Real-Time Detection of Selfish Behavior in IEEE 802.11 Wireless Networks

The open and distributed nature of the IEEE 802.11 based wireless network makes it easy for selfish nodes to gain unfair share on the networks by manipulating the protocol parameters. In this paper, we address the detection of such selfish behavior. The two main challenges associated with the detection problem are the unknown selfish behavior strategy and real-time detection of the behavior. While the two challenges are correlated for efficient detection, existing solutions can not address both of them well at the same time. In our work, we propose a new observation method monitoring the number of successful transmissions of the tagged node. This enables us to capture the short-term dynamic of the traffic behavior which is crucial for real-time detection. Integrating our the observation method with the CUSUM test, we develop a detection scheme to deal with both the challenges without any modification to the existing protocols. Moreover, we utilize a discrete Markov chain based model to characterize the behavior of the CUSUM test statistic, which enables us to quantitatively analyze the tunable parameters in the scheme for guaranteed detection performance. The performance of the proposed scheme is validated through ns-2 simulation. We show that the scheme is capable of quickly and accurately detecting the selfish behavior without knowledge of the selfish strategy.