Secure Data Collection in Wireless Sensor Networks Using Randomized Dispersive Routes



  T-secure data collection in wireless sensor network tranction.pdf (Size: 1.84 MB / Downloads: 6)

INTRODUCTION
Motivations

OF the various possible security threats encountered in a
wireless sensor network (WSN), in this paper, we are
specifically interested in combating two types of attacks:
compromised node (CN) and denial of service (DOS) [22].
In the CN attack, an adversary physically compromises a
subset of nodes to eavesdrop information, whereas in the
DOS attack, the adversary interferes with the normal
operation of the network by actively disrupting, changing,
or even paralyzing the functionality of a subset of nodes.
These two attacks are similar in the sense that they both
generate black holes: areas within which the adversary can
either passively intercept or actively block information
delivery. Due to the unattended nature of WSNs, adversaries
can easily produce such black holes [1]. Severe CN
and DOS attacks can disrupt normal data delivery between
sensor nodes and the sink, or even partition the topology.


Contributions and Organization

The key contributions of this work are as follows:
1. We explore the potential of random dispersion for
information delivery in WSNs. Depending on the
type of information available to a sensor, we develop
four distributed schemes for propagating information
“shares”: purely random propagation (PRP),
directed random propagation (DRP), nonrepetitive
random propagation (NRRP), and multicast treeassisted
random propagation (MTRP). PRP utilizes
only one-hop neighborhood information and provides
baseline performance. DRP utilizes two-hop
neighborhood information to improve the propagation
efficiency, leading to a smaller packet interception
probability.


RANDOMIZED MULTIPATH DELIVERY
Overview

As illustrated in we consider a three-phase approach
for secure information delivery in a WSN: secret sharing of
information, randomized propagation of each information
share, and normal routing (e.g., min-hop routing) toward
the sink. More specifically, when a sensor node wants to
send a packet to the sink, it first breaks the packet into
M shares, according to a ðT;MÞ-threshold secret sharing
algorithm, e.g., Shamir’s algorithm [20]. Each share is then
transmitted to some randomly selected neighbor.


Random Propagation of Information Shares

To diversify routes, an ideal random propagation algorithm
would propagate shares as dispersively as possible.
Typically, this means propagating the shares farther from
their source. At the same time, it is highly desirable to have
an energy-efficient propagation, which calls for limiting the
number of randomly propagated hops. The challenge here
lies in the random and distributed nature of the propagation:
a share may be sent one hop farther from its source in a
given step, but may be sent back closer to the source in the
next step, wasting both steps from a security standpoint. To
tackle this issue, some control needs to be imposed on the
random propagation process.


ASYMPTOTIC ANALYSIS OF THE PRP SCHEME
The random routes generated by the four algorithms in
Section 2 are not necessarily node-disjoint. So, a natural
question ishow good these routes are in avoiding black holes.
We answer this question by conducting asymptotic analysis
of the PRP scheme. Theoretically, such analysis can be
interpreted as an approximation of the performancewhenthe
node density is sufficiently large. It also serves as a lower
bound on the performance of the NRRP, DRP, and MTRP
schemes. Note that the security analysis for the CN and DOS
attacks are similar because both of them involve calculating
the packet interception probability. For brevity,weonly focus
on theCNattack model. The same treatment can be applied to
the DOS attack with a straightforward modification.