
up
s
into
2
m
n
subg
rou
p
s
a
nd
use the
sam
e
p
r
o
c
e
d
u
re
as the
ro
und
1. The
fo
llowing
roun
d
s
work
as
above. And the proto
c
ol d
oes n
o
t stop
until the num
ber of subg
ro
ups i
s
one.
4. Securit
y
A
n
aly
s
is
4.1. Implicit
Ke
y
Authentication
Implicit key a
u
thentication
to a use
r
A implie
s that o
n
ly the use
r
s
with wh
om A want
s to
agre
e
u
pon
a
comm
on
key
may be
able
to com
pute a
particula
r key. This i
s
a
n
id
eal p
r
ope
rty for
se
cure g
r
ou
p com
m
uni
cation si
nce it gets rid
of the nee
d for a
sep
a
rat
e
authe
nticat
ion
mech
ani
sm
key sh
arin
g a
nd
can
with
stand th
e ma
n
-
in-th
e
-mi
ddle
attacks. In
o
u
r p
r
oto
c
ol th
e
se
cret
se
ssio
n key is
co
m
puted
by ea
ch u
s
er
’
s
l
ong
-term private
key and ep
heme
r
al
p
r
iv
ate
key. Therefore, in any r
un of the protoco
l
, each user a
s
sure
s t
hat n
o
other pa
rtn
e
r exce
pt for the
intende
d on
e
who
ha
s hi
s o
w
n lo
ng-te
rm
private
key a
nd the valu
e
of ephe
meral
private
key can
learn the
se
cure sessio
n key. If an attacker want
s to
imperson
a
te
all other use
r
s to the user U
i
(1
≤
i
≤
m
)
, the
attacker ju
st
sele
cts (m
-1
) eph
eme
r
al
p
r
ivate
keys r
j
’ (
1
≤
j
≤
m,
j
≠
i
)
and se
nd
s
T
j.i
’= r
i
-
1
r
j
’Q
i
(1
≤
j
≤
m, j
≠
i) to the user U
i
. Figure 1
sho
w
s the imperson
a
tion a
ttack.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 10, Octobe
r 2014: 747
1
– 7477
7474
Figure 1. Impersonatio
n Attack o
n
Proto
c
ol O
R
-AGK
A
However, in
Figure 1, the
attacker M
a
lice c
annot co
mpute the final secret se
ssion key
without the kn
owledge of other us
ers’ long-term private keys.
4.2. Ke
y
-
Compromise Impersona
tio
n
Key-com
p
ro
mise
impe
rso
nation state
s
that
the atta
cker
who
ha
s
comp
romi
se
d
the lo
ng-
term private
key of use
r
A
can not o
n
ly imperso
n
a
te
A but also imperso
nate the othe
r users to
A. In our p
r
ot
ocol
O
R
-AGK
A, sup
p
o
s
e t
hat an
adve
r
sary
ha
s
got
the lon
g
-te
r
m
private
key o
f
a
certai
n u
s
e
r
U
i
(1
≤
i
≤
m), h
e
ca
n imp
e
rsonate
U
i
. Ho
wever, if h
e
want
s to ma
squerade
a
s
u
s
er U
j
(1
≤
j
≤
m
)
, he
can choo
se
an
ephem
eral p
r
ivate key
r
j
’
and sen
d
T
j’,i
to
U
i
. This attack
is
s
h
own i
n
Figure 2.
Figure 2. Key-co
mpromi
se
Imperson
a
tio
n
Attack on O
R
-AGKA
But without
U
i
’s ephe
meral
private ke
y
r
i
, the adv
ersary
can
n
o
t
comp
uter t
he K
i
.
Mean
while, upon re
ceivi
ng
T
t, j
(1
≤
t
≤
m, t
≠
j)
from
other partners,
the adversary
still cannot
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Perfect Fo
rward Se
cure ID-b
ased Ke
y Agreem
ent
Protocol in Gro
up… (Pe
n
g
s
h
uai Qiao
)
7475
comp
uter K
j
without
u
s
e
r
U
j
’s l
ong
-term private
key
S
j
and ep
he
meral
private
key r
j
. Th
ere
f
ore
even the adversary has
got the
long-term
private key
of a
certain
user,
he still cannot
imperso
nate
as othe
r users.
4.3. Perfec
t For
w
a
r
d
Sec
r
ec
y
If the long-te
rm private
keys of so
me
par
tici
pant
s are com
p
rom
i
sed,
the se
crecy
of
previou
s
se
ssion
key
s
sh
ould not b
e
a
ffected. A
nd
we
say that a proto
c
ol
ha
s pa
rtial forward
se
cre
c
y if co
mpromi
se
of the long
-term keys of
o
ne o
r
more but no
t all the participants d
o
e
s
n
o
t
comp
romi
se previou
s
ly
e
s
tablish
ed se
ssion ke
y
s
, a
nd we say that a protocol ha
s pe
rfe
c
t
forwa
r
d
se
crecy if com
p
romise
of the
long-te
rm
p
r
ivate keys of
all the parti
cipa
nt doe
s
not
comp
romi
se
any sessio
n key previo
usl
y
estab
lish
e
d
by these participa
nts. And KGC forwa
r
d
se
cre
c
y is
an
other
se
cu
rity issue. If at any run
the K
G
C’
s private
key is
com
p
romise
d, it do
es
not com
p
rom
i
se the
previo
usly e
s
tabli
s
h
ed sessio
n keys [13]. In o
u
r p
r
oto
c
ol, the comp
romi
se
of the entire
partne
r
s’ lo
ng-te
rm p
r
iva
t
e key
or K
G
C’
s private
key give
s n
o
help
about
the
se
ssi
on key, sin
c
e the
se
ssion
key is
co
mputed n
o
t
o
n
ly from long
-term key but also from u
s
e
r
s’
ephem
eral private keys. By this feature
,
our pr
oto
c
ol
can provide perfe
ct forwa
r
d se
crecy a
n
d
K
G
C f
o
r
w
a
r
d
se
cre
c
y
.
4.4. Kno
w
n
Session Ke
y
Securit
y
Each run of the proto
c
ol
should result in
a unique
secret se
ssi
on
key. The co
mpromi
se
of one
se
ssio
n key
sh
ould
not comp
ro
mise
other
session
key
s
and the
kn
owledge
of previous
se
ssi
on
keys do not all
o
w dedu
ction
of future
se
ssi
on keys. Because in o
u
r p
r
otocol, the fina
l
se
ssi
on
key comp
ri
se
s e
v
ery partn
er’
s
ep
heme
r
al
private key r
i
(1
≤
i
≤
m
)
, it is uniqu
e. It’s
impossibl
e for adversa
ry to com
pute th
e cu
rre
nt
se
ssion
key fro
m
the com
p
romise
d sessi
on
key
s
.
4.5. Unkno
w
n Ke
y
-
share
d
Resis
t
anc
e
Un
kno
w
n ke
y-sha
r
e
d
me
ans user
A share
s
a
key
with a diffe
re
nt party u
s
er C than
intende
d u
s
e
r
B a
nd A
d
oes not
kn
o
w
it. To
ou
r
proto
c
ol, at l
east it i
s
req
u
ired
to
kno
w
two
use
r
s’ l
ong
-te
r
m private
ke
ys to initial u
n
kn
own
k
e
y
-
s
h
ared
attack. However, it’s
diffic
u
lt for t
h
e
adversa
ry ex
cept fo
r KG
C to have
mo
re than
tw
o
u
s
ers’ l
ong
-term private
ke
ys at the
sa
me
time.
4.6. No Key
Con
t
rol
No
key
co
ntrol me
ans n
o
any p
a
rti
c
ip
ant in
th
e
group
ca
n influ
ence a
nd
co
ntrol th
e
outcom
e
of the se
cret session key. Beca
use eve
r
y
run
of our proto
c
ol, the secret se
ssi
on key is
determi
ned b
y
all users in the gro
up, an
d no one
can
control or p
r
e
-
dete
r
mine th
e se
ssi
on key.
5. Efficienc
y
Analy
s
is
As illustrated in Barua et
al.’s literature
[6], the efficiency of AGKA protocols mainly
involves the
communi
catio
n
and
com
put
ation cost
s.
