IOSR Journals

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

IOSR Journal of Computer Engineering (IOSRJCE)

ISSN: 2278-0661, ISBN: 2278-8727 Volume 5, Issue 2 (Sep-Oct. 2012), PP 01-12


www.iosrjournals.org
www.iosrjournals.org 1 | P a g e

AACT: Anonymous and Accountable communication topology
for Wireless Mesh Networks

M.Narasimha Rao
1
, Shaik Jaffar
2

1
(M.Narasimha Rao currently doing his M.Tech in VLSI, Dept of E.C.E at Madina Engineering College,
Kadapa, AP, India
2
(Shaik Jaffar is currently working as Associate professor and HOD, Dept Of EIE, Madina Engineering
College, Kadapa, AP, India

ABSTRACT: Recently, multihop wireless mesh networks (WMNs) have involved increasing attention and
deployment as a low-cost move toward to give broadband Internet access at an urban scale. Security and
privacy issues are of the major concern in pushing the success of WMNs for their broad deployment and for
behind service-oriented applications. Regardless of the required, partial security research has been conducted
towards privacy protection in WMNs. This motivates us to develop Anonymous and Accountable communication
topology (AACT), a novel secure communication framework, tailored for WMNs. On one hand, AACT
implements harsh user access control to cope with both free riders and spiteful users. On the other hand, AACT
offers complicated user privacy protection beside both adversaries and a range of other network entities. AACT
is accessible as a suite of authentication and key agreement protocols built upon our AACT. Our analysis
demonstrates that AACT is resilient to a number of security and privacy related attacks. Additional methods
were also discussed to further improve scheme efficiency.
Keywords: Wireless Mesh network, Anonimity, Onion ring cryptography, user security , user accountability

I. INTRODUCTION
Wireless mesh networks (WMNs) have recently concerned rising attention and deployment as a
promising low-cost approach to give last-mile high speed Internet access at urban scale [2], [3]. Typically, a
WMN is a multihop layered wireless. The first layer consists of access points, which are high-speed wired
Internet way in points. In the second layer, stationary mesh routers form a multihop spine via long-range high-
speed wireless methods such as WiMAX [6]. The wireless spine connects to wired access points at some mesh
routers through high speed wireless links. The third layer consists of a huge number of mobile network users.
These network users access the network either by a direct wireless link or through a chain of other peer users to
a nearby mesh router. WMNs correspond to a unique marriage of the ubiquitous coverage of large area cellular
networks with the ease and the speed of the local area Wi-Fi networks [4]. The compensation of WMNs also
contains low deployment costs, self-configuration and self maintenance, good scalability, high robustness, etc.
[2].Security and privacy issues are of mainly a concern in pushing the success of WMNs for their large
deployment and for supporting service-oriented applications. Due to the essentially open and distributed nature
of WMNs, it is necessary to enforce network access control to cope with both free riders and spiteful attackers.
Dynamic access to WMNs should be subject to successful user authentication based on the correctly pre
recognized trust among users and the network operator; otherwise, network access should be forbidden. On the
other hand, it is also dangerous to provide good provisioning over user privacy as WMN communications
regularly contain a vast amount of sensitive user details. The wireless standard, open network structural design,
and be lacking in of physical protection over mesh routers render WMNs extremely vulnerable to different
privacy-oriented attacks. These attacks range from passive eavesdropping to active message Phishing,
interception, and modification, which could simply lead to the leakage of user information. Obviously, the wide
deployment of WMNs can succeed only after users are assured for their capability to manage privacy risks and
preserve their desired level of anonymity. Included with sensors and cameras, the WMN may also be used to
gather information of interest. Perceptibly, all these communications include different kinds of sensitive user
information like individual identities, actions, position information, fiscal information, transaction summaries,
social/business connections, and so on. Once disclosed to the attackers, this information could negotiation any
user's privacy, and when further associated together, can cause even more overwhelming consequences. Hence,
securing user privacy is of paramount practical importance in WMNs. Moreover, for both billing purpose and
avoiding the neglect of network resources, it is also necessary to exclude free riders and let only legitimate
residents access WMNs.
Despite the need and significance, limited research has been conducted to address security mechanisms
for anonymous and accountable communication in WMNs. This encourages us to propose AACT, a novel
Anonymous, Accountable communication topology for WMNs. Our assistance is fourfold as follows:
AACT: Anonymous and Accountable communication topology for Wireless Mesh Networks
www.iosrjournals.org 2 | P a g e
Security: It achieves explicit mutual authentication and key organization among users and mesh routers and
among users themselves. It, thus, excludes both illegitimate network access from free riders and spiteful users
and Phishing attacks due to rogue mesh routers.
Anonymity: It concurrently enables independent anonymous authentication among users and mesh routers and
bilateral anonymous authentication between any two users. It, thus, make sure user anonymity and privacy.
Accountability: It enables user accountability, at regulating user behaviors and defending WMNs from being
harmed and attacked. Network communications can always be audited in the cases of disagreements and
deceptions. It in addition allows adaptive user revocation so that spiteful users can be ejected.
Sophisticated user privacy: It allows users to disclose minimum information probabilistically while preserves
accountability. In AACT, the user characteristics is a comprehensive information as network users as society
members always interact with WMNs in different roles and contexts. Therefore, a dispute about a given
communication session should only be attributed according to the role/context information about the user
without disclosing his full identity information (unless necessary).
For the finest of our knowledge, AACT is the first attempt to set up an accountable security framework
with a complicated privacy protection model tailored for WMNs. AACT also lays a solid background for
designing other upper layer security and privacy solutions, e.g., Anonymous communication.
The rest of the paper is prearranged as follows: Section 2 is the introduction of the cryptographic
knowledge entailed by AACT. Section 3 describes the problem formulation. Then, in Section 4, the details of
AACT are described. We further analyze in Section 5 the security and privacy properties of AACT, as well as
its presentation. Section 6 is about related work. Finally, we conclude the paper in Section 7.

