Distributed Machine-Learning for Early HARQ Feedback Prediction in Cloud RANs

In this work, we propose novel HARQ prediction schemes for Cloud RANs (C-RANs) that use feedback over a rate-limited channel (2 - 6 bits) from the Remote Radio Heads (RRHs) to predict at User Equipment (UE) decoding outcome BaseBand Unit (BBU) ahead of actual decoding. particular, Dual Autoencoding 2-Stage Gaussian Mixture Model (DA2SGMM) is trained in an end-to-end fashion whole C-RAN setup. Using realistic link-level simulations sub-THz band 100 GHz, show DA2SGMM scheme clearly outperforms all other adapted and state-of-the-art schemes. The shows superior performance terms blockage detection as well no-blockage single-blockage cases. with 4~bit achieves more than 200 % higher throughput average compared its best alternative. Compared regular HARQ, reduces maximum transmission latency by 72.4 %, while maintaining 75 scenario. scenario, significantly increases most evaluated Signal-to-Noise-Ratios (SNRs) HARQ.