Deep Learning-Based Estimator for Fast HARQ Feedback in URLLC

Autonomous systems and mission-critical applications demand ultra-reliable low-latency communication (URLLC). To build wireless communication networks capable of accommodating such applications, optimization of the airinterface characteristics is vital. This paper leverages recent advancements in the field of Artificial Intelligence (AI) technologies to optimize specific aspects of the air interface design to satisfy these stringent link reliability and latency requirements. The precise aim of this research is to reduce the link latency caused by the presence of the Hybrid Automatic Repeat reQuest (HARQ) mechanism. To this end, we propose a novel deep learning-based algorithm (Deep-HARQ), employing a deep neural network (DNN) with fully connected layers to estimate the decodability of the coded-received in-phase and quadrature (I/Q) signals prior to accomplishing the majority of the complex reception tasks. This enables the receiver to respond faster, allowing for the reduction of the signal round-trip time (RTT). To evaluate Deep-HARQ with a realistic dataset, we collected training and validation samples from a waveform compatible with 3GPP 5G NR Release 15 standards. The simulation results reveal a faster estimation response, with an accuracy enhancement of 12% compared to relevant algorithms in the literature.