Access Management Mechanism In Air-Ground Integrated Networks

With the growing of communication demands,air-ground integrated network can effectively address the bottlenecks encountered in existing wireless communication system as a promising communication structure with low cost,high scalability,and swift deployment.Nevertheless,the node heterogeneity and topology changes in air-ground integrated network pose significant challenges to designing an efficient and adaptive access management mechanism.How to provide personalized services to users and how to achieve flexible and reasonable resource allocation have become the key issue in the development of air-ground integrated networks.Fortunately,the development of artificial intelligence technology provides novel approaches to solve these problems.Therefore,this paper studies the base station deployment strategy,network selection strategy,and resource allocation strategy under the access management mechanism from the perspective of intelligence.The main contributions of this thesis are summarized as following:First,the intelligent deployment mechanism of lift-off mobile base stations is studied.Since the coverage range,movement speed,and resource capacity of lift-off base stations are limited,and mobile terminals’ locations dynamically change in real-time,poor deployment strategy can prevent users from obtaining effective signal coverage and communication services for a long time.To address this challenge,this paper proposes an intelligent deployment strategy based on the MADDPG framework and the attention mechanism to achieve dynamic and optimal adjustment of lift-off base station deployment locations.Numerical results show that the algorithm can not only expand the network coverage but also effectively enhance the long-term transmission performance under different network scales.Second,the user access selection strategy in the air-ground integrated network is investigated.Due to the difference in network ability between lift-off base stations and ground base stations,nodes have to select appropriate site for network access based on site ability and user demands when they are covered by multiple access sites.Therefore,this paper proposes an access selection algorithm based on DQN,and an intelligent network optimization access selection algorithm with the long-term goal of maximizing user satisfaction is also designed by comprehensively considering service characteristics,network attributes,and user preferences.The simulation results show that the designed algorithm outperforms traditional heuristic algorithms in terms of user access success rate and network transmission performance.Finally,the time slot resource allocation mechanism of the lift-off base station is studied in this paper.The limited time slot resources require dynamic adjustments of the resource allocation strategy to allocate free time slots to users for service transmission.Therefore,this paper firstly proposes a heuristic time slot resource allocation scheme that supports multiple services according to different services’ Quality of Service(QoS)requirements.Furthermore,an improved heuristic time slot allocation algorithm is developed through the optimal design of control time slots and priority criteria.On the basis of above scheme and algorithm,an Actor-Critic based intelligent time slot resource allocation algorithm is proposed by combining the deep reinforcement learning algorithm.Simulation results validate that the proposed dynamic resource allocation algorithm can meet the transmission requirements of each service,and the intelligent algorithm can effectively reduce the service delay and improve the access success rate compared with the heuristic algorithm.