In ea
ch roun
d
,
a use
r
may
have to tran
smit
data to so
me
or all the oth
e
r u
s
ers. Add
i
tionally
, each
use
r
ha
s to
perfo
rm som
e
ope
ration
s l
i
ke
scalar m
u
ltipli
cation
s, pairi
ng com
putati
ons. Co
mmu
nicatio
n
overhead i
s
affect
ed by the nu
mber
of roun
ds, total grou
p ele
m
ent se
nt, total me
ssage
s exchan
ged.
And com
puta
t
ion co
sts in
cl
ude
total of pairin
g
comp
utatio
n, total of sca
lar multipli
cati
ons.
In this
s
e
c
t
ion, we us
e notations
as
follows
:
-
R(n): Th
e total numbe
r o
f
round
s for n
users
-
S
i
: The nu
mber of scal
a
r
multiplicatio
ns in ro
und i
-
P
i
: The nu
mber of pai
rin
g
-comp
u
tatio
n
s in ro
und i
-
B
i
: The nu
mber of me
ssage
s tran
smit
ted in roun
d i
In proto
c
ol
OR-AGKA, m use
r
s
can
finish key
agre
e
me
nt in one roun
d
,
and the
effic
i
enc
y
is
:
S
1
= m
2
, P
1
=3
m,
B
1
=
1
1
m
1
m
C
C
2
. Fo
r
n u
s
ers to
ge
nerate
a
co
m
m
on
se
ssi
on
key, in
roun
d i, if
we
take
an i
n
teg
e
r
numb
e
r
m
as th
e b
a
sed
numbe
r fo
r g
r
oup
s divi
sion,
N
i
su
bgroup
s
will be divided into
m
N
i
new
subgroups, R(n) will be
n
log
m
(2
≤
m
≤
n).
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 12, No. 10, Octobe
r 2014: 747
1
– 7477
7476
In
the ca
se
s whi
c
h have
t
he sam
e
tota
l
num
ber
of round
s, whe
n
every
roun
d N
i
(mod
m)
=
0, the cost for gener
ating the common session key will be maximum.
The com
putati
onal
overhe
ad of
prop
osed p
r
o
t
ocol i
s
sum
m
ari
z
ed
and
co
mp
ared
with other
proto
c
ol
s in Ta
ble
1. As
sho
w
n
in
Ta
ble 1,
even
i
f
our p
r
oto
c
o
l
nee
ds
mo
re roun
ds, it i
s
p
o
ssibl
e
to
provide l
o
wer
comp
utation
co
st if we cho
o
se a
n
app
ro
priate m.
Table 1. Co
m
pari
s
on
with other Proto
c
o
l
s (2
≤
m
≤
n)
R: Total No.
of round
s.
S: Total No. of scalar multi
p
lications.
P: Total No. of pair-comp
utations.
B: Total No. of transmitted
messag
es.
6. Conclusio
n
In this pa
per,
a
se
cu
re,
efficient a
n
d
flex
ible ID-ba
s
e
d
on
e
rou
nd
authenti
c
ated
grou
p
key ag
reem
e
n
t proto
c
ol u
s
ing bili
nea
r
pairin
g
s i
s
propo
sed. Th
e prop
osed p
r
o
t
ocol focuses on
roun
d, mutua
l
authenti
c
ati
on, ban
dwidt
h
efficien
cy
a
nd provide
s
perfe
ct forwa
r
d secre
c
y. After
se
curity an
al
ysis a
nd p
e
rf
orma
nce an
al
ysis, it sh
ows that the prop
ose
d
sch
e
me
provide
s
stro
ng
se
curity and
lower
comp
utation co
st than prev
io
u
s
ly kno
w
n A
G
KA protoco
l
s. In the future
scope, the
compa
r
ison of
schem
es
wa
s given
by
two
que
stion
s
: whi
c
h schem
e
is suitabl
e for
different scenario and to whi
c
h degree these schemes will impa
ct the systems’ perform
a
nce
con
s
um
ption.
Referen
ces
[1]
W Diffie, M Hellman. Ne
w
d
i
re
ctions in cr
ypto
grap
h
y
.
IEEE Transactions on
Information Theory
. 197
6;
22(6): 64
4-6
4
5
.
