Design And Implementation Of GPU High-Performance Computing In Large Scale Communication System Simulation

With the deepening and maturity of the research on 5G system,in addition to the improvement of communication technology,researchers are trying to seek a substantial improvement of 5G system performance by improving the architecture scale of the system.Therefore,large-scale communication system emerged.Large-scale communication system is a communication system with multiple cells,multiple users and multiple base stations,and the base stations use a large-scale antenna array with more than ten antennas and even hundreds of antennas and serve multiple users at the same time.However,Due to the large amount of simulation data of large-scale communication system,the simulation time of communication system is too long.In this paper,GPU(Graphics Processing Unit)high-performance computing is introduced in the process of large-scale communication system simulation to solve the "time bottleneck" problem of communication system simulation brought by large-scale data.The topic of the paper is selected in the national major project "A New Generation of Broadband Wireless Mobile Communication Network," the"Large-scale Complex Network Dynamic Simulation Verification Platform" issue.In order to solve the problem of long simulation time in large-scale communication system simulation,this paper presents a solution to GPU high-performance computing based on CUDA(Compute Unified Device Architecture)technology.The design and implementation of GPU high performance computing in large-scale communication system simulation are studied.The main work of the paper is as follows:First of all,review the large-scale communication system and GPU high performance computing,analyze the present status of large-scale communication system and its simulation platform,introduce GPU high performance computing based on CUDA architecture and analyze its principle of acceleration.Review the GPU high performance computing in large-scale communication system,and analyze its key simulation technologies,provide theoretical basis for the application of GPU high performance computing in large-scale communication system simulation.Then,the design objectives and design principles of GPU high performance computing in large-scale communication system simulation are clarified.By analyzing the existing simulation platform of large-scale communication system,select initialization module and interference calculation module in large-scale communication system to simulate in GPU.Complete GPU high-performance computing design for initialization and interference calculation modules.Finally,the realization process of GPU high performance computing in large-scale communication system simulation is clarified.Based on the commonality and characteristics of the initialization module and the interference calculation module in GPU high-performance computing,are described respectively.The GPU accelerated computing module is added to a large-scale communication system simulation with a CPU(Central Processing Unit)as a main development environment to realize simulation acceleration of a large-scale communication system as a whole.Then,the acceleration result of the sub-module and the overall acceleration are obtained:the acceleration effect of the sub-module is 40 times to several hundred times,the whole acceleration effect is 38%and 43%,respectively,in the two acceleration cases of single GPU and multi-GPU.Finally,the simulation results are analyzed.