Performance on Autoencoder-Based MIMO Quantize-Forward Relay System for Various Learning Parameters

In multiple input multiple output (MIMO) quantize-forward (QF) relay systems, an autoencoder comprising an encoder, a decoder, and a channel component has been employed, demonstrating commendable performance. In the QF relaying, the relay quantizes the phases of received signals and forwards them to the destination. A neural network is subsequently integrated into the relay after quantization, introducing a non-linear beamforming effect. In assessing the efficacy of the autoencoder-based MIMO QF relay system applying phase quantization with neural network at the relay, we conduct a comprehensive analysis of bit error rates. This evaluation compares system performance related to diverse learning parameters, such as batch size, number of epochs, and neural network size at the relay. Simulation results clearly illustrate that these learning parameters significantly influence the overall performance of the system.