Network coding combined with onion routing for anonymous and secure communication in a wireless mesh network

International Journal of Computer Networks & Communications (IJCNC) Vol.6, No.6, November 2014
NETWORK CODING COMBINED WITH ONION
ROUTING FOR ANONYMOUS AND SECURE
COMMUNICATION IN A WIRELESS MESH
NETWORK
Abhinav Prakash, Dharma P. Agrawal, and Yunli Chen
ABSTRACT
This paper presents a novel scheme that provides high level of security and privacy in a Wireless Mesh
Network (WMN). We combine an approach of Network Coding with multiple layered encryption of onion
routing for a WMN. An added superior feature provides higher level of security and privacy. Sensitive
network information is confined to 1-hop neighborhood which is available anyways in a wireless medium
with nodes using a bivariate polynomial. The only routing information divulged to a relay node is about
next hop. No plain text is ever transmitted and all data can only be decrypted by its source and destination.
Prior work finds it difficult to enforce encryption with network coding without divulging in complete
routing information,hence losing privacy and anonymity. We compare our scheme with other existing
approach for several networks. The preliminary results show this work to provide superior security and
anonymity at low overhead cost.
INDEX TERMS
Network Coding, Mesh Networks,Onion Routing, Privacy, Security.
I. INTRODUCTION
A Wireless Mesh Network (WMN) consists of Internet Gateways (IGWs), Mesh Routers (MRs)
and Mesh Clients (MCs). MCs are served by MRs which are connected together using wireless
links in an ad hoc mode and constitute as the backbone of a WMN.Some MRs act as IGWs to
provide access to the Internet and makes the network very costeffective. MRs could be of
different types, such as Wi-Fi, WiMax routers etc. and could have different interfaces. This leads
to two levels of hierarchy, MRs at higher level with many different wireless interfaces, and MCs
constitute the lower level and are served by MRs using separate interfaces with the presence of
few IGWs wired to the Internet, the cost is low, while easy to expand using additional MRs. In
this way, it is easy to expand WMNs, especially in the sparsely populated areas and makes them
easily scalable. Another very important application of a WMN is to provide different medium of
wireless access. Hence, WMN can be said to be dynamically self-organized and auto-healing.
Versatilities of a WMN is especially important in an open wireless medium where MRs are
owned by different independent entities.Such networks mainly rely on an ad-hoc packet
transmission mechanism as MRs forward along the route from source to destination acting
as relay nodes. Several possible misbehaviorscould be identified as impersonations, packet
sniffing, selfish behavior etc. There are several works to date focusing on identifying and
rectifying many such misbehavior [1]. In this paper we introduce a scheme to utilize the concept
of network coding that maximizes efficiency of the WMN with low overheads and high level of
security and anonymity by using encryption at each link along the path.
DOI : 10.5121/ijcnc.2014.6601 1

