Channel Sensing Algorithm Based on Neural Networks for Cognitive Wireless Mesh Networks

Spectrum is always a kind of rare resource in wireless communication especially in wireless mesh networks (WMNs) which own the wireless connectivity backbone. But in most cases we seldom take full advantage of the spectrum. This problem can be alleviated by the cognitive radio paradigm that aims at devising spectrum sensing and management techniques, thereby allowing radios to intelligently locate and use frequencies other than those in the ISM band. In cognitive radio, channel sensing should be able to detect the signals from other systems which are using the same frequency band in order to avoid the mutual interference, which is so important that determine the cognitive radio success. In this paper, we propose a channel sensing algorithm based on ART-2 (adaptive resonance theory) neural networks which well satisfy the WMNs structure. Simulation results show that the proposed algorithm can give the more correct sensing results and remarkably improves the cognitive WMNs performance.