Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
430975 | Journal of Discrete Algorithms | 2007 | 19 Pages |
The IEEE 802.11 MAC protocol now prevailing in wireless LANs is vulnerable to selfish backoff attacks consisting in selection of short backoff times in the constituent CSMA/CA procedure. Administrative prevention of such attacks fails in ad hoc configurations, where stations' behavior cannot be mandated. In this paper we take an incentive-oriented game-theoretic approach whereby stations are allowed to maximize their payoffs (achieved success rates). Using a fairly accurate performance model we show that a noncooperative CSMA/CA game then arises with a payoff structure characteristic of a Prisoners' Dilemma. For a repeated CSMA/CA game, a novel SPELL strategy is proposed and shown to yield to simple algorithmic design. Assuming that the stations are rational players and wish to maximize a mean-value-type long-term utility, SPELL is further shown to deter a single attacker by providing a disincentive to deviate from SPELL.