2
A. Background and Related Work
1) Security: The field of Wireless Mesh Network is still in its early stage and hence security is
open. The protocol for WMNs are still to be developed. In a WMN the MRs are pretty static
while MCs move around at different speed and get connected to MRs in their vicinity. In the year
2006 [2] described various attacks like sinkhole and wormhole attacks and also look into
numerous vulnerabilities of a WMN. Many ideas have been proposed to combat these attacks
using different techniques, including shared private keys and public-private key pairs. RSA-based
public key cryptography is one example. The Asymmetric key system are complex for
authentication and data communication as it is computationally very expensive. Hence, to
maintain low energy consumption in MCs with limited energy sources, we want our scheme to be
light weighted.
Use of symmetric keys seems to be an attractive option. In [3], Eschenauer and Gligor proposed a
random key distribution scheme such that each device randomly selects a group of keys from a
large pool of P keys.A major disadvantage of this scheme is that if the number of keys given is
small, most of the devices are sparsely connected and are not capable of communicating with
each other as they may not have even one key in common. Hence, lots of devices remain
disconnected. When the number of keys given is large, the scheme becomes susceptible to a
device capture attack. This could force a large amount of secret to be lost, thereby compromising
a large portion of the network. Furthermore, each pair of devices to have a shared symmetric key
each device must be given at least n-1 common keys to spawn capability of complete
connectivity. Chan et al. [4], revised the Eschenauers and Gligors model by having at least q
(where q>1) keys in common between two adjacent wireless devices instead of just one common
key, so that they can have a secure wireless link between them. They call their scheme to be q-composite
random key pre distribution scheme. This is one way to enhance the network resilience
against any device capture attack. Blom also introduced a symmetric key generation scheme
(SKGS) [5], where a pair of devices can independently generate a common key between them by
exchanging a small amount of secret information. However there could be dependencies between
the keys in this particular approach and a number of users have to collaborate in resolving the
ambiguity of unknown keys. This makes this scheme vulnerable to device capture attack when the
number of captured devices exceed a given threshold value which is directly proportional to the
amount of secret divulged. In [6], Blundo et al. proposed a secure key distribution scheme for
systems where a device may leave or enter the network dynamically, by that constantly changing
the network topology. Therefore, they proposed a t-degree bivariate symmetric polynomial pre
distribution scheme. Such a strategy is equally pertinent to any hierarchical networks. Usually, the
communicating wireless devices swap polynomials by substituting the variables with their
respective IDs. It is possible to compute a common secret key between them due to the symmetric
attribute of the polynomial. This scheme is called as k-secure where k pertains to the degree of the
symmetric polynomial.
Yi Cheng et al. [7] have presented a pair wise key establishment mechanism (EPKEM) by
generating keys and arranging them in a matrix so that a common shared key between any
communicating nodes can be achieved. Each user is allocated a row and a column of keys to form
(2m - 1) set of keys. The selected (2m-1) elements from the matrix are then loaded into each
device along with the(i; j) coordinates of the respective elements to form its key ring and then
they are deployed randomly. In this scheme, two devices discover a common key between them
by broadcasting their respective IDs while the indices (i; j) of keys are exchanged that were used
at the time of key pre-distribution. This scheme drastically reduces the number of keys that are
needed to be pre-stored on the devices during the deployment phase, while assuring at least two
common keys among a pair of any adjacent devices.

2) Preliminaries: Polynomial Based Scheme: In [8], we proposed a bivariate polynomial
function based security scheme that is highly scalable at a very low cost. In this scheme, we
introduced a novel scheme to allow a secured authenticated connection between any two entities
in a WMN. The two adjacent nodes can be an IGW, one MR or any MC. The crucial step of this
scheme is to furnish each node a set of bivariate polynomials during the pre-deployment phase
by the central authority. Once deployed, this secret polynomial are used to independently generate
symmetric secure key and once achieved, we say that the two nodes have an generated an
authenticated association with each other. It should be noted that during the pre-deployment
phase, three different sets are given to each node as follows:
3
1) A shared key K for initial secure information exchange.
2) A set of Bivariate Polynomial Functions Fi,j,k(x,y)(where 0 i < l; 0 j < m; 0 k < m) picked
randomly from a 3D matrix of polynomials and the indices of the selected polynomial functions.
3) A function H( ) known as the hash function to compute the shared key from the values received
by the bivariate polynomial functions. This three dimensional matrix has been adapted from Yi
Cheng‘s scheme in [7] and [9] by randomly selecting the polynomials. From the get go each
device undergoes three stages of
1) Acquiring Secrets.
2) Authenticated Association.
3) Pair-wise secure channel establishment from a Mesh Client to an IGW or the AAA server.
A standard Bivariate Polynomial distribution is construed as follows:
Where the coefficients ars are randomly selected over a Finite Field Gf(X) where X is a
very large prime number and i; j; k are the indices for the position of the polynomial in the
three-dimensional matrix used at pre-deployment and p is the degree of the function Fi;j;k (x,y).
A Polynomial Function is called a Bivariate Polynomial if it satisfies the following elementary
prerequisite:
A three dimensional matrix is conceived containing a bivariate polynomial at each unit position of
this matrix. The bivariate polynomi-als selected to populate this three dimensional matrix are
randomly chosen from a large pool of such possible bivariate polynomials. For ex-plicit
understanding we can contemplate such a matrix as a set of i two dimensional mxm ma-trices
with degree t. This allows us to compute

