[Paper Review] Online Channel Assignment in Multi-Radio Wireless Mesh Networks Using Learning Automata

Ziaeddin Beheshtifard, Mohammad Reza Meybodi|arXiv (Cornell University)|May 14, 2014

Mobile Ad Hoc Networks17 references3 citations

TL;DR

This paper proposes a learning automata-based online channel assignment scheme for multi-radio wireless mesh networks that dynamically optimizes channel selection to maximize network throughput and minimize interference. By using a distributed, adaptive mechanism that evaluates upstream link capacity and interference levels, the algorithm converges to a stable, optimal channel configuration with proven convergence via Lyapunov analysis.

ABSTRACT

In this paper, we look into the problem of channel assignment in multi-channel multi-radio wireless mesh networks. We propose a new learning automata based channel assignment scheme that adaptively improve network overall throughput by expecting channel state. Since the ability of sending packets via upstream links will be evaluation bases for assigning channels to radio interfaces on each node. We use a link capacity function that potentially reflects degree of interferences imposed by selected channels by each node. According to dynamics of system, proposed algorithm assigns channels to radio interface in distributed fashion such that minimize interference in neighborhood of a node. We analyze the stability of the system via appropriate Lyapunov-like trajectory; we show that stability and optimum point of the system is converged.

Motivation & Objective

• To address the challenge of dynamic channel assignment in multi-radio wireless mesh networks with high interference and varying traffic loads.
• To improve network throughput by enabling each node to adaptively select optimal channels based on real-time interference and link capacity feedback.
• To design a distributed, online channel assignment algorithm that minimizes interference in the neighborhood of each node without centralized coordination.
• To ensure system stability and convergence to an optimal channel assignment through formal Lyapunov-based analysis.

Proposed method

• The algorithm uses learning automata to model each node’s channel selection process, where actions correspond to channel assignments.
• Each node evaluates the performance of its selected channel based on upstream link capacity, which reflects interference and transmission success.
• A reward-penalty mechanism updates the learning automaton’s action probabilities based on observed link performance, favoring high-throughput, low-interference channels.
• The system employs a Lyapunov-like trajectory to mathematically prove convergence to a stable, optimal channel assignment state.
• Channel assignment decisions are made independently per radio interface, enabling distributed operation without global knowledge.
• The link capacity function quantifies interference impact, allowing nodes to estimate the quality of channel choices in real time.

Experimental results

Research questions

• RQ1 How can channel assignment in multi-radio wireless mesh networks be optimized in a distributed, online manner to maximize throughput?
• RQ2 What learning mechanism enables nodes to adaptively select channels that minimize interference and improve link performance?
• RQ3 Can a learning automata-based approach ensure convergence to a stable and optimal channel assignment under dynamic network conditions?
• RQ4 How does the system maintain stability and avoid oscillations during online channel reassignment?

Key findings

• The proposed learning automata-based channel assignment scheme achieves stable convergence to an optimal channel configuration, as proven via Lyapunov analysis.
• The algorithm effectively reduces interference in the neighborhood of each node by dynamically adapting channel choices based on real-time link performance.
• Throughput improvements are achieved by prioritizing channels with higher upstream link capacity and lower interference levels.
• The distributed nature of the algorithm eliminates the need for centralized control or global channel state information.