A Distributed Neural Linear Thompson Sampling Framework to Achieve URLLC in Industrial IoT

One of the most prominent requirements of future Industrial Internet of Things (IIoT) networks will be to provide Ultra-Reliable Low-Latency Communication (URLLC) in support of critical physical processes underlying the production chains. However, standard protocols for allocating wireless resources may not be able to optimize the latency-reliability trade-off, especially for uplink communication. For example, centralized (e.g., grant-based) scheduling can ensure almost zero collisions, but introduces delays in the way resources are requested by the User Equipments (UEs) and then granted by the Next Generation Node B (gNB). On the other hand, distributed scheduling (e.g., based on random access), in which UEs autonomously choose the physical resources to transmit uplink data, may lead to potentially many collisions especially when the density of UEs (and so the traffic) increases. Along these lines, in this work we propose DIStributed combinatorial NEural linear Thompson Sampling (DISNETS), a novel scheduling framework that combines the best of the two worlds. By leveraging a feedback signal sent from the gNB and reinforcement learning, the UEs are trained to autonomously optimize their uplink transmissions by selecting the available physical resources so as to minimize the number of collisions, disaggregated from the network and without additional message exchange to/from the gNB. DISNETS is a distributed, multi-Agent adaptation of the Neural Linear Thompson Sampling (NLTS) algorithm, which has been further extended to admit multiple actions in parallel. We apply DISNETS to the context of IIoT, and demonstrate by extensive simulations the superior performance of the proposed approach in addressing URLLC compared to other baselines.