Joint Design And Implementation Of Precoding And Power Allocation For Massive MIMO System

The development of the Internet of Things continues to drive wireless communication into every aspect of human social life,it not only brings a new direction of economic development to countries,but also puts forward unprecedented new demands on the performance of communication system.As one of the core technologies to promote the performance of 5G communication,massive Multiple Input Multiple Output(massive MIMO)technology has achieved significant improvement in energy efficiency and spectral efficiency by deploying a very large number of antennas at the base station,prompted massive MIMO technology becomes a key research hotspot in the field of communication.The signals sent by multiple antennas in MIMO systems will interfere with each other.In order to reduce interference and improve system capacity,this paper focuses on the precoding and power allocation issues on the base station side of massive MIMO systems.At present,there are many studies on precoding and power allocation algorithms for massive MIMO systems,and obtained a number of results,but there are still some problems.Firstly,many precoding algorithms do not fully consider the interference specificity existing in different communication system environments,so the effectiveness of precoding cannot be fully played.Secondly,the existing power allocation optimization goals usually choose one of the maximum sum rate or the maximum minimum rate,which cannot simultaneously balance the overall system and individual user performance;Finally,when the optimization objective is the maximum sum rate,it has been proved to be a non-convex problem,while when the optimization objective is to maximize the minimum rate,the convex optimization problem will be generated,and the optimal solution needs to be calculated by the convex optimization solution method,with high computational complexity.In order to solve the above problems,this paper firstly proposes a hierarchical precoding scheme.According to the correlation of the channel between users in the cell,users are grouped and implements corresponding precoding schemes according to the grouping situation,this can effectively eliminate Multi-User Interference(MUI),InterCell Interference(ICI)and residual MUI are suppresses by precoding based on maximum signal-to-leakage-and-noise ratio.Secondly,in order to give consideration to the performance of the whole system and individual users,the optimization objective function is designed to maximize-the product of the minimum rate and average rate of users in each cell.And the artificial immune algorithm is introduced to solve the power allocation problem.It optimizes the search process of power allocation by applying biological immune mechanism,which offers the benefits of fast convergence speed and high accuracy.Experimental results show that compared with several general precoding algorithms,the proposed algorithm can improve the user rate by nearly 2.3 times at the position with 5% interrupt probability,and execution power allocation can not only ensure the minimum user rate,but also improve the average user rate,so that the overall rate can be improved by about 2bit/s/Hz.Finally,in order to adapt to the dynamic changes of users in the system to perform power allocation instantly,a power allocation algorithm based on Deep Neural Network(DNN)is proposed.The user location information and the corresponding power allocation results are regarded as a mapping relationship.The pre-processing and sample label calibration were carried out to construct a complete data set,and the power allocation DNN model is built and trained.Experimental results show that compared with the traditional power allocation optimization algorithm,the proposed power allocation algorithm based on DNN can not only take into account the performance of the whole system and individual users,but online power allocation can also significantly reduce the computation time to meet the dynamic needs of users.