[2]
A Shamir. Id
en
tit
y
-bas
ed cr
yp
tos
y
stem
an
d
sign
ature sc
he
mes.
Lecture
Notes i
n
C
o
mp
uter Scie
nce
.
198
4; 196: 47-
53.
[3]
A Jou
x
. A
on
e
roun
d pr
otocol
for tripartite
Di
ffie-Hellm
an.
L
e
cture N
o
tes i
n
Co
mputer Sc
ienc
e
. 20
00
;
183
8: 385-
394.
[4]
D Bone
h, M Frankli
n. Identit
y-b
a
se
d encr
y
ption from the
W
e
il pari
ng.
Lecture N
o
tes
in Co
mp
ute
r
Scienc
e.
2001
; 2139: 21
3~
22
9.
[5]
K Ch
oi, J
H
w
a
ng, D
L
ee. Effi
cient ID-
bas
ed
grou
p k
e
y a
g
re
ement
w
i
th
bi
li
near
maps.
Le
cture N
o
tes
i
n
Co
mp
uter Scie
nce
. 200
4; 294
7: 130-1
44.
[6]
X Du, Y
W
a
n
g
, J Ge, Y
W
ang. ID-
bas
ed
Auth
e
n
ticate
d
T
w
o Rou
n
d
Multi-Part
y
K
e
y Agre
eme
n
t.
Cryptol
ogy ePr
i
nt Archive
. 20
03; Rep
o
rt 200
3/247.
[7]
M Burmester,
Y Desmedt. A se
cure
and
effi
cient co
nfere
n
c
e
ke
y d
i
stributi
on s
y
stem.
Le
cture Notes
in
Co
mp
uter Scie
nce
. 199
5; 95
0: 275-2
86.
[8]
F
G
Z
hang, XF
Chen. Attack on T
w
o ID-
base
d
Authe
n
t
i
cated Gro
up
Ke
y
A
g
re
eme
n
t Schemes.
Cryptol
ogy ePr
i
nt Archive
. 20
03; Rep
o
rt 200
3/259.
[9]
K
y
u
ng-A
h
SHI
M
. F
u
rther Anal
ysis of ID-Ba
s
ed Auth
entica
t
ed Group Ke
y Agreeme
n
t Protocol fro
m
Bilin
ear Ma
ps.
IEICE TRANS
A
CTIONS.
2007; E90-A(1); 29
5-29
8.
[10]
CH Lin, HH
Li
n,
JH Ch
ang.
Multip
art
y
Ke
y Agre
eme
n
t fo
r Secur
e
T
e
lec
onfere
n
cin
g
.
S
ystem
s. M
a
n
and Cy
bern
e
tic
(SMC)
.
T
a
ipei. 2006; 5: 37
02-
370
7.
[11]
Jian Z
h
ou
an
d
Xian
w
e
i
Z
h
o
u
. Ke
y Agr
eeme
n
t Procoto
l
in
DSN.
T
E
LKOMNIKA Indo
ne
sian J
our
nal
of
Electrical E
ngi
neer
ing
. 2
013;
11(2): 80
9-8
1
8
.
[12]
R Baru
a, R D
u
tta, P Sarkar. Exte
ndi
ng J
o
ux’s pr
otocol t
o
multi p
a
rt
y k
e
y
agr
eeme
n
t.
Cryptogr
aphy
ePrint Archiv
e
. 200
3; Repr
ot 2003/0
62.
[13]
L Ch
en,
C Ku
dla.
Id
entity b
a
s
ed
authe
ntica
t
ed key
agr
ee
me
nt pr
otocols
from pa
irin
g
. Procee
din
g
of
16th IEEE Sec
u
rit
y
Fou
n
d
a
tio
n
s Wor
kshop.
Califor
ni
a. 200
3: 219-2
33.
Schemes R
S
P
B
[12]
n
log
3
< 5(n
-
1)
≤
9(n
-
1)
≤
5n
n
log
3
+3
[20] 1
n
2
n
n(n
-
1)
[6] 2
n(n+5
4n 3(n-1
[10]
2
n
2n
2n
[19] 1
n
0
2n
EO
R-A
G
KA
n
log
m
≤
(n
-
1)
m
2
/ (m
-
1)
≤
3(n
-
1)
m
.
