Modeling Multi-Cell IEEE 802.11 WLANs with Application to Channel Assignment

TL;DR

This paper introduces an accurate analytical model for dense multi-cell IEEE 802.11 WLANs, enabling efficient channel assignment through a decentralized, throughput-based Nash equilibrium algorithm without requiring topology knowledge.

Contribution

The paper develops a scalable cell-level model for dense WLANs and proposes a simple decentralized channel assignment algorithm based on game theory and throughput optimization.

Findings

Model accurately predicts WLAN performance compared to ns-2 simulations.

Proposed channel assignment algorithm achieves Nash equilibria.

Algorithm requires no topology information or message passing.

Abstract

We provide a simple and accurate analytical model for multi-cell infrastructure IEEE 802.11 WLANs. Our model applies if the cell radius, $R$, is much smaller than the carrier sensing range, $R_{cs}$. We argue that, the condition $R_{cs} >> R$ is likely to hold in a dense deployment of Access Points (APs) where, for every client or station (STA), there is an AP very close to the STA such that the STA can associate with the AP at a high physical rate. We develop a scalable cell level model for such WLANs with saturated AP and STA queues as well as for TCP-controlled long file downloads. The accuracy of our model is demonstrated by comparison with ns-2 simulations. We also demonstrate how our analytical model could be applied in conjunction with a Learning Automata (LA) algorithm for optimal channel assignment. Based on the insights provided by our analytical model, we propose a simple decentralized algorithm which provides static channel assignments that are Nash equilibria in pure strategies for the objective of maximizing normalized network throughput. Our channel assignment algorithm requires neither any explicit knowledge of the topology nor any message passing, and provides assignments in only as many steps as there are channels. In contrast to prior work, our approach to channel assignment is based on the throughput metric.