Ultra-high reliable optimization based on Monte Carlo Tree Search over Nakagami-[formula omitted] Fading

Abstract

Supporting the ultra-reliable and low-latency communications (URLLCs) has become one of the major goals for future wireless networks. In this paper, we present an analytical reliability model for user equipment (UE) connecting multiple base stations (BSs) with multi-stream carrier aggregation (MSCA) technology. We first derive a closed-form expression for reliability characterization using signal-to-interference-plus-noise (SINR) model over Nakagami-m fading. We then formulate a joint resource allocation problem to maximize reliability, considering UE association, sub-carrier assignment and discrete power allocation. With distributed decision making (DDM) theory, we decouple it into two sub-problems, and each sub-problem is formulated as a separate Markov Decision Problem (MDP). We further propose the Monte Carlo Tree Search (MCTS) method to find the optimal solution for each sub-problem. The iterative joint optimization method based on DDM is also proposed. Our simulation results show that the reliability with MSCA increases with increasing the number of sub-carriers, as well as increasing the power allocation. We also show that our algorithm is an effective way in finding the optimal resource allocation for reliability improvement.