II. CRYPTOGRAPHY SPECIFICS
2.1 Onion ring strategy [31]
The Onion routing [31] achieves communication privacy by making communication ends as unable to
link. An Onion routing network consists of a number of interconnected Onion routers (ORs); each OR has a pair
of public/private keys. Each OR knows the topology of the Onion network as well as the public keys of other
ORs. An end user that requires an anonymous communication will send a request to an OR that it trusts; this OR
is known as the Onion Proxy (OP) for the user. The communication between an end user and its OP is protected
from the adversaries. The OP determines a route that consists of a series of ORs and constructs an "Onion" using
the public keys of the routers en route. The "Onion" is constructed in a way such that the most inner part is the
message to the intended destination. The message is wrapped, i.e., encrypted using the public keys of the ORs in
the route, in the same order as the ORs appears in the route. Once an OR receives the Onioned message, it uses
its private key to peel, i.e., decrypt, the "Onion", to obtain the information such as the next hop and the session
key. It then forwards the rest of the "Onion" to the next hop. This process is repeated until the "Onion" reaches
the last OR, which peels the last layer of the "Onion" and obtain the exit information, i.e., the destination.
For example, if the private route is
1 2
....
n
R R R , where
i
R is the
th
i OR , and the last router
n
R will
connect to the exit funnel of the ORs, which will further communicate with the address requested by the
session initiator; the message flow and the "Onion"(s) received at each router in the route are as follows:

( )
( ) ( )
( )
p 1 p 2 p n
k R 2 1 k R k R n
E R , k ,E ....E k , exit .... 1
( )
( )
( )
p 2 p n p n
k R k R n k R n
E ...E k , exit ... ... E k ,exit .
k R
p i
and
i
k are the public key and assigned session key for the
th
i router. After the route is built up,
session keys are used for constructing "Onion"s, and anonymous circuit ID (ACI) is used for routing. For the
reverse path, data packet was encrypted with the session keys. The OP receives the "Onion" in the reverse path
and peels it using the session keys it assigned to the ORs, and sends the raw data to the end user.
For an Onion route, only the proxy knows the rst and the last router. Any OR in the route only knows its
previous hop and next hop. For both outside attackers and inside attackers (i.e., compromised ORs), as
encryption or decryption is processed at every OR, it is difficult to link any two links (a link is a connection
between two Onion routers) to the same route. Therefore, for a communication going through the Onion routers,
the entry OR and exit OR are unable to link. When there are a large number of connections, it is difficult to nd
out the two communication ends for any connection that applies Onion routing.
To avoid that the change of "Onion" size in the route built-up stage may give adversary hints about
routing in- formation, an "Onion" has to be padded when part of its information has been read and removed, so
that the length of the "Onion" keeps the same and it is difficult for an inside observer to obtain the routing
information. Refer to [10], if the maximum number of Onion routers in a private route is N, the OP will
construct a message of N "Onions" to build an Onion route. When an router receives the "Onion"s, it decrypts
AACT: Anonymous and Accountable communication topology for Wireless Mesh Networks
www.iosrjournals.org 3 | P a g e
all the "Onion"s and obtain the routing information only from the rst one. It then adds a dummy packet at the
end, and forward the "Onion"s further.
For example, if the maximum hop count N is 5, and the private route is as
1 2 3
OP R R R , the message
flow and the messages sent at each router are as follows:
( ) ( )
( )
( )
p 1 p 1 p 2
1 k R 2 1 k R k R 3 2
OP R : E R , k ,E E R , k , 2
( )
( ) ( )
p 1 p 2 p 3
k R k R k R 3
E E E exit, k ,
dummy,dummy
( ) ( )
( )
( )
p 2 p 2 p 3
p 3
1 2 k R 3 2 k R k R 3
2 3 k R 3
R R : E R , k ,E E exit, k ,
dummy;dummy;dummy
R R : E exit, k ,
dummy;dummy;dummy;dummy


2.2 Group Signature
Group signature schemes are a comparatively recent cryptographic concept introduced by Chaum and
van Heyst in 1991 [9]. A group signature scheme is a technique for allowing a member of a group to sign a
message on behalf of the group. In contrast to ordinary signatures, it gives anonymity to the signer, i.e., A
verifier can only tell that a member of any group signed. However, in outstanding cases, such as a legal
argument, any group signature can be "opened" by a designated group manager to make known clearly the
identity of the signature's originator. Some group signature schemes support revocation, where group
membership can be disabled. One of the most recent group signature schemes is the one proposed by Boneh and
Shacham [8], which has an extremely short signature size that is similar to that of an RSA-1024 signature [10].
This scheme is based on the following two problems that are believed to be hard. Let
1 2
, G G ,
1 2
, g g as defined
above.
q-Strong Diffie-Hellman problem: The q-SDH problem in (
1 2
, G G ) is defined as follows: given a (q + 2)-tuple
2
( ) ( )
1 2 2 2 2
( , , , ,...,
q
g g g g g

as input, output a pair
1/( )
1
( , )
x
g x
+
, where
p
x Z
-
e .
Decision linear on
1
G : Given random generators u, v, h of
1
G and , , 1
a b c
u v h G e as input, output yes if a + b =
c, and no, otherwise.

III. PROBLEM FORMULATION AND THE SCHEME OVERVIEW
3.1 Network Architecture and System Assumptions
In the three-layer architecture consider a metropolitan-scale WMN under the manage of a network
operator (NO). The network operator deploys a number of APs and mesh routers and forms a well-connected
WMN that covers the whole area of a city and gives network services to network users, i.e., the citizens.
Network users, on the other hand, subscribe to the network operator for the services and use their mobile clients
to freely access the network from wherever within the city. The membership of network users may be 1)
completed/renewed according to user- operator agreement in an episodic manner or 2) dynamically revoked by
NO in case of argument/attack.
Similar to [4], [11], we assume that the downlink from a mesh router to all users within its reporting is
one hop. However, the uplink from a user to a mesh router may be one or several hops. That is, a network user
wants to transmit packets in multiple hops to a mesh router beyond his direct transmission range. In this case,
network users cooperate with each other on relaying the packets to mesh routers. We further assume that all the
network traffic has to go through a mesh router except the communication between two direct neighboring
users. We assume so as it is probable that communications to and from a mesh router will constitute the majority
of traffic in a WMN [12]. Moreover, this assumption would considerably reduce the routing complexity from
the users' point of view as mesh routers will take the responsibility.
We assume that NO can always communicate with mesh routers through pre recognized secure
channels, and so are mesh themselves. The WMN is assumed to be deployed with redundancy in mind so that
revocation of individual mesh routers will not affect network connection. We assume the survival of an offline
trusted third party (TTP), which is trusted for not disclosing the information it stores. TTP is essential only
during the system setup. We further assume that there is a secure channel among TTP and each network user.

