Bio-inspired link quality estimation for wireless mesh networks

In this paper, the problem of estimating the link quality in mesh networks has been considered. Such a process is a major task to develop an efficient network layer, since it allows routing protocols to efficiently use neighbors as relays for multi-hop communications. In the last years, a number of link-quality aware routing metrics have been proposed and analyzed. However, such metrics usually adopt simple link-quality estimators based on moving average filters, which lead to poor performances due to their static nature. In this paper, we propose to improve the estimation of the link quality resorting to a bio-inspired estimator based on the neural network paradigm. The effectiveness of the proposal has been proved by means of a numerical performance comparison between the proposed estimator and the traditional ones under several environmental conditions.