Wavelet Neural Network Based Link Quality Prediction for Fluctuating Low Power Wireless Links

Abstract

Low power wireless links are prone to fluctuate when the channel environment changes. In order to reduce the impact of link fluctuations on data transmission, it is necessary to predict the link quality quickly and accurately and make dynamic adjustments according to prediction results. However, existing link quality prediction mechanisms lack sufficient consideration of the impact of link fluctuations, which leads to high prediction errors under the links with large fluctuations such as moderate and sudden changed links. In response to this problem, this paper proposed WNN-LQP, a more effective link quality prediction mechanism under the links with large fluctuations. By taking advantage of the higher resolution of link quality indicator in the transition region as well as the stronger learning ability and higher prediction accuracy of wavelet neural network, WNN-LQP could reduce the prediction errors under moderate and sudden changed links effectively. Compared with the similar mechanism, its prediction errors are reduced by 26.9% under both moderate and sudden changed links.