AACT: Anonymous and Accountable communication topology for Wireless Mesh Networks
www.iosrjournals.org 4 | P a g e
3.2 Threat Model and Security Requirements
Due to the open medium and spatially distributed nature, WMNs are susceptible to both passive and
active attacks. The passive attacks include eavesdropping, while active attacks range from message relaying,
bogus message injection, Phishing, active imitation to mesh router cooperation. Hence, for a practical threat
model, we consider an adversary that is able to eavesdrop all network communications, as well as inject random
fake messages. In addition, the adversary can compromise and control a small number of users and mesh routers
subject to his option; it may also set up rogue mesh routers to phish user accesses. The purposes of the adversary
contain 1) illegal and unaccountable network access, 2) the privacy of genuine network users, and 3) denial-of-
service (DoS) attacks against service accessibility.

In light of the above threat model, the following security requirements are necessary to make sure that a WMN
functions correctly and strongly as purposed.
User-router shared authentication and key agreement: A mesh router and a user should equally
authenticate each other to stop both unauthorized network access and Phishing attacks. The user and the
mesh router should also set up a shared pairwise symmetric key for session authentication and message
encryption.
User-user mutual authentication and key agreement: Users should also authenticate each other by
cooperation in observing to message relaying and routing. Moreover, symmetric keys should be established
and efficiently maintained to give session authentication and message encryption over the equivalent traffic.
Sophisticated user privacy protection: The privacy of users should be well secluded, and we distinguish
user privacy against dissimilar entities such as the adversary, NO, and the law authority, as will be
complicated in the next section.
User accountability: In the cases of attacks and argument, the responsible users and/or user groups should
be capable to be audited and pinpointed. On the other hand, no innocent users can be framed for
disputes/attacks they are not concerned with.
Membership maintenance: The network should be capable to handle membership dynamics with
membership revocation, renewing, and addition.
DoS resilience: The WMN should maintain service accessibility despite of DoS attacks.

IV. AACT: ANONYMOUS, ACCOUNTABLE COMMUNICATION TOPOLOGY
When designing AACT, we find that none of the obtainable anonymous accountable cryptographic
primitives, such as blind signature and group signature schemes, suits our purpose given the security and
privacy requirements discussed above. Blind signature and group signature schemes can only give binding
anonymously, while AACT demands user accountability, and hence, revocable anonymity. Existing group
signature schemes do give revocable secrecy, but cannot support complicated user privacy. This inspiring us to
tailor a group signature scheme by combining with onion ring strategy to convene all the necessities. AACT is
then built on this onion ring based group signature difference by further integrating it into the authentication and
key agreement protocol design.

4.1 AACT Key Management
The following setup operations are performed in an offline manner by all the entities in AACT, namely
NO, a TTP, mesh routers, network users, and user group managers. AACT works under bilinear
groups
1 2
( , ) G G with isomorphism and respective generators
1
g and
2
g , as in Section 2.1. AACT also
employs hash functions
0
H and H, with respective ranges
2
2
G and
p
Z . The notation below mainly follows [8].
NO is responsible for the key generation operation. Specifically, NO proceeds as follows:
1. Select a generator
2
g in
2
G uniformly at random and set
1 2
( ). g g Select
p
RZ
-

and set
2
w g

= .
2. Select
*
i p
grp RZ


For a registered user group I.
3. Using , generate an SDH tuple
,
( , , )
i j i j
A grp x by selecting
*
j p
x RZ

such that 0
i j
grp x + + = , and
setting
1/( )
1
i j
grp x
ij
A g
+ +
.
4. Repeat Step 3 for a prearranged number of times that are mutually agreed by NO and the user group
manager
i
GM .
5. Send {[ , ], , ) }
i i j
GM i j grp x j via a secure channel.
6. Repeat Steps 2, 3, and 4 for every user group.
AACT: Anonymous and Accountable communication topology for Wireless Mesh Networks
www.iosrjournals.org 5 | P a g e
7. Send TTP:
,
{[ , ], ) , }
i i j j
GM i j A x i j via a secure channel, where 0 denotes bitwise exclusive OR
operation.
The above operation generates the group public key gpk and a number of private keys gsk:
1 2
,
( , , )
{ [ , ] ( , , ) , }.
i j i j
gpk g g w
gsk i j A grp x i j
=


Furthermore, NO obtains a set of revocation tokens, grt, with grt[i,j] =
, i j
A and also keep the mapping among
group id i and
i
grp for all user groups. Note that is the system secret only known to NO. For the purpose of
non denial, NO signs on Steps 5 and 7 under a standard digital signature scheme, such as ECDSA [13]. In
AACT, we suppose that ECDSA-160 is used. For the same purpose,
i
GM and TTP also sign on these messages
upon receiving and send the resulted signature back to NO.
Additionally, NO prepares every mesh router
k
MR a public/private key pair, denoted by ( , )
k k
RPK RSK . Each
mesh router also gets an accompanied public key
A certificate signed by NO to prove key authenticity. The signing key pair of NO is denoted by (NPK, NSK).
The certificate has the following fields at the minimum:
{ , , , },
k k k NSK
Cert MR RPK ExpT Sig =
Where ExpT is the expiration time and Sig, denotes an ECDSA-160 signature signed on a given message using a
private key .
Before accessing the WMN, a network user has to validate himself to his fit in user groups. For each such user
group i, a network user
j
uid is assigned a casual group private key as follows:
1.
i
GM sends ( , , , )
j i j
uid i j grp x as well as the related system parameters.
2.
i
GM requests TTP to send
,
( , , )
j i j j
uid i j A x by providing the index [i, j].
3.
j
uid assembles his group private key as
,
[ , ] ( , , )
i j i j
gsk i j A grp x = .
Note that in our setting,

i
GM only keeps the mapping of ( ( , , , ))
j i j
uid i j grp x but has no knowledge of the
corresponding
, i j
A .
NO only knows the mapping of ( , [ , ])
i
GM gsk i j but has no knowledge about to whom gsk [i, j] is
assigned.
TTP has the mapping of
,
( ( , ))
j i j j i
uid A x grp as it sends
j
uid this information through a safe channel
among the two upon the request from
i
GM . But TTP has no knowledge of the corresponding
j
x or
, i j
A .
Here, we use
j
uid the user's necessary attribute information. For the purpose of non repudiation,
j
uid signs on
the messages it receives from
i
GM and TTP under ECDSA-160, and sends back
i
GM the equivalent
signature.

