Location Similarity based Replica Node Detection for Sensor Networks
Lijun Yang
College of Internet of Things
Nanjing University of Posts and
Telecommunications
Nanjing, China
e-mail: yanglijun@njupt.edu.cn.com
Chao Ding
College of Computer Science
Nanjing University of Posts and
Telecommunications
Nanjing, China
e-mail: dingchao_129@163.com,
Meng Wu
Key Lab of “Broadband Wireless
Communication and Sensor Network
Technology” of Ministry of Education
Nanjing University of Posts and
Telecommunications
Nanjing, China
e-mail: wum@njupt.edu.cn
Abstract—The node replica attack is known to be dangerous to
wireless sensor networks (WSNs) because it enables the
adversary to extend the damage throughout the network with
very low cost. To stop such attack, we propose a similarity
estimation based scheme with group deployment knowledge.
Compared with prior works, our proposal provides extra
functionality that prevent replica from generating false
location claims. Through simulation experiment, we evaluate
the performance of the proposed scheme and demonstrate that
our scheme achieves effectiveness and efficiency under
different situations and attack strategies.
Keywords-sensor networks; security; replica detection;
location anomaly detection; deployment knowledge
I.
I
NTRODUCTION
Low-power wireless sensor networks (WSNs) are known
to be capable of rapid deployment in large geographical area
in a self-organized manner, which makes them particularly
suitable for real-time large-scale data collection and event
monitoring for mission-critical applications such as target
tracking and in-network aggregation. However, such
wireless sensor network applications are usually vulnerable
to a variety of attacks both from inside and outside of the
network due to their unattended nature and poor security
guarantee.
Among such attacks, the node replica attack [1] may be
especially dangerous to wireless sensor networks since the
adversaries with a small number of compromised nodes can
easily generate a large number of replica nodes which share
the keying materials and IDs with the original compromised
ones, and spread such replicas throughout the network. The
replicas, which are regarded as legal members of the network
since their keying materials and IDs are accepted by the
security mechanisms, are capable of assisting the
compromised ones in launching the inside attacks. By
injecting large number of replicas into the target network, the
adversaries manage to determine the network without being
detected, while the cost of injecting replicas is much lower
than that of injecting equal quantities of compromised nodes.
Hence, it is extremely significant to detect replica nodes in
an early stage.
A straightforward solution to the node replica problem is
to equip the tamper-proof hardware on each node in the
network against illegal loading of security materials and
malicious program rewriting. However such solution is much
too expensive for most of sensor network scenarios.
Another class of solutions [2-4] identifies the replica
nodes based on the location claims reported by the sensor
nodes themselves. These solutions deduce the location
anomalies based on the conflicts existing in the location
claims. Parno et al. [1] is thought to be the first work on the
detection of node replica attacks, in which the randomized
and line-selected multicast based detection schemes are
proposed for static sensor networks. In those two schemes,
nodes report the location claims that identify their physical
positions and attempt to find out the conflict reports that
indicate that one node in multiple locations. Conti. et al. [3]
then propose a improved scheme to enhance the linear-
selected multicast scheme of [1] in terms of replica detection
probability, as well as storage and computation overheads by
using trusted random values. Ho et al. [2] propose two
distributed approaches which take advantage of group
deployment knowledge to reduce the communication,
computation and storage overheads. However, these
location-claim-based schemes are vulnerable to the falsified
location claims generated by the replicas. The replica nodes
manage to elude the detection by reporting the same location
as the original compromised nodes to the basestation.
To address the limitation of the existing research, we
propose a novel location-similarity-based detection scheme
against node replica attacks in WSNs. The basic idea behind
the proposed scheme is that it is reasonable to treat a node as
replica when its difference between the location claim and
true position is far beyond the reasonable range. However,
the exact physical position of each node is hard to get due to
the various potential uncertainties existing in WSNs. In order
to solve this problem, we design a new metric named
location similarity based on the knowledge of deployment
and neighborhood to quantify the position deviation between
the node’s true position and location claim, and proposed
threshold decision scheme to identify the replica nodes based
on the location similarity. Additionally, we proposed a
Locality Sensitivity Hashing (LSH) based algorithm to
reduce the complexity of the metric computation and
threshold comparison to a level of several bit XOR
operations.
The rest of paper is organized as follows: in Section II,
2016 9th International Symposium on Computational Intelligence and Design
2473-3547/16 $31.00 © 2016 IEEE
DOI 10.1109/ISCID.2016.127
56