Decode-and-Forward Based Cooperative MIMO System using Deep Learning for Wireless Body Area Networks

In this article, we introduce a Decode-and-Forward (DF) based cooperative Multiple-Input Multiple-Output (MIMO) system using autoencoder (AE) technique, abbreviated as AE-DF, for Wireless Body Area Network (WBAN). In our scheme, the source, relay, and destination nodes are designed using neural networks, forming an AE. The learning parameters of the AE are trained synchronously through two sequential phases, allowing for optimal synchronization of the entire system. Additionally, in order to effectively reduce co-channel interference (CCI) in the received signal streams and enhance the bit error rate (BER) performance of the AE-DF system, we introduce the Minimum Mean Square Error Network (MMSEnet) detector employing deep learning at the relay and destination node. We assess the performance of these systems in scenarios with perfect channel state information (CSI) and imperfect CSI. The simulation results demonstrate that our system significantly outperforms the baseline system which is the conventional DF-based cooperative MIMO system using Minimum Mean Square Error (MMSE) detector. © 2023 IEEE.