4.2 User-Router Mutual Authentication and Key Agreement
To access the WMN, a network user follows the user-router common authentication and key agreement
protocol as particular below, when a mesh router is within his direct communication range.
1. The mesh router
k
MR first picks a random nonce
*
R p
r RZ

and a random generator g in


1
G and then
computes .
R
r
k
g MR further signs on g
R
r
g , and the current time stamp
1
ts , using ECDSA-160. MR
k
then
broadcasts
1
, , , , ,
R
k
r
RSK k
g g ts Sig Cert CRL URL (M.1)
As part of beacon message that is periodically broadcast to declare service existence. Here, CRL and
URL denote the mesh router certificate revocation list and the user revocation list, respectively. Specifically,
URL contains a set of revocation tokens that corresponds to the revoked group
private keys, which is a subset of grt. Both CRL and URL are signed by NO.
Upon receipt of (M.1), a network user uidj proceeds as follows:
AACT: Anonymous and Accountable communication topology for Wireless Mesh Networks
www.iosrjournals.org 6 | P a g e
Check the time stamp ts
1
to prevent replay attack. Examine
k
Cert to confirm public key authenticity and the
certificate expiration time; examine CRL and see if
k
Cert has been revoked by applying NPK. Further verify
the authenticity of
RSK
Sig by applying
k
RPK .
Upon positive check results,
j
uid believes that
k
MR is legitimate and does the following:
Pick two random nonce
*
,
j p
r r RZ

, compute
j
r
g , and prepare the current timestamp
2
ts . Further get two
generators ( , ) u v in
2
G from
0
H as
2
0 2 2
( , ) ( , , , ) ,
j
R
r
r
u v H gpk g g ts r G e (1)


And compute their images in
1
: ( ) ( ). G u u andv v
Compute
1 2 , i j
T u andT A v
o o
by selecting an exponent . ( )
p i j p
RZ Set grp x Z o o o + e

. Pick
blinding values ,
x
r r
o
, and
p
r RZ
o
.
Compute helper values
1 2
, R R , and
3
R :
1 2 2 2 2
, ( , ) . ( , ) . ( , ) ,
x
r r r r
R u R e T g e v w e v g
o o o

and
3 1
. .
x
r r
R T u
o

Compute a challenge
value
p
c Z e using H:
2 1 2 1 2 3 .
( , , , , , , , , , )
j
R
r
r
p
c H gpk g g ts r T T R R R Z e
Compute , ( )
x x i j
s r c s r c grp x
o o
o = + = + + and .
p
s r c Z
o o
o = + e Obtain the group signature
on
2
{ , , }
j
R
r
r
g g ts as
[ , ] 1 2
( , , , , , , ).
gsk i j x
SIG r T T c s s s
o o

Compute the shared symmetric key with
k
MR :
,
( )
j
R
r
r
k j
K g = .
Unicast back to
k
MR
2 [ , ]
, , ,
j
R
r
r
gsk i j
g g ts SIG . (M.2)
Upon receipt of (M.2),
k
MR carries out the following to authenticate
j
uid :
Check
R
r
g and
2
ts make sure the freshness of (M.2).
Check that
[ , ] gsk i j
SIG is a valid signature by applying the group public key gpk as follows:
Compute u and v using (1), and their images
u and v in
1
: ( ) ( ). G u u andv v
Retrieve
1 2
, R R and
3
R as:
1 1
/
s c
R u T
o


2 2 2 2 1 2
( , ) . ( , ) .( ( , ) / ( , )) ,
z
s s c
R e T g e v w e T w e g g
o


And
3 1
. .
z
s s
R T u
o


Check that the challenge c is correct:
2 1 2 1 2 3
? ( , , , , , , , , , ).
j
R
r
r
c H gpk g g ts r T T R R R
=

(2)
For each revocation token A eURL, check whether A is encoded in
1 2
( , ) T T by checking if
2 1
( / , )? ( , ). e T A u e T v
=
(3)
If no revocation token of the URL is encoded in
1 2
( , ) T T , then the signer of
[ , ] gsk i j
SIG has not been revoked.
If all the above checks succeed,
k
MR is now assured that the current user is a legitimate network user,
although
k
MR does not know which particular user this is. Note that
j
uid is never disclosed or transmitted
during protocol execution.
AACT: Anonymous and Accountable communication topology for Wireless Mesh Networks
www.iosrjournals.org 7 | P a g e
a.
k
MR Further computes the shared symmetric key as
,
( )
j
R
r
r
k j
K g = and sends back
j
uid :
,
, , ( , , ),
j j
R R
k j
r r
r r
K k
g g E MR g g (M.3)
Where E denotes the symmetric encryption of the given message within the brackets using key .
The above protocol allows explicit mutual authentication among a mesh router and a genuine network user; it
also enables unilateral anonymous authentication for the network user. Upon successful completion of the
protocol, the mesh router and the user also create a shared symmetric key used for the succeeding
communication session. And this session is uniquely identified through ( , )
j
R
r
r
g g .
Remarks
Equation (2) holds because
1 1 1
/ / ( ) .
s r c c
R u T u u u R
o o
o o o +
= = = =


( ) . . 2 2
2 2 2 2 2 2 2 2 2 2
1 2 1 2
2 2
2 2
1 2
( , ) ( , )
( , ) . ( , ) . ( , ) ( ( , ) . ( , ) . ( , ) ).( ( , ) . ( , ) . ( , ) .
( , ) ( , )
(
( , )
. .
( , )
i j i j z z
i j
c
grp x grp x s s r s s r
c
grp x
e T w e T w
R e T g e v w e v g e T g e v w e v g e T g e v w e v g c
e g g e g g
e A
e T v wg
R R
e g g
o o o o
o o
o
+ +
+
| |
= =
|
\ .
| |
= =
|
|
\ .

, 2
1 2
2 2
1 2 1 2
, )
( , )
. .
( , ) ( , )
i j
c
c grp x
i j
wg
e g g
R R
e g g e g g
+
| |
| |
= = |
|
|
\ .
\ .

