Research On Antenna Selection Security Algorithm For Massive MIMO Systems

With the development of fifth-generation mobile communication technology(5G),terminal users are increasingly demanding network stability,signal quality,and transmission rate.This increase in demand is due to a number of factors,such as the massive number of device connections,the rapid growth of data volume,the emergence of new communication services,and more complex application scenarios.Massive MIMO(Multi-Input Multi-Output)technology has become one of the key technologies of the next-generation communication network by equipping a large number of antennas to make full use of space resources and provide more multiplexing gain without consuming additional spectrum resources,thus improving the system capacity.However,traditional massive MIMO systems require a dedicated RF(Radio Frequency)chain for each antenna,which results in high hardware costs and energy consumption for actual deployment.Antenna selection can greatly reduce the RF link overhead and the system cost by selecting part of the antenna for signal transmission.As the application of ML(Machine Learning)becomes more and more widespread,the application of machine learning to antenna selection is of great significance to improve system performance and reduce computational complexity.In this thesis,we study the antenna selection algorithm for the Massive MIMO system,and the main work is described as follows:Firstly,this thesis introduces the eavesdropping channel model of the MIMO system and deduces the channel security capacity and security interrupt probability as the performance evaluation indicators of the system.Five classical antenna selection algorithms are introduced and their realized channel security capacity and computational complexity are compared by simulation.Then the workflow of ML and the structure of deep learning especially CNN(Convolutional Neural Network,CNN)are introduced.Secondly,this thesis proposes a transmit antenna selection algorithm based on GBDT(Gradient Boosting Decision Tree).The optimization problem of transmit antenna selection is transformed into a multi-class classification problem,and the GBDT classifier is constructed to solve the problem.With the help of the ability of ML to solve the multi-class classification problem,the channel matrix features are extracted and normalized.Based on the derived channel security capacity,labels are defined for the channel matrix samples,and a complete data set is generated for training the GBDT classifier.Simulation results show that the proposed algorithm is close to the traditional exhaustive antenna selection algorithm in terms of channel security capacity while ensuring high accuracy of antenna selection.Thirdly,in order to ensure the continuity of communication between the transmitter and receiver,this thesis proposes a joint transceiver antenna selection algorithm based on CNN.By taking advantage of the multi-channel network structure of CNN,which can process multiple features at the same time.Combined with the target of extracting features at both sides,this thesis constructs a twochannel CNN.Two kinds of convolutional kernels are designed for the two-channel to extract the features of the transmitter and the receiver respectively,and labels are re-defined for the channel matrix samples by using the multi-label method.The experimental results show that the channel security capacity of the proposed algorithm is close to that of the traditional exhaustive search algorithm and better than that of the GBDT-based antenna selection algorithm.In addition,the loss function and classification accuracy of the two-channel CNN are simulated.The classification accuracy of the proposed algorithm is up to 97.6%.Finally,considering the multi-antenna and multi-user system model,a joint users scheduling and transmit antennas selection scheme is proposed.The algorithm considers system sum rate maximization as the optimization objective and finally selects both the transmit antenna and user subsets.In addition,in order to allow more users to have the opportunity to participate in communication,an improved algorithm based on ratio fairness is proposed,and the communication rate is sacrