Study On Non-uniform Constellation Optimization Based On Group Intelligence Algorithm

The global informationization and the transmission of massive information drives the rapid development of communication technology.The modulation and demodulation module in the physical layer serves as the basis and key module of the communication system,and has an important influence on the accuracy of the entire system information transmission.The constellation mapping process in the modulation module is a necessary step of symbol modulation,and the optimization of the constellation performance plays an important role in improving the demodulation effect.At present,the constellation diagram commonly used in LTE systems is a normalized QAM uniform constellation,and under the AWGN channel condition,the uniform constellation point distribution is not optimal.If a certain algorithm is adopted to optimize the uniform constellation distribution,it can be made.The system obtains better demodulation error rate performance under the same channel and parameter conditions.Nowadays,non-uniform constellation diagrams have become one of the key technologies discussed by 3GPP.If we want to further improve the performance of the non-uniform constellation,we need to start with the specific implementation algorithm,but there are few literatures on the specific optimization algorithm and implementation of the non-uniform constellation.Based on the above background,this paper studies and designs the optimization scheme of the non-uniform constellation diagram.The main work completed and the simulation results achieved are as follows:(1)Design a non-uniform constellation optimization scheme based on genetic algorithm,and design and improve the key modules such as selection operator,crossover operator and mutation operator in the scheme to determine the optimal operator type.Research and improvement on the influence of modulation order and the type of fitness function algorithm.The optimal parameter configuration is finally determined and verified for simulation of the dermodulation error rate performance.Under the optimal parameter conditions and AWGN conditions,the optimized non-uniform constellation error perfomrmance is improved by about 4-5 percentage points.(2)Designing a non-uniform constellation optimization scheme based on particle swarm optimization algorithm,studying the effects of inertia weight coefficient,learning factor,evaluation function and modulation order,on the particle swarm position update algorithm and the best historical position update module.The design and improvement are elaborated and verified by simulating the error rate performance.Under the optimal parameter conditions and AWGN conditions,the optimized non-uniform constellation error performance is improved by about 3-4 percentage points.(3)Compare different non-uniform constellation optimization schemes,and compare the similarities and differences between the two groups of intelligent algorithms and gradient descent algorithms.This paper can be used as a reference for the further improvement of the non-uniform constellation optimization scheme and the further optimization of the performance of the non-uniform constellation,which is effective and practical.