Research On Key Technologies For The Convergence Of Communication And Computing

The future-oriented B5G and 6G networks achieves global threedimensional coverage of wireless access through the convergence of spaceair-ground network,providing seamless roaming and consistent services for mobile users and ensuring the reliability of network services and green intensive resource.In the converged space-air-ground network,users evolving application and service demands and the generated,distributed data and computing needs drive the deep integration of communication and computing,which deploying computing nodes at different levels in the mobile network,the cloud-edge-end collaboratively processes the massive data and computational demands generated by users,thus reducing the network transmission traffic,accelerating local data processing decisions,optimizing the efficiency of resource utilization,reducing the cost of network infrastructure construction,and meeting users real-time analysis and personalized needs.The research on the key technologies for the convergence of communication and computing includes cloud-edge collaborative computing,joint optimization of communication and computing resources,and integrated mobility management technology,etc.,which collaboratively utilizes the communication and computing capabilities of cloud computing centers,edge computing nodes and user equipment to reasonably match user needs and wireless environments to meet the changing needs of emerging services.However,the current network infrastructure cannot meet the needs of all vertical fields,communication and computing resources are always limited,the convergence of communication and computing has challenges in the realtime analysis and processing of distributed data,dynamic migration of user computing tasks,which needs to consider the dynamic changes of user service requirements,application scenarios,channel conditions,computing power,etc,and only useful data can be uploaded to cloud computing centers to ensure continuity of service and low-cost coverage worldwide to meet the explosive traffic and connectivity growth needs of future networks.In order to improve the performance of user equipment and user experience,and expand the coverage of user services,this paper focuses on three aspects:the cloud-edge collaborative low-rank singular value decomposition regularized discriminant method in mobile communication,the joint Lyapunov optimization method for satellite-ground collaboration in satellite communication,and the improved adoptive niche genetic alternating joint optimization method in UAV communication.The three research contents respectively consider the integration of communication and computing technologies in mobile communication,satellite communication and UAV communication to reduce network redundancy,migrate computing task,optiize network performance so that users can obtain low-cost,high reliability and low-latency computing services through network connection at any time and place.The main research results are as follows:(1)Considering that computation-intensive or delay-sensitive applications in mobile communications such as autonomous driving,telemedicine,virtual reality,augmented reality and security require frequent identification and perception and generate a large amount of real-time data,uploading to the cloud computing center will increase network burden and delay response time,and the local view and limited resources provided by edge computing nodes cannot meet users requirements for high-quality services.Therefore,this paper proposes a cloud-edge collaborative low-rank singular value decomposition regularized discriminant method for the distributed data processing problem generated by user equipment in the terrestrial network coverage scenarios.First,the global view and historical data provided by the cloud computing center are used to guide the edge computing nodes to extract the useful features of the data collected by the user equipment.Then,the edge computing node uploads the extracted features to the cloud computing center for further processing.Finally,the prototype system of cloud-edge collaboration is implemented with image recognition as an example,where feature matching is performed at the cloud computing center and recognition results are returned to the edge computing nodes and user equipment.The simulation results show that the proposed method can achieve similar or even better recognition accuracy than uploading the original image data with only a small amount of feature data,and reduce the network traffic and feature matching time.(2)Considering that the communication,computation resources and battery life of low-earth orbit satellites in satellite communication are limited,and the high-speed flight of satellites in low-earth orbits makes it difficult to obtain prior knowledge related to communication links and solar radiation energy harvesting processes.Meanwhile,the large amount of raw earth or space data identified and sensed by satellites are directly downloaded to the ground station for processing,which easily leads to longer time delay and energy consumption.Therefore,aiming at the problems of computation migration cost and energy harvesting in the scenario of low Earth orbit satellite network coverage,this paper proposes a Lyapunov dynamic computation migration method for handling satellite computing task.First,the satellite battery energy level constraint in the original problem of minimizing the system total computing task cost is converted to a virtual energy level queue,and the system total computing task cost of the optimization objective is converted to a penalty function.Then,the sum of the length drift of the queue backlog and the penalty function is used as the objective function to transform the long-term time average optimization problem into a deterministic optimization problem under the current time slot and reduce the complexity of the original problem.Finally,the optimized decision is made based on the deterministic state information of each time slot in satellite networks,so that the satellite battery virtual energy queue is in long-term time average stability after several iterations and the computing task cost reaches the optimization goal of long-term time average minimization.The simulation results show that the proposed method can reduce the task cost of the system and make the energy level of the satellite battery stable in long-term time average by reasonably setting the lower bound of satellite battery discharge,control parameters and time slot length.(3)Considering the impact of wireless channel quality on system performance,the characteristics of intelligent reflecting surface to reconfigure the wireless propagation environment,the characteristics of MIMO system beamforming,and the limitations of UAV payload and energy supply,when user equipment run frequent identification,sensing,and other computing tasks that require a large amount of computing power and cache or consume a large amount of energy,UAV can assist users in migrating computing tasks to nearby edge computing nodes of ground-based stations.Therefore,this paper proposes an improved adoptive niche genetic alternation optimization method for the problem of minimizing the delay of MIMO-based intelligent reflecting surface-assisted computation migration in UAV network coverage scenarios.First,when the wireless channel between user equipment and base stations in UAV communication is blocked by obstacles such as buildings,the intelligent reflecting surface composed of many low-cost passive reflecting elements is deployed in UAV to expand the intelligent reflecting surface coverage.Then,the intelligent reflecting surface control circuit obtains real-time location and channel status information of the user equipment through the communication components of the UAV to control the intelligent reflecting surface to assist the user equipment in computing migration.Finally,the active beamforming of the user equipment,the passive beamforming of the intelligent reflecting surface,the allocation of computing task,and the hovering position of the UAV are jointly optimized so that the user can migrate the computing task to the edge computing node of the base station.The Simulation results show that the proposed method can suppress genetic drift to a certain extent,maintain the population diversity,and provide better global search capability and convergence speed to reduce user task latency.