Optimal Design Of Receiving Module For Wireless Communications Based On Machine Learning

Due to the increasing demand for mobile communication network capacity in the future,researchers have carried out a lot of research to support large capacity communication.Among them,the multiple input multiple output(MIMO)technology is considered to be one of the most promising technologies.However,the application of MIMO technology still faces many problems.It is especially difficult to provide an efficient detection algorithm for the receiver.Although maximum likelihood(ML)detection is able to obtain the theoretically lowest bit error rate,its computational complexity significantly increases as the number of antennas increases.It is difficult to provide a detection algorithm that can obtain a lower bit error rate with lower computational complexity,which is a long-term pain point in the field of MIMO technology.At the same time,artificial intelligence technology represented by machine learning has developed rapidly in recent years and has gradually become a general mathematical tool that can solve complex problems.This technology is particularly suitable for solving multi-classification problems.Classification problems have natural similarities with signal detection.This paper mainly studies the scheme of using machine learning algorithms to improve traditional detection algorithms.The detailed works are as follows:1.This paper introduces the research status and progress of MIMO signal detection in recent years and analyze the application prospects of artificial intelligence technology represented by machine learning in the field of signal detection.2.Machine learning algorithms are data-driven while traditional communication theories are model-driven,so machine learning algorithms cannot be simply applied to communication systems.In this paper,the author analyzes the problems and difficulties of neural network in the application of communication signal detection.Then,a neural network based intelligent detection scheme for constant channels is proposed.Specifically,by sorting and dividing the channel matrix,the signal detection process is divided into two parts: linear detection and non-linear detection.The suitable algorithms are used for each part.By combining the neural network algorithm and the traditional communication model,the bit error rate and the computational complexity of the algorithm can be balanced.Compared with traditional algorithms,the algorithm proposed in this paper can obtain a lower bit error rate and the bit error rate and computational complexity can be adjusted by adjusting the parameters.3.Two MIMO signal detection algorithms for multiusers uplink signal detection under fading channels are proposed in this paper.The first algorithm is the algorithm improved from the algorithm under the constant parameter channel in this paper,so that it can be used under fading channel.The second algorithm is based on the block-diagonalization(BD)detection algorithm.By converting the channel matrix of the transmission system into an equivalent channel matrix for each user,the problem scale of the detection algorithm is reduced and the users' interference is suppressed.Then,a small neural network is trained to detect the signal for each user individually.Compared with traditional algorithms,detection schemes using intelligent algorithms perform better in simulations.At the same time,this paper gives separate simulations to the intelligent detection algorithm under different parameters and makes necessary explanations on the performance,advantages and disadvantages of the algorithm based on the experimental results.Finally,the reinforcement learning schemes are introduced and discussed.Based on the above discussion,the limitations of artificial intelligence technology represented by machine learning in the field of wireless communication signal detection is discussed.Machine learning technologies should not stack calculation models and cost long-term tuning to obtain good performance in communication systems.And model-driven communication theory can make up for the lack of interpretability of intelligent algorithms.