( ) (
3 1 1 3
( ) . ( ) . . .
z i j i j
z z z
r c grp x r c grp x
s r r s r r
R T u u u u u T u R
o
o o o
o
o o
+ + +

= = = = =


Equation (3) holds when there is an element A of URL encoded in
1 2
( , ) T T because of the following.
We know that :
2 1
G G is an isomorphism such that
2 1
( ) . g g = According to the
definition of isomorphism, we have ( ) ( ) ( ) PQ P Q = for any P, Q
2
. G e Using this property and
mathematical induction, it is easy to know the following fact: For any natural number
2 1
, ( ) .
m m
m N g g e =
Hence, if a group private key
,
( , , )
i j i j
A grp x with
, i j
A URL e signed the group signatureo . For
simplicity, let
2 2

b
u g andv g
o
= = for some integers a and b. On one hand,
2 , , , 2 1 2 1 2
( / , ) ( / , ) ( , ) (( ( )) , ) (( ( )) , ) (( ) , ) ( , ) .
b b ab
i j i j i j
e T A u e A v A u e v u e v u e g u e g g e g g
o o o o o o o
= = = = = =

On the other hand,
1 2 1 2 1 2
( , ) ( , )) (( ( )) , ) (( ( )) , ) (( ) , ) ( , ) .
b ab
e T v e u v e u v e g v e g g e g g
o o o o o o o
= = = = =
Therefore,
2 , 1
( / , ) ( , ).
i j
e T A u e T v =


4.3 User-User Mutual Authentication and Key Agreement In AACT
Adjacent genuine network users may help to relay each others traffic. To this end, two network users
within each others direct communication range first authenticate each other and create shared secret pairwise
key as follows:
1.
j
uid picks a random nonce
*
j p
r RZ

and computes where


j
r
g is obtained from the inspirational
messages broadcasted by the current service mesh router.
j
uid further signs on ,
j
r
g g , and current time stamp
ts1, using his group private key gsk[i,j] following Steps 2b(i) to 2b(iv), as in Section 4.2.
j
uid Then locally
broadcasts
1 [ , ]
, , , .
j
r
gsk i j
g g ts SIG (M.1)
2. Upon receipt of ( .1) M

,
I
uid checks the time stamp and verifies the authenticity of
[ , ] gsk i j
SIG by
applying the group key gpk following Step 3b, as in Section 4.2.
I
uid further checks if the signature is
generated from a revoked group private key following Step 3c, as in Section 4.2. Note that URL can always be
obtained from the beacon messages.
If all checks succeed,
I
uid is assured that the current user it communicates with is legitimate.
I
uid proceeds to
pick a random nonce
*
I p
r RZ

and computes
I
r
g .
I
uid further signs on ,
j
I
r
r
g g , and current time stamp
2
ts ,
AACT: Anonymous and Accountable communication topology for Wireless Mesh Networks
www.iosrjournals.org 8 | P a g e
using an appropriate group private key gsk[t, I] of his.
I
uid also computes the shared pairwise session key
as
,
( ) .
j
I
j I
r
r
r r
K g = then replies
I
uid
2 [ , ].
, , ,
j
I
r
r
gsk t I
g g ts SIG ( .2) M


3. Upon receipt of ( .2) M

,
j
uid first delay window.
j
uid checks whether
2
ts -
1
ts is within the acceptable
delay window.
j
uid also examines
[ , ] gsk i j
SIG and URL as
j
uid did above. If all checks succeed,
j
uid is also
assured that its communicating counterpart is legitimate.
j
uid Computes the shared pairwise session key
as
,
( )
j
I
j I
r
r
r r
K g = .
j
uid Finally replies
I
uid
,
1 2
, , ( , , , ).
j
I I I
r r
j I
r
r r r
K
g g E g g ts ts = ( .3) M


Upon receipt of ( .3) M

and successful decryption of


,
1 2
( , , , ).
I I
r r
j I
r r
K
E g g ts ts =
I
uid is assured that
j
uid has
successfully completed the authentication protocol and recognized the shared key for their subsequent
communication session, which is uniquely identified through ( , )
j
I
r
r
g g .
This design of AACT protects user privacy in a complicated manner, while still maintaining user accountability.
User Anonymity against the Adversary, the User Groups, and TTP
In AACT, a user only authenticates himself as a genuine service subscriber without disclosing any of
his identifying information by make use of the group signature method. Neither the adversary nor the user group
managers can tell which meticulous user generates a given signature. The adversary, even by compromising
mesh routers and other network users, that is, knowing a number of group private keys in addition to the group
public key, still cannot infer any information concerning the meticulous group private key used for signature
generation. This is due to the rigidity of the underlying q-SDH problem, where q is a 1,020-bit prime number.
Due to the similar reason, neither a user group manager can distinguish whether or not one of his group
members has signed a meticulous signature as he has no knowledge of the corresponding
, i j
A s nor can he
compute them. The same termination also holds for TTP as TTP can compute neither
j
x
nor
, i j
A given
, i j j
A x . Furthermore, each data session in AACT is identified only through pairs of fresh
random numbers, which again discloses nothing concerning the user identity information. In addition, AACT
needs a network user to refresh session identifiers and the shared symmetric keys for each different session.
This further eliminates the ability to link among any two sessions initiated by the same network user. We note
that even with the help of compromised mesh routers and other network users, the opponent still cannot judge
whether two communication sessions are from the similar user. This is because, basically, none of them can tell
whether two signatures are from the same user, given q- SDH problem and decision linear on G problem are
hard.
User Privacy against NO and User Accountability:Since NO knows grt, it can always tell which gsk[i,
j] produces a given signature. However, NO has no knowledge about to whom gsk[i, j] is assigned as AACT
allows a late compulsory among group private keys and network users. Furthermore, it is user group managers'
sole responsibility to assign group private keys to every network user without any participation of NO.
Therefore, NO could only map gsk[i, j] to the user group i based on
i
grp . Because no other entities except NO
and the key holder himself has the knowledge of the corresponding
, i j
A , and can therefore, generate the given
signature, the key holder must be a member of the user group i. This audit result serves us both necessities. On
one hand, the result only discloses partial nonessential attribute information of the user and still protects user
privacy to an extent. On the other hand, the result is adequate for user accountability purposes for NO.
When NO (on behalf of mesh routers) finds a certain communication session disputable or suspicion, it
conducts the following protocol to audit the responsible entity:
1. Given the link and the session identifier, find the equivalent authentication session
message
2 [ , ]
( .2) , , ,
j
R
r
r
gsk i j
M g g ts SIG = from the network log file.
2. For each revocation token
, i j
A grt e , check whether
2 , 1
( / , )? ( , )
i j
e T A u e T v
=
. Output the first element
, i j
A grt e such that
2 , 1
( / , )? ( , )
i j
e T A u e T v
=
.
3. For the found revocation token
, i j
A , output the corresponding mapping between
, i j
A and
i
grp .
AACT: Anonymous and Accountable communication topology for Wireless Mesh Networks
www.iosrjournals.org 9 | P a g e
Since
i
grp maps to a particular user group i, now a responsible entity has been found from the perspective of
NO.
From the user's perspective, only part of his unneeded attribute information is disclosed from the audit.
But such unneeded attribute information will not reveal his necessary attribute information. For example, the
above audit may find that the dependable user is a member of Company XYZ but cannot reveal any other
information about the user. Yet NO still has adequate proof to prove to Company XYZ that one of his members
violates certain network access rule so that Company XYZ should take the corresponding responsibility
specified in their service contribution agreement.
Revocable User Anonymity against Law Authority: When law authority decides to track the meticulous
attacker that is responsible for a certain communication session, the following procedure is taken: NO reports to
the law authority
,
( , )
i j i
A grp by executing the above protocol against the session in audit.
,
( , )
i j i
A grp is then
further forwarded to
i
GM .
i
GM Checks its local record, finds out the mapping between ( )
i i
grp andx , and
hence, the corresponding user uniqueness information
j
uid , to whom gsk[i,j] is assigned during the system
setup.
i
GM then replies
j
uid to the law authority. At this point, law authority and only law authority get to
know about which particular user is conscientious for the communication session in the audit. We point out that
this tracing procedure has the non denial property because 1)
i
GM signed on all gsks that are assigned from NO
as the proof of receipt; 2)
j
uid also signed on the messages when obtaining gsk[i, j] from
i
GM and TTP as the
proof of receipt. AACT also not able to frame because no one else knows gsk[i, j] except NO and
j
uid or is
able to forge a signature on behalf of
j
uid .

