Cross-Layer Optimization Using Game Theory to Alleviate Unfairness in Wireless Networks

Abstract

The IEEE 802.11 Distributed Coordination Function (DCF) does not require the presence of a central controller or access point. It is therefore, the standard protocol used for resolving medium access contention in mobile ad hoc networks. IEEE 802.11 DCF is based on carrier sense multiple access with collision avoidance and Binary Exponential Back off Algorithm (BEBA). BEBA can improve the system throughput but increases the capture effect, permitting one node to seize the channel. This is because BEBA inherently favours the last successful node by providing it a smaller contention window after each successful transmission. This unfairness issue at the Medium Access Control (MAC) layer also affects the Transmission Control Protocol (TCP) layer as it leads to the most active connection dominating the shared channel. We have two types of instability issues that arise -- intraflow and interflow instabilities. In this paper, we take a game theory based approach for solving these instability issues that arise as a result of the capture effect. A modified back off mechanism for the DCF game called Optimized Back off Mechanism (OBM), in which nodes adopt smooth dynamics in changing their channel access probabilities by taking into account the actions of the other nodes in the neighborhood has been proposed. To ensure that no particular node gains undue access to the channel, we design the mechanism so as to achieve Nash equilibrium in the neighborhood. A node which unilaterally increases its channel access probability will eventually be penalized and incur a higher payoff. The mechanism will prevent a particular node or flow from dominating the shared channel in the intraflow and interflow scenarios respectively.