Cognitive Diversity Equalization for 2x2 HF MIMO Channels

Abstract

The ionosphere has been both a hindrance and attraction of communication solutions that require usage of the High Frequency (HF) band for propagation. While the ionosphere is able to propagate signals at long ranges, it can also introduce significant distortion via multipath and fading. One approach investigated by works in the literature as a means of helping improve communications in the HF band has been the usage/design of MIMO systems. Cognitive equalization can be used to further help mitigate these issues by using reinforcement learning to determine the optimal equalizer configuration (i.e. tap length, filter type, etc.) for a specific channel. The objective of this paper is to investigate the performance of cognitive equalization for a simulated 2x2 HF MIMO channel when the Epsilon-greedy and Softmax Strategy algorithms are used to learn the optimal equalizer structure when having access to Linear and Decision-Feedback equalizers in diversity Mode.