V. PERFORMANCE ANALYSIS OF AACT
5.1 System Security Analysis
As its basic security functionality, AACT enforces network access control. Hence, we are the majority
concerned with the following three different types of attacks, i.e., Bogus data injection attacks, data Phishing
attacks, and DoS attacks.
Bogus data injection attacks: In such attacks, the opponent needs to inject bogus data to the WMN
aimed at using the network service for free. The sources of the bogus data could be outsiders, revoked users, or
revoked mesh routers.
However, such bogus data traffic will be all instantly filtered in AACT. First, with respect to outsiders,
they do not know any group private keys. Thus, they cannot produce correct message signatures, when
attempting to initialize a communication session with NO and/or other network users. They also cannot bypass
the authentication procedure and straightly send out bogus data to others as they do not possess any shared
symmetric session keys with them, and thus, cannot produce correct MACs. Then, regarding revoked users,
there are two situations: 1) they do not have any group private key at present in use due to group public key
update or 2) the corresponding group private keys owned by them are previously revoked and are published in
the URL in beacon messages. Obviously, the revoked users cannot increase network access in neither cases.
Finally, for revoking mesh routers, they are no longer valid members of the WMN. By checking CRL, no
genuine mesh routers will accept/relay data traffic from revoking mesh routers. Also, since the downlink from a
mesh router to its service range is only one hop, network users never require to and will not relay data traffic for
mesh routers in AACT.
Data phishing attacks: In such attacks, the opponent may set up bogus mesh routers and try to phish
user connections to such routers. In this way, the opponent could control network connection and analyze users'
data traffic for their benefits. The Phishing mesh routers can be either completely new mesh routers or revoked
mesh routers both at the adversary's control. In the former case, the mesh router will not be capable to
authenticate itself to the network user. Therefore, no network user will set up any session with such a mesh
router. Even if the mesh router could stop the network traffic among a network user and a genuine mesh router,
it will not be able to decrypt the message and obtain any useful information. In the latter case, a newly revoked
mesh router, however, will possibly be capable to authenticate itself to a network user, if such a user does not
possess the most recent version of CRL. The network user may be deceived in this case but only for up to
(inverse of the update frequency(current timelast periodically update time)) time period. This is because the
revoked mesh router will not be capable to give a legal CRL update at the next periodical CRL update time
point.
DoS attacks: In such attacks, the opponent may flood a huge number of illegal access request messages
to mesh routers. The purpose is to exhaust their resources and render them less capable of serving legitimate
users. In AACT, for every access request message (M.2), the corresponding mesh router has to confirm a group
AACT: Anonymous and Accountable communication topology for Wireless Mesh Networks
www.iosrjournals.org 10 | P a g e
signature and check the validity of the signer. Both operations involve costly pairing operations, which, hence,
can simply be exploited by the opponent. To deal with this issue, we assume the same client- puzzle approach as
adopted in [18]. The idea of this approach is as follows: When there is no proof of the attack, a mesh router
process (M.2) usually. But, when under a suspected DoS attack, the mesh router will attach a cryptographic
puzzle to every (M. 1 ) and need the solution to the puzzle be attached to every (M.2). The mesh router commits
resources to process (M.2) only when the solution is correct. Typically, solving a client puzzle needs a brute-
force search in the solution space, while the solution conformation is trivial [18].
Therefore, the opponent must have abundant resources to be capable to promptly compute a huge
efficient number of puzzle solutions in line with his sending rate of bogus access request (M.2). In contrast,
although puzzles slightly increase genuine users' computational load when the mesh router is under attack, they
are still able to obtain network accesses despite the subsistence of the attack. We refer the readers to [18] for the
complete design.

5.2 User Privacy and Accountability Analysis
AACT protects user privacy in a complicated manner, while still maintain user's responsibility. First,
AACT enables user anonymity against the opponent, the user group managers, and TTP. In AACT, a network
user only authenticates himself as a genuine service subscriber without disclosing any of his identity information
by using the group signature method. Neither the opponent nor the user group managers can tell which
meticulous user generates a given signature. The adversary, even by compromising mesh routers and other
network users, that is, knowing a number of group private keys in addition to the group public key, still cannot
deduce any information about the particular group private key used for signature generation. This is due to the
rigidity of the underlying q-SDH problem, where q is a 1,020-bit prime number. Due to the same reason, a user
group manager also cannot differentiate whether or not one of his group members has signed a particular
signature as he has no knowledge of the corresponding
, i j
A s nor can he compute them. The same finish also
holds for TTP as TTP can compute neither Xj nor
, i j
A given
, i j j
A x . Furthermore, every data session in
AACT is recognized only through pairs of fresh random numbers, which again discloses nothing about user
identity information. In addition, AACT requires a network user to refresh session identifiers and the shared
symmetric keys for every different session. This further eliminates the linkage among any two sessions
originated from the same network user. We note that even with the help of compromised mesh routers and other
network users, the adversary still cannot judge whether two communication sessions are from the same user.
This is because, basically, none of them can tell whether two signatures are from the same user, given q- SDH
problem and decision linear problems on
1
G are hard.
Second, AACT gives adequate user privacy protection against NO while maintaining user
accountability. Since NO knows grt, it can always tell which gsk[i, j] produces a given signature. However, NO
has no knowledge about to whom gsk[i, j] is assigned as AACT allows a late binding among group private keys
and network users. Furthermore, it is the user group managers' sole liability to assign group private keys to each
network user without any participation of NO. Therefore, NO could only map gsk[i, j] to the user group i based
on
i
grp . Because no other entities except NO and the key holder himself has the knowledge of the
corresponding
, i j
A , and can therefore, generate the given signature, the key holder has to be a member of the
user group i. This audit result serves us both necessities. On one hand, the result only reveals partial unneeded
attribute information of the user and still protects user privacy to an extent. On the other hand, the result is
adequate for user accountability purposes for NO.
Finally, AACT gives revocable user anonymity against the law authority. As discussed in Section 4.5, the law
authority could track any particular user through the cooperation from both NO and the corresponding user
group manager.

