Combined Relay Selection Enabled by Supervised Machine Learning

Combined relay selection only requires two relays to forward signals transmitted on multiple subcarriers, but the optimal outage performance is almost surely achievable in high signal-to-noise ratio (SNR) region. However, because combined involves generation of full set two-relay combinations, complexity much higher than that per-subcarrier when number goes large. This drawback restricts implementation dense networks. To overcome this drawback, we propose enable by supervised machine learning (ML). Because training procedure off-line, proposed scheme can considerably reduce and processing latency. We carry out extensive experiments TensorFlow 2.1 over a graphics unit (GPU) aided computing cloud server validate effectiveness scheme. The experimental results confirm ML provide near-optimal with lower latency well matches provided brute-force search manner.