/ (m
-
1)
≤
2m(n
-
1)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Perfect Fo
rward Se
cure ID-b
ased Ke
y Agreem
ent
Protocol in Gro
up… (Pe
n
g
s
h
uai Qiao
)
7477
[14]
Li
Xi
eh
ua, W
a
ng Y
ong
ju
n. Securit
y
En
ha
nc
ed A
u
thentic
ati
on a
n
d
Ke
y A
g
reem
ent
Prot
ocol in Ne
xt
Generati
o
n
Mo
bile
N
e
t
w
ork.
I
n
ternati
o
n
a
l J
o
urna
l of
Adva
n
c
ements
in
Co
mp
utin
g T
e
c
h
n
o
lo
gy
. 2
012;
4(3): 215-
22
2.
[15]
Lipi
ng Zha
ng, Guilin
g Li, Co
n
g
Xi
on
g, Shao-
Hui Z
hu. A Pai
r
ing-fre
e
Identit
y-b
a
se
d Authe
n
ticated Ke
y
Agreem
ent Pr
otocol for W
i
r
e
less a
nd Mo
bile
Net
w
orks.
Internatio
nal
Journ
a
l of Ad
vance
m
ents in
Co
mp
uting T
e
c
hno
logy
. 2
012;
4(5): 287-2
94.
[16]
Bin Hao, Yu
Yang, Sho
u
sh
an
Lu
o, Yixi
an
Yang, F
uqia
n
g
Liu.
An Authentic
ated Cl
u
s
tering-
base
d
Group Ke
y
Ag
reeme
n
t for La
rge Ad
Hoc
N
e
t
w
o
r
ks.
Adva
nces i
n
Infor
m
ation Sc
ie
nces
and S
e
rvic
e
Scienc
es
. 201
2; 4(7): 281-2
9
1
.
[17]
Z
i
y
i
Yo
u an
d Xi
ao
ya
o
Xi
e. A Novel Grou
p Ke
y
A
g
reem
ent Protocol f
o
r W
i
reless M
e
sh Net
w
o
r
k.
Journ
a
l of Co
n
v
erge
nce Infor
m
ati
on T
e
ch
no
logy
. 20
11; 6(2
)
: 86-101.
[18]
Abde
l Alim Ka
mal. Cr
yptan
a
l
y
sis of a Pol
y
n
o
mial-
bas
ed K
e
y
Ma
na
geme
n
t Scheme for Secure Grou
p
C
o
mmu
ni
ca
tion
.
Internatio
nal
Journa
l of Netw
ork Security
. 201
3; 15(1): 68
-70.
[19]
Marimuthu Rajaram, T
h
ilagav
a
th
y Dor
a
ira
j
Suresh. A
n
Interval-b
ase
d
C
ontrib
u
tor
y
Ke
y A
g
reem
ent.
Internatio
na
l Journ
a
l of Netw
ork Security
, 2
011; 13( 2): 92
-97.
[20]
Yiju
an S
h
i, Go
ngli
a
n
g
C
h
e
n
, Jia
nhu
a
Li. I
D
-bas
ed on
e
r
oun
d authe
ntic
ated gro
up
k
e
y agr
eeme
n
t
protoco
l
w
i
t
h
b
i
line
a
r
pair
i
ngs.
Informatio
n
T
e
chno
log
y
:
Co
di
ng
and
Com
p
u
t
ing (IT
CC). 20
05; 1:
75
7 -
761.
[21]
S Blake-W
ilso
n
, D Johnso
n
, A Menezes. Ke
y
A
g
reem
e
n
t Protocols a
nd their Sec
u
r
i
t
y
An
al
ysis
.
Lecture N
o
tes i
n
Co
mp
uter Scienc
e.
1997; 1
355: 30-
45.
[22]
A Menez
es, P
van Oorsc
hot, S Vansto
ne. H
and
bo
ok
of Ap
plie
d Cr
ypt
ogr
aph
y. F
i
fth
Edit
ion. F
l
or
id
a
:
CRC Press. 20
01.
Evaluation Warning : The document was created with Spire.PDF for Python.