5.3 Performance Analysis
Communication overhead: In AACT, Both authentication and key agreement protocols need only
three-way communication among mesh routers and network users and among network users. This is the
minimal communication rounds essential to achieve mutual authentication, and therefore, AACT incurs a
compact authentication delay. Furthermore, by design, AACT poses minimum additional communication
overhead on network users as they may carry their mobile clients such as PDAs and smart phones other than
laptops to access the WMN. These mobile clients are much less powerful as evaluate to mesh routers with
regard to their communication ability. In messages (M. 1 ), ( .1) M

, and( .2) M

, a network user only needs to


broadcast a group signature to accomplish the authentication function. As we base our group signature
difference in the scheme proposed in [8], the signature comprises two elements of
1
G and five elements of
1
G .
AACT: Anonymous and Accountable communication topology for Wireless Mesh Networks
www.iosrjournals.org 11 | P a g e
When using the curves described in [19], one can take p to be a 170-bit prime and as a group
1
G , where each
element is 171 bits. Thus, the total group signature length is 1,192 bits or 149 bytes. With these parameters,
security is about the same as a standard 1,024-bit RSA signature, which is 128 bytes [8]. That is, the length of
the group signature is almost the similar as that of a standard RSA-1024 signature.
Computational overhead: In AACT, the most computationally expensive operations are the signature
generation and verification. Signature generation requires two applications of the isomorphism . Computing
the isomorphism takes roughly the similar time as an exponentiation in
1
G (using fast computations of the trace
map) [8]. Thus, signature generation needs about eight exponentiations (or multi exponentiations) and two
bilinear map computations. Signature verification takes six exponentiations and 3 2 URL + computations of
the bilinear map. By design, AACT adopts an asymmetric-symmetric hybrid approach for session authentication
to decrease computational cost. Network entities (both mesh routers and network users) execute exclusive group
signature operation to authenticate each other only when establishing a new session; all subsequent data
exchanging of the same session is authenticated through a highly efficient MAC-based approach.
More specifically, AACT requires a network user executing exactly one signature generation and one
signature verification when performing mutual authentication for establishing a new session. It can be seen that
the actual computational cost of signature verification depends on the size of the URL, while signature
generation cost is fixed. AACT can proactively control the size of the URL. Moreover, a far more efficient
revocation checks algorithm, whose running time is independent of URL can be adopted as described in [8]
with a little bit sacrifice on user privacy. This technique could further bring the total cost of signature
verification to six exponentiations and five bilinear map computations. On the other hand, AACT requires a
mesh router to perform mutual authentication with every network user within its coverage for each different
session and sign on every beacon message being periodically broadcasted.
Storage overhead: In AACT, network users may carry resource-constrained persistent devices such as
PDAs and smart phones to access the WMN. Therefore, storage overhead for each network user should be
reasonable to modern pervasive devices. As is shown in our scheme description, each network user in AACT
needs to store two pieces of information: his group private key and the related system parameters. The group
private key for each user just contains 1 group element of
1
G and 2 elements of
*
p
Z . If we choose p to be a 170-
bit prime and as a group
1
G with each group element of 171 bits, the group private key for every user just
consumes 511-bit memory, which is insignificant for modern pervasive devices. The most memory-consuming
parts are the system parameters, which may contain codes to describe the bilinear groups (
1
G and
2
G ), the
bilinear pairing function (e), the isomorphism , the hash functions
0 1
( ) H andH , and the signing function
ECDSA-160. Fortunately, the needed code size for each part could be in the magnitude of kilobytes as is studied
in prior work such as [20]. Therefore, it should be affordable to most of the modern pervasive devices.

VI. RELATED WORK
Security study in WMNs is still in its early stage, particularly with respect to user privacy protection. Ben
Salem and Hubaux [21] discussed specifics of WMNs and identified basic network operations that needed to be
secured. Siddiqui and Hong [22] surveyed the threats and vulnerabilities faced by WMNs and also recognized a
number of security goals. Cheikhrouhou and Chaouchi [23] discussed a security architecture for WMNs based
on IEEE 802.1X. [5] And Zhang and Fang [4] discussed how to support secure user roaming in a number of
WMNs belonging to dissimilar domains. Wu and Li [24] presented an anonymous routing scheme for static
WMNs. Wan et al. [25] proposed two privacy-preserving routing schemes to give anonymity, unlinkability, and
security for WMNs. The authors of [26], [27] presented an authentication scheme for WMNs, which is resilient
against mesh router compromise. Other general privacy-aware authentication methods are described in [28],
[29], [30].

VII. CONCLUSION
In this paper, we proposed AACT, which, to the most excellent of our knowledge, is the first attempt to
set up an liable security framework with a complicated user privacy protection model tailored WMNs. We
tailored group signature scheme[8] that combined with onion ring strategy [31]. We then built AACT on this
new model by further integrating it into the authentication and key agreement protocol design. On one hand,
AACT enforces strict user access control to cope with both free riders and spiteful users. On the other hand,
AACT offers complicated user privacy protection against both adversaries and different other network entities.
Our analysis showed that AACT is elastic to a number of security and privacy related attacks. Additional
methods were also discussed to further improve the scheme efficiency.
AACT: Anonymous and Accountable communication topology for Wireless Mesh Networks
www.iosrjournals.org 12 | P a g e

