An adaptive channel assignment in wireless mesh network: The learning automata approach

In wireless mesh networks, random changes in the environment can increase the complexity of the multi-channel assignment. In this work, a new channel assignment scheme based on learning automata is proposed, which adaptively improves the network's overall performance by predicting network dynamics. First, we use a practical utility function that reflected the user's preference regarding the signal-to-interference-and-noise ratio is applied. In the multi-automata learning algorithm, each user evaluates a channel selection strategy by computing a utility value in a stochastic iterative procedure. The utility function that potentially reflects a measure of satisfaction is used by every node as an environmental response to the current selected strategy. In the proposed algorithm, by changing network traffic pattern, the channel allocation varies adaptively with dynamic conditions of the network. Extensive simulation-based evaluation of our algorithm demonstrates that the proposed algorithm converges to an equilibrium point, which is also optimal for channel assignment policy.