Optimal Planning of Emergency Communication Network Using Deep Reinforcement Learning

摘要

Aiming at the problems of traditional algorithms that requirehigh prior knowledge and weak timeliness, this paper proposes anemergency communication network topology planning method based ondeep reinforcement learning. Based on the characteristics of the emergencycommunication network, and drawing on chess, we map the nodelayout and topology planning problems in the network planning to chessgame problems; The two factors of network coverage and connectivity areconsidered to construct the evaluation criteria for network planning; Themethod of combining Monte Carlo tree search and self-game is used torealize network planning sample data generation, and the network planningstrategy network and value network structure based on residual networkare designed. On this basis, the model was constructed and trained basedon Tensorflow library. Simulation results show that the proposed planningmethod can effectively implement intelligent planning of network topology,and has excellent timeliness and feasibility.