Dynamic Frequency Management in 802.11-based Multi-Radio Wireless Networks

TL;DR

This paper introduces ARACHNE, a distributed, routing-aware channel selection protocol for 802.11 mesh networks that optimizes end-to-end performance by intelligently assigning channels based on load and interference, significantly improving throughput.

Contribution

The paper presents ARACHNE, a novel, distributed channel selection protocol that considers end-to-end link loads and interference, addressing a gap in existing solutions for mesh networks.

Findings

ARACHNE approaches optimal channel selection in simulations.

The protocol improves total network throughput over existing methods.

It effectively manages interference and load balancing across routes.

Abstract

Efficient channel selection is essential in 802.11 mesh deployments, for minimizing contention and interference among co-channel devices and thereby supporting a plurality of QoS-sensitive applications. A few protocols have been proposed for frequency allocation in such networks, however they do not address the problem end-to-end. In this paper, we present a general formulation of the channel selection problem taking into account the performance of both mesh-access and mesh-backhaul. Moreover, we propose ARACHNE, a routing-aware channel selection protocol for wireless mesh networks. ARACHNE is distributed in nature, and motivated by our measurements on a wireless testbed. The main novelty of our protocol comes from adopting a metric that captures the end-to-end link loads across different routes in the network. ARACHNE prioritizes the assignment of low-interference channels to links that (a) need to serve high-load aggregate traffic and/or (b) already suffer significant levels of contention and interference. Our protocol takes into account the number of potential interfaces (radios) per device, and allocates these interfaces in a manner that efficiently utilizes the available channel capacity. We evaluate ARACHNE through extensive, tracedriven simulations and we show that approaches the optimal channel selection. We observe that our protocol improves the total network throughput, as compared to three other representative channel allocation approaches in literature.