Attack and Flee: Game-Theory-Based Analysis on Interactions Among Nodes in MANETs

[attachment=18317]

I. INTRODUCTION
THE COLLABORATION between the participants is the
foundation for mobile ad hoc networks (MANETs) to
achieve the desired functionalities. The topologies in MANETs
change dynamically because of node movement. Nodes in
MANETs usually have no predefined trust between each other.
Moreover, all nodes tend to maximize their own utility (also
referred to as payoff) in activities. Among existing research,
different mechanisms (e.g., reputation systems, virtual currency,
and barter economy) have been developed to stimulate
cooperation and mitigate nodes’ selfish behavior.
Aside from regular nodes’ selfish behavior, malicious nodes
also exist in the network. The common objective of malicious
nodes is maximizing the damage to the network while avoiding
being caught. Their utility comes from activities that disrupt
the operation of the network and waste the resources of regular
nodes.


RELATED WORK
The incentives for nodes to cooperate are analyzed and
presented in [1]–[3]. However, in these works, malicious nodes
are modeled as never cooperative, without any further sophistication,
since their main focus was discouraging selfish
nodes. There is no degree of selfishness that can approximate
the behavior of malicious nodes. In this paper, we model the
malicious nodes with their own utility functions, which are
different from regular nodes. In other words, we assume that
malicious nodes are also rational concerning their goals.
Some recent works have studied the incentives for malicious
nodes and modeled their behavior more rationally.


REGULAR/MALICIOUS NODE GAME
We model the regular/malicious node game as a multistage
dynamic Bayesian signaling game to find the optimal strategy
of regular and malicious nodes.


COUNTERMEASURES
The regular node needs to balance the possible loss for false
alarm and gain in order to yield a correct report. It needs an
evidence accumulation process to make a confident reporting
decision. The malicious node clearly gains advantages by fleeing
before the end of this process. Therefore, shortening the
length of this process and making it less predictable become
the networks’ main countermeasures against malicious nodes.