Dishonest reporting in queue-based cross-layer network optimization

Queue-based cross-layer optimization algorithms have recently been a subject of intensive research in wireless networks. Their purpose is to guarantee stable operation and to achieve some form of fairness among users, whenever the traffic demand exceeds network capacity. Despite the plethora of work in this field, the scenario where one or more nodes declare false queue backlog values in order to gain throughput advantage remains unexplored. In this paper we examine this type of selfish misbehavior, concentrating on a specific class of algorithms, the so-called quadratic Lyapunov-function-based algorithms (QLA). In particular, the effect of backlog misreporting on a single-hop access network with contending stations is evaluated through simulations. A simple framework for the detection of misbehaving nodes is proposed, under the assumption that the access-point is aware of the utility functions of the stations. The detection approach exploits the fact that under QLA the throughput of a node must be approximately equal to an “expected” value, derived from the reported queue backlogs.