REFERENCES
[1] K. Ren and W. Lou, "A Sophisticated Privacy-Enhanced Yet Accountable Security Framework for Wireless Mesh Networks," Proc.
28th Int'l Conf. Distributed Computing Systems (ICDCS '08), June 2008.
[2] I.F. Akyildiz, X. Wang, and W. Wang, "Wireless Mesh Networks: A Survey," Computer Networks, vol. 47, no. 4, pp. 445-487,
Mar.2005.
[3] "Self Organizing Neighborhood Wireless Mesh Networks," http://www.research.microsoft.com/mesh/, 2009.
[4] Y. Zhang and Y. Fang, "A Secure Authentication and Billing Architecture for Wireless Mesh Networks," ACM Wireless Networks,
to be published.
[5] Y. Zhang and Y. Fang, "ARSA: An Attack-Resilient Security Architecture for Multi-Hop Wireless Mesh Networks," IEEE J.
Selected Areas in Comm., vol. 24, no. 10, pp. 1916-1928, Oct.2006.
[6] "The Wimax Forum,"http://www.wimaxforum.org. 2009.
[7] "Boston Suburb Secures Metro-Scale Wireless Mesh Network with Bluesocket," http://www.tmcnet.com/usubmit/2006/09/27/
1936581.htm, Sept. 2006.
[8] D. Boneh and H. Shacham, "Group Signatures with Verifier-Local Revocation," Proc. ACM Conf. Computer and Comm. Security
(CCS), pp. 168-177, 2004.
[9] D. Chaum and E. van Heyst, "Group Signatures," Proc. Conf. Eurocrypt, pp. 257-265, 1991.
[10] R. Rivest, A. Shamir, and L. Adleman, "A Method for Obtaining Digital Signatures and Public-Key Cryptosystems," Comm. ACM,
vol. 21, no. 2, pp. 120-126, 1978.
[11] M. Jakobsson, J. Hubaux, and L. Buttyan, "A Charging and Rewarding Scheme for Packet Forwarding in Multi -Hop Cellular
Networks," Proc. Seventh Int'l Conf. Financial Cryptography (FC), 2003.
[12] N. Salem, L. Buttyan, J. Hubaux, and M. Jakobsson, "A MicroPayment Scheme Encouraging Collaboration in Multi-Hop Cellular
Networks," Proc. ACM MobiHoc, 2003.
[13] D. Hankerson, A. Menezes, and S. Vanstone, Guide to Elliptic Curve Cryptography. Springer-Verlag, 2004.
[14] Y. Zhang, W. Liu, and W. Lou, "Anonymous Communications in Mobile Ad Hoc Networks," Proc. IEEE INFOCOM, Mar. 2005.
[15] Y. Zhang, W. Liu, W. Lou, and Y. Fang, "MASK: Anonymous On- Demand Routing in Mobile Ad Hoc Networks," IEEE Trans.
Wireless Comm., vol. 5, no. 9, pp. 2376-2385, Sept. 2006.
[16] J. Kong, X. Hong, and M. Gerla, "An Identity-Free and On- Demand Routing Scheme against Anonymity Threats in Mobile Ad
Hoc Networks," IEEE Trans. Mobile Computing, vol. 6, no. 8, pp. 888-902, Aug. 2007.
[17] B. Zhu, Z. Wan, M.S. Kankanhalli, F. Bao, and R.H. Deng, "Anonymous Secure Routing in Mobile Ad-Hoc Networks," Proc. 29th
Ann. IEEE Int'l Conf. Local Computer Networks (LCN '04), pp. 102-108, Nov. 2004.
[18] A. Juels and J. Brainard, "Client Puzzles: A Cryptographic Countermeasure against Connection Depletion Attacks," Proc. Sixth
Network and Distributed System Security Symp. (NDSS), 1999.
[19] D. Boneh, H. Shacham, and B. Lynn, "Short Signatures from the Weil Pairing," J. Cryptology, vol. 17, no. 4, pp. 297-319, 2004.
[20] TinyECC Library, http://discovery.csc.ncsu.edu/software/ TinyECC/index.html, 2009.
[21] N. Ben Salem and J.-P. Hubaux, "Securing Wireless Mesh Networks," IEEE Wireless Comm., vol. 13, no. 2, pp. 50-55, Apr.2006.
[22] M. Siddiqui and C. Hong, "Security Issues in Wireless Mesh Networks," Proc. IEEE Int'l Conf. Multimedia and Ubiquitous
Eng.,2007.
[23] A. Cheikhrouhou and H. Chaouchi, "Security Architecture in a Multi-Hop Mesh Network," Proc. Fifth Conf. Security Architecture
Research, 2006.
[24] X. Wu and N. Li, "Achieving Privacy in Mesh Networks," Proc. ACM Workshop Security of Ad Hoc and Sensor Networks
(SASN),2006.
[25] Z. Wan, K. Ren, B. Zhu, B. Preneel, and M. Gu, "Anonymous User Communication for Privacy Protection in Wireless
Metropolitan Mesh Networks," Proc. ACM Symp. Information, Computer and Comm. Security (AsiaCCS), 2009.
[26] X. Lin, R. Lu, P.-H. Ho, X. Shen, and Z. Cao, "Tua: A Novel Compromise-Resilient Authentication Architecture for Wireless Mesh
Networks," IEEE Wireless Comm., vol. 7, no. 4, pp. 1389-1399, Apr. 2008.
[27] X. Lin, X. Ling, H. Zhu, P.-H. Ho, and X. Shen, "A Novel Localized Authentication Scheme in ieee 802.11 Based Wireless Mesh
Networks," Int'l J. Security and Networks, vol. 3, no. 2, pp. 122132, 2008.
[28] K. Ren, W. Lou, K. Kim, and R. Deng, "A Novel Privacy Preserving Authentication and Access Control Scheme for Pervasive
Computing Environment" IEEE Trans. Vehicular Technology, vol. 55, no. 4, pp. 1373-1384, July 2006.
[29] K. Ren and W. Lou, "Privacy-Enhanced, Attack-Resilient Access Control in Pervasive Computing Environments with Optional
Context Authentication Capability," ACM Mobile Networks and Applications (MONET) (special issue on wireless broadband
access), vol. 12, pp. 79-92, 2007.
[30] Y. Zhang and K. Ren, "On Address Privacy in Mobile Ad Hoc Networks," ACM/Springer Mobile Networks and Applications
(MONET), vol. 14, no. 2, pp. 188-197, Apr. 2009.
[31] M. Reed, P. Syverson, and D. Goldschlag, Anonymous Connections and Onion Routing, IEEE Journal on Selected Areas in
Communication Special Issue on Copyright and Privacy Protection, 1998.

You might also like