Wireless channel prediction using ensemble of Extreme Learning Machines

Abstract

In this article, we examine the possibilities and provide justification for extreme learning machines (ELMs) ensemble application in prediction of wireless channel condition. Single-input single-output (SISO) system in environments classified as microcellular and picocellular is used for analysis of the prediction model. Effectiveness and accuracy of ensemble based ELM algorithm to predict signal-to-noise ratio (SNR) in the channel is confirmed using, as performance indicators, the normalized mean squared error (NMSE) and time consumption. Moreover, the ensemble can effectively improve the generalization of the model compared to the single ELM. The results also show that ELM scan generates diverse prediction results, even when using the same training set.