4
Fig. 1: A three dimensional matix of bivariate polynomials Fig. 2: A M M Matrix
the total number of bivariate polynomials from a three dimensional matrix as explained below:
Total number of polynomials = l*m*m such that: 0t m and 0 i l We can easily say that a
function at the position i; j; k can be written as Fi;j;k(x; y) as displayed in the Figure 1. Now by
selecting a set S randomly of matrices containing bivariate polynomials from l, i.e., the total
number of m m two dimensional matrices can be obtained as: S d(l + 1)=2e . where dxe is the
ceiling function that gives the smallest integer x.Using the ceiling function, We ensure that S is
an integer always greater than or equal to l=2. This further guarantees that two different randomly
selected sets Sa and Sb always have atleast one m m Matrix in common. After selecting the
random set S of matrices with one random column and one random row from each of these
matrices in S, all the functions contained in the selected row and column are given to a mesh
entity. For example refer Figure2.
Now, since this matrix is of the order m mit has m rows and m columns. So, the number of
polynomials contained in one row and one column selected randomly are m + (m 1). Now,
there are S such matrices hence total number of bivariate polynomials given to each MC are:
The total number of bi-variate polynomials = S (m + (m 1)) = S (2m 1) Now, Let us analyze

how two clients on the mesh can have common functions to establish a secure communication
channel. Since we know two different set of matrices Sa and Sb have atleast one matrix in
common. Let the two common matrices be as shown in Figure 3 Assuming the highlighted rows
and columns were randomly selected row and column for sets Sa and Sb respectively. It is
obvious that both these sets will have atleast two functions in common. In a better case, there
could
5
(a) Sa (b) Sb
Fig. 3: Common Matrix between sets Sa and Sb
be more common matrices, leading to more common functions between two mesh entities. So,
this technique of allocating polynomials guarantees any two MCs to have atleast two common
bivariate polynomial functions which are used for secured communication.
Achieving a Secure Link for Communi-cation: When the network is formed, each MC identifies
its neighbors and exchanges informa-tion to generate a secure key for communica-tion. Say for
example, Once two neighbors C and D find each other, they share each others node ID and the
indices of the polynomial functions they possess. This information is encrypted using a common
Key K which is given to each node for an initial handshake and exchange information to generate
a secured key on the fly. Using the polynomial function indices, both the nodes separately
determine which function they have in common.
Assuming the node IDs are IDC and IDD and the common functions are F2;7;6() and F2;3;4().
At node C, seed values will be computed using the common functions and node IDs of C and D.
Seed 1C = F2;7;6(IDC , IDD)
Seed 2C = F2;3;4(IDC , IDD)
Similarly, at node D, it would compute its seed values:
Seed 1D = F2;7;6(IDD , IDC)
Seed 2D = F2;3;4(IDD , IDC)
Since functions F2;7;6() and F2;3;4() are bivari-ate polynomial we get using this property:
F2;7;6(x; y) = F2;7;6(y; x)
The same is true for any other common func-tion therefore:
Seed 1C = Seed 1D
Seed 2C = Seed 2D
This is applicable to any further seeds.
So, the seeds generated independently at both the nodes would be identical. Each node uses a one
way hashing function Hf() that is assigned during the deployment phase. All the seed values are
hashed to generate a final secured key for communication and since the same hash function is
used at both the nodes and seed values being identical, they both have the same identical unique
key for encryption or decryption. This key is never sent over the network and is just used for

encryption at the sender end and decryption at the receiver which ensures the key not to be stolen
by other entities that might be overhearing the communication.
SecureKey = Hf(Seed 1C ; Seed 2C ; : : :)
If the two nodes have more than two common functions, they generate more than two seed values.
This provides more than two seed val-ues to the hashing function which makes it even stronger
and more secure. In this fashion, all the nodes establish secure encryption technique with their
one hop neighbors. Assuming a node A needs to communicate to the Internet Gateway which is
five hops away from it. Each of the four links on the way would be a secured connection using
pairwise key. This pairwise secure key establishment ensures end-to-end secured transfer.
3) Privacy: A lot of work has been done in the field of security [7], [9] for a WMN. But, multiple
vulnerabilities still remain open in the field of privacy. Several researchers have attempted to
address the privacy issues related to a wired network and there seems to have a lot of scope in this
field especially in a wireless network scenario. For example, in a wired network, Onion Routing
[10] was invented by Michael G. Reed, Paul F. Syverson, and David M. Goldschlag to provide
anonymity of destination. In this scheme, a path is pre-computed at the source and the data packet
is encrypted in multiple layers with the public key of the forwarding node along the path to
destination in a sequential order and each node removes their layer of encryption after receiving
the packet and forward the remaining packet called the onion to the next hop and finally the
decrypted data with all the layers of encryption removed, received by the des-tination. The work
in [11] resorts to a quan-titative approach towards privacy. Authors in [12] specifically approach
provisions for pri-vacy preservation but lack efficiency of gains achieved from the network
coding. The basis of privacy performance parameters are well defined in [13].
In [14], Wu and Li introduced an onion ring protocol for wireless mesh networks where onion
rings are formed starting at the IGW and using all the cycles in the network. Data only travels in
one direction and data sessions are only initiated at the gateway router. This provides a good
anonymity, while fails on several issues such as a bottleneck is created at the gateway router as all
the scheduling is done at the gateway. Moreover, this scheme works on the concept of finding
cycles. In a dynamic WMN, uplink messages are initiated at the MR to traverse through the
network to the IGW, which could fail in a realistic scenario. Another problem in this protocol is
all the other nodes in a ring have to wait while one node communicates.
The work [15] talks about a layered onion ring approach in which some routers are con-sidered
trusted nodes and is similar to [14]. Communication starts and ends at the MR and
6
Fig. 4: Coding Gain with oppurtunistic listen-ing Example 1

7
Fig. 5: Coding Gain with oppurtunistic listen-ing Example 2
some level of anonymity is present in the net-work. But, this work strongly relies on finding the
cycles in the network.
4) Random Linear Network Coding: In Random Linear Network Coding (RLNC), par-ticipating
nodes combine their incoming pack-ets linearly using randomly chosen coefficients. It has been
observed to be very efficient with reference to the network coding exemplar [16]. The basic idea
behind network coding is when a node overhears multiple incoming packets, it simultaneously
places a linear combination of all the overheard packets and retransmits this single coded packet
which when received, can be decoded to obtain original packets if sufficient information is
available at the decod-ing node. Example of coding gain achieved as explained in [17] is shown
in Figures 4 and 5. Intuitively, these gains are huge and open new doors to multicast networks
like P2P and WMN’s. As any new superior technology not only does it create several new
possibilities and strengths to wireless networks, it also opens new doors for exploiting new
security and privacy schemes. Inherently network coding approach does provide some security
due to its intrinsic nature. But, being an application for content distribution efficiently it works on
co-operative networking and all nodes are trusted with sharing several parameters for information
extraction which faced with any smart mali-cious node fails miserably. Intrinsically Net-work
Coding approach has a huge conflict with strong anonymous schemes like onion routing and are
actually unusable in their original form. Our approach modifies both these well known schemes
and combines them together to work cooperatively while removing any conflicts sys-tematically.
Some existing similar approaches employing network coding and achieving pri-vacy can be found
in [18] and [19]. Authors in [20] introduce an enhanced scheme for network coding and promise
even superior gains. The work in [21] achieves groundbreaking results in deploying network
coding for a realistic wireless network and provides huge gains over conventional packet
forwarding mechanisms. In our proposed scheme, we use the foundation of network coding using
XOR scheme. This work is our further enhancement to the work previously published in [22].
II. PROPOSED SCHEME: NETWORK CODING ON MODIFIED ONION ROUTING
We assume that all MCs have Omni-directional Antenna with uniform transmis-sion range.
Network Initiation takes place with neighborhood discovery by MCs, MRs and IGW’s. Each MC
discovers its one hop neigh-bor MR’s which we define as a neighborhood as given in Table I. Our
scheme has two
Fig. 6: Message Propagation

8
Fig. 7: A Wireless Mesh Network
salient features. First, a group key is generated using bivariate polynomial among devices in a
neighborhood. The second step is to merge the message with the group key similar to network
coding so that actual message is never transmitted, but extracted at each intermedi-ate MR as
illustrated in Figure 6. When the message propagates from MR1 to MR2 and so on. As MR2
knows the group key K1 between two MRs, the message can be easily extracted. Then, MR2
encrypts message by mixing with K2 similar to network coding and so on till the message reaches
the IGW.
A. Network Initiation
Network initiation starts with one hop neigh-bor discovery and creating authenticated secure
communication paths. Each MC registers its membership with the MR over a secure com-munication
channel. After MC’s registration is complete MR processes the network map creating
a uniform spanning tree with branches of the tree that gives routes going through various MRs
and eventually to an IGW.
Having multiple MRs makes our scheme
TABLE I: A Branch Table Entry
BID Nextid Previd
34 A3E C5F
even stronger as it enables routers to work cooperatively while providing higher privacy and
sharing the load. In such cases branches starting and ending in a MR can be created for the
network topology of Figure 7 and the corre-sponding network model of Figure 9. Creating such
Neighborhoods gives the scheme multiple strengths. All the members of the neighborhood that
are a part of a path in which data flows, is illustrated in Figure 8.

9
III. IMPLEMENTATION DETAILS
This section describes how our scheme reacts in different network scenarios and how network
coding is intertwined with onion routing using onion packets providing very high level of privacy
at a low cost and achieving the cod-ing gain utilizing opportunistic listening and coding as in
[17]. In our proposed scheme we have segregated MRs into one-hop neighbor-hoods. This logical
network sectioning makes our scheme robust and distributed hence ideal for a Wireless Mesh
Scenario.
We use a similar network coding scheme as proposed in COPE in [17]. Additionally we empower
the relay nodes by providing just enough information to be capable of decipher-ing routing
information to the next hop in the correct direction towards the destination node. This is achieved
without disclosing a source-destination pair and the payload of the data packet. At the same time
a relay node
Fig. 8: Route for Mesh Routers to IGW
Fig. 9: Network Map at MR (center)
is capable of opportunistically decoding and encoding data packet again together but with a very
different key. This makes the COPE [17] coding scheme even more efficient and utilizes lower
overheads while keeping high gains.
IV. PERFORMANCE ANALYSIS
In [8], we have shown that our PBS (polynomial based scheme) is highly scalable and perfectly
secure which is provided at a very low cost. With very few functions stored on a MC, we can
support a very large network. The proposed Network Coding combined with Onion
Routing(NC&OR) scheme takes this to the next level and fills the gap of privacy in case of an
attack by a global adversary. Very high level of anonymity is achieved at the cost of an
incremental overhead, like decryption and encryption with a new key at each stage. The
overheads are caused by Multiple Encryption as per onion routing. It can be easily observed that

the onions incur a heavy cost of usage. But, in NC&OR, we only use forward path. This helps us
in keeping the cost to bare minimum for multiple encryptions as compared to a pure onion ring
routing in [14] and layered onion ring routing approach of [15] where they always use onions for
10
Fig. 10: A Packet Propagation Path from Clientto IGW Fig. 11: Regular MC
Deployment in Hexagon, Triangle
and Square Patterns
any type of communication. Redundancy in Network Coding and Modified Onion Routing is
much more efficient than Phantom Routing of [23] which is based on flooding to ensure privacy
among a group of nodes that necessitates too many re-transmissions.
Anonymity:
To an outside observer, all the MRs in a WMN act exactly the same way. The encrypted
communication combined with the usage of dummy packets makes it impossible even for the
global observer [24] to isolate the node initiating the communication session. Additionally, a
session is never initiated at the MC as it can only request for a session to a MR. In case of
presence of an inside attacker, information can not be leaked as each MC only knows about the 1-
hop information among the branch (previous and next entity). The data packets are forwarded by
an inside attacker and are encrypted. Hence, a dummy packet and a data packet are
indistinguishable by an inside attacker as they appear exactly the same in their encrypted form.
In our scheme, the flow of traffic goes through two layers of branches which makes it impossible
to isolate the session initiator. It randomly propagates the packet towards the IGW. In our case,
the selection of branches on different levels is totally independent which makes it difficult to
predict what path the packet is going to take by each MR. Addi-tionally, availability of multiple
branches adds further randomization to the selection of the final path taken. Multiple layers and
availability
Fig. 12: Regular Network Topology
of several branches makes our scheme more or less private and secure. Furthermore, each
encryption key can be changed dynamically over time, using the bivariate polynomial.
Experimental Results:

