Design Of Resource Scheduling Algorithm For Maritime Wireless Network

The strategic concept of E-Navigation driven by the demand of maritime communications has attracted wide attention from academics and industrial circles in various countries since its introduction.China is also actively formulating E-Navigation strategic development plans.In order to solve the bandwidth problem,cognitive radio technology is applied to marine communication systems.Resource scheduling is one of the key technologies of cognitive radio.Graph coloring theory and Q-Learning algorithm are two commonly used methods in the cognitive network resource scheduling.They have good adaptability to the centralized and distributed maritime cognitive network resource scheduling.The dissertation takes the maritime broadband wireless network as background,and proposes corresponding resource scheduling algorithms for different application scenarios.The main research contents include:1)Based on the introduction of cognitive radio resource scheduling technology and reinforcement learning theory,the network topology characteristics and application environment of the maritime communication network are analyzed in depth,and the maritime cognitive network is divided into two scenarios.Under this premise,the problems faced when designing resource scheduling algorithms are presented.2)A satisfaction fairness algorithm based on graph coloring theory for centralized maritime cognitive networks is proposed.The algorithm and the sensitized graph coloring algorithm are compared in terms of users' average satisfaction,fairness of satisfaction,total system benefit,and vulnerable users.The simulation results show that the satisfaction fairness algorithm has a better effect in fairness and other aspects.3)According to the characteristics of remote coast scenes in the maritime wireless communication network,the disadvantages of the traditional Q-Learning-based resource scheduling algorithm under the dynamic change of channel service are analyzed.Based on this,a resource scheduling algorithm applied to the dynamic channel service of distributed marine cognitive networks is proposed.The simulation results verify the significant advantages of the algorithm in dynamic channel service feature scenarios.4)Based on the independent Q-Learning resource scheduling algorithm and the idea of multi-agent system.A collaborative Q-Learning resource scheduling algorithm is proposed for multi-user distributed distributed maritime cognitive networks.Simulation and comparison of this algorithm and the independent Q-Learning resource scheduling algorithm show that this algorithm has better performance in spectrum conflict rate and total system capacity.