We now evaluate our scheme using realistic wireless settings. Our evaluation is based on
applying real network metrics into the well known ns-2 network simulator. In our exper-iment
two types of nodes are deployed. A MR acting as the sync node and rest of the MRs are all relay
nodes with network coding capabilities and few nodes are randomly selected to act under an
active session with the MR. We compare our scheme with the following two routing protocols:
1. COPE, the routing protocol with network coding (simulated by using our algorithm with-out
11
any encryptions under same constraints and parameters),
Fig. 13: Average Throughput in Cross Topol-ogy
Fig. 14: Coding Gain Achievement Compari-son in Cross Topology
Fig. 15: Average Throughput in Grid Topology

12
Fig. 16: Coding Gain Achievement Compari-son in Grid Topology
2. TOR (The Onion Routing) protocol be-tween initiating MR and IGW with no network coding
(simulated by using our algorithm with-out any network coding under same constraints and
parameters).
Our results are compared on the basis of performance in these three parameters:
1.Average Throughput over entire runtime.
2.Coding Gain (Fraction of total packets that were coded before forwarding)
3.Encryption/Decryption Cost, the computa-tional cost of packet processing when encryp-tion is
used.
With our earlier experience with regu-lar network deployments [25], we considered three
standard deployments namely: Triangular, Square and Hexagonal as displayed in Figure 11.
Triangular and Hexagonal deployments es-sentially break down into the cross topology and
square deployment is represented by the grid topology as shown in Figure 12. We used all default
seed values of 802.11b wireless networking specification using UDP data pack-ets with
transmission range is set to 250m. Results are averaged over 15 simulations for each setup.
Graphs 13, 14, 15, and 16 show how our scheme Network Routing and Onion Routing (NC&OR)
gives high level of coding gain almost equivalent to COPE while main-taining Onion Routing’s
high level or security and privacy. Further we deduce from our results that Cross topology
outperforms Grid topology.
V. CONCLUSION
In this paper, we propose solution to an sub-stantial problem that when a wireless network
unfolds network coding, previously operational privacy-sustaining methods no longer operate
correctly or are left non-functioning. To this end, we propose a superior hybrid scheme which
combines the strengths of onion rout-ing and embedding it with Network Coding , a
contemporary anonymous communication protocol which can function for distributed dynamic
wireless networks. Both analytical and experimental results suggest that our scheme not only
keeps the advantage of network coding for an effective use of the capacity, but also ensures
enhanced privacy for MCs without much overhead.
VI. FUTURE WORK
Our proposed scheme encounters various complex network scenarios in a random net-work
topology scenario. The research work is ongoing in a realistic random deployed WMN.

Preliminary results obtained experi-mentally show promising prospects for high gains and
efficient performance of our scheme NC&OR in a random WMN.
ACKNOWLEDGMENT
The authors would like to thank members of the Center of Distributed and Mobile Comput-ing
Lab (CDMC) at the University of Cincin-nati for their technical feedback which helped improve
the presentation.
REFERENCES
[1] N. Nandiraju, D. Nandiraju, L. Santhanam, B. He, J. Wang, and D. Agrawal, “Wireless mesh
networks: Current challenges and future directions of web-in-the-sky,” Wireless Communications,
IEEE, vol. 14, pp. 79 –89, august 2007.
[2] Y. Zhou and Y. Fang, “Security of ieee 802.16 in mesh mode,” in Military Communications
13
Conference, 2006. MILCOM 2006. IEEE, pp. 1 –6, oct. 2006.
[3] L. Eschenauer and V. D. Gligor, “A key-management scheme for distributed sensor networks,” in
Proceedings of the 9th ACM conference on Computer and communica-tions security, CCS ’02, (New
York, NY, USA), pp. 41– 47, ACM, 2002.
[4] H. Chan, A. Perrig, and D. Song, “Random key predis-tribution schemes for sensor networks,” in
Security and Privacy, 2003. Proceedings. 2003 Symposium on, pp. 197 – 213, may 2003.
[5] R. Blom, “An optimal class of symmetric key generation systems,” in Proc. of the EUROCRYPT 84
workshop on Advances in cryptology: theory and application of crypto-graphic techniques, (New
York, NY, USA), pp. 335–338, Springer-Verlag New York, Inc., 1985.
[6] C. Blundo, A. D. Santis, A. Herzberg, S. Kutten, U. Vac-caro, and M. Yung, “Perfectly-secure key
distribution for dynamic conferences,” in CRYPTO, pp. 471–486, 1992.
[7] Y. Cheng and D. Agrawal, “Efficient pairwise key estab-lishment and management in static wireless
sensor net-works,” in Mobile Adhoc and Sensor Systems Conference, 2005. IEEE International
Conference on, pp. 7 pp. –550, nov. 2005.
[8] A. Gaur, A. Prakash, S. Joshi, and D. P. Agrawal, “Polynomial based scheme (pbs) for establishing
au-thentic associations in wireless mesh networks,” Journal of Parallel and Distributed Computing,
vol. 70, no. 4,
pp.338 – 343, 2010.
[9] Y. Cheng, M. Malik, B. Xie, and D. P. Agrawal, “En-hanced approach for random key pre-distribution
in wire-less sensor networks,” in Proceedings of International Conference on
Communication, Networking and Infor-mation Technology, 2008.
[10] M. G. Reed, P. F. Syverson, and D. M. Goldschlag, “Onion routing network for securely moving data
through communication networks,” 07 2001.
[11] Y. Li, J. Ren, and J. Wu, “Quantitative measurement and design of source-location privacy schemes
for wireless sensor networks,” IEEE Transactions on Parallel and Distributed Systems, vol. 23, no. 7,
pp. 1302–1311, 2012.
[12] Z. Wan, K. Ren, B. Zhu, B. Preneel, and M. Gu, “Anonymous user communication for privacy
protection in wireless metropolitan mesh networks,” Vehicular Tech-nology, IEEE Transactions on,
vol. 59, pp. 519–532, Feb 2010.
[13] K. Vu, R. Zheng, and J. Gao, “Efficient algorithms for k-anonymous location privacy in participatory
sensing,” in INFOCOM, 2012 Proceedings IEEE, pp. 2399–2407, March 2012.
[14] X. Wu and N. Li, “Achieving privacy in mesh networks,” in Proceedings of the fourth ACM
workshop on Security of ad hoc and sensor networks, SASN ’06, (New York, NY, USA), pp. 13–22,
ACM, 2006.
[15] R. Li, L. Pang, Q. Pei, and G. Xiao, “Anonymous com-munication in wireless mesh network,” in
Computational Intelligence and Security, 2009. CIS ’09. International Conference on, vol. 2, pp. 416
–420, dec. 2009.
[16] T. Ho, M. Medard, R. Koetter, D. Karger, M. Effros, J. Shi, and B. Leong, “A random linear network
coding approach to multicast,” Information Theory, IEEE Trans-actions on, vol. 52, pp. 4413–4430,
Oct 2006.

[17] S. Katti, H. Rahul, W. Hu, D. Katabi, M. Medard, and J. Crowcroft, “Xors in the air: Practical
wireless network coding,” Networking, IEEE/ACM Transactions on, vol. 16, pp. 497–510, June 2008.
[18] P. Zhang, C. Lin, Y. Jiang, P. P. C. Lee, and J. C. S. Lui, “Anoc: Anonymous network-coding-based
commu-nication with efficient cooperation,” IEEE Journal on Selected Areas in Communications, pp.
1738–1745, 2012.
[19] J. Wang, J. Wang, C. Wu, K. Lu, and N. Gu, “Anony-mous communication with network coding
14
against traffic analysis attack,” in INFOCOM, 2011 Proceedings IEEE,
pp.1008–1016, IEEE, 2011. 11
[20] T. Cui, L. Chen, and T. Ho, “Energy efficient opportunis-tic network coding for wireless networks,”
in INFOCOM 2008. The 27th Conference on Computer Communica-tions. IEEE, April 2008.
[21] J. C. Corena, A. Basu, S. Kiyomoto, Y. Miyake, and T. Ohtsuki, “Xor network coding pollution
prevention without homomorphic functions,” in Consumer Com-munications and Networking
Conference (CCNC), 2014 IEEE 11th, pp. 293–300, Jan 2014.
[22] A. Prakash, A. Gaur, and D. P. Agrawal, “Multilevel onion tree routing for anonymous and secure
commu-nication in a wireless mesh,” Cyber Journals: Multidis-ciplinary Journals in Science and
Technology, Journal of Selected Areas in Telecommunications (JSAT), vol. 3, January 2013.
[23] P. Kamat, Y. Zhang, W. Trappe, and C. Ozturk, “Enhanc-ing source-location privacy in sensor
network routing,” in Distributed Computing Systems, 2005. ICDCS 2005. Proceedings. 25th IEEE
International Conference on, pp. 599 –608, june 2005.
[24] K. Mehta, D. Liu, and M. Wright, “Protecting location privacy in sensor networks against a global
eavesdrop-per,” Mobile Computing, IEEE Transactions on, vol. 11, pp. 320 –336, feb. 2012.
[25] Y. Wang and D. P. Agrawal, “Optimizing sensor networks for autonomous unmanned ground
vehicles,” Optics/Pho-tonics in Security and Defense, (Cardiff, Wales, United Kingdom), vol. 7112,
pp. 15 –18, Sep. 2008.
AUTHORS
Abhinav Prakash is a PhD student in the Computer Science Division at the University
of Cincinnati. He is working for his doctorate under the guidance of Dr. Dharma P.
Agrawal at the Center for Development of Mobile Comput-ing(CDMC) Lab a part of
EECS depart-ment at University of Cincinnati. He has a Bachelors in Computer Science
from Banaras Hindu University in India and a Masters of Science in Computer Science
from University of Cincinnati. His research interests include distributed systems,
security in sensor and mesh networks, and applications of parallel computing.
Dharma P. Agrawal is the Ohio Board of Regents Distinguished Professor and the
founding director for the Center for Distributed and Mobile Computing in the EECS
department, University of Cincinnati, OH. He is a coauthor of textbooks on
Introduction to Wireless and Mobile Systems, 4th edition and Ad hoc and Sensor
Networks, 2nd edition.His recent research interests include resource allocation and
security in mesh networks, efficient deployment and security in sensor networks, use of
Femto cells, and heterogeneous wireless networks. He has seven approved patents, two
personal pending patents and twenty four patent filings in the area of wireless cellular
networks. He has been the Program Chair and General Chair for numerous international conferences and
meetings. He is a Fellow IEEE 1987; ACM 1998; AAAS 2003; WIF 2004, and Charter Fellow, National
Academy of Inventors, 2012 and is recipient of the IEEE-CS 2008 Harry H. Goode Memorial Award.
Yunli Chen received her BS, MS and Ph.D. degrees from Zhejiang University, China,
Beijing Institute of Technology, China and University of Cincinnati in 1994, 1999 and
2004 respectively. From 2004 to 2008, she was with Motorola. Currently, she is an
associate professor with Beijing University of Technology, where she conducts research
and development of MAC, routing, and transport protocols for wireless mesh networks.
Her research interests also include cross-layer design, energy efficient power allocation
and communication protocols for cellular, mobile ad hoc, and sensor networks

Network coding combined with onion routing for anonymous and secure communication in a wireless mesh network

More Related Content

What's hot (19)

Viewers also liked (19)

Similar to Network coding combined with onion routing for anonymous and secure communication in a wireless mesh network (20)

More from IJCNCJournal (20)

Recently uploaded (20)

Network coding combined with onion routing for anonymous and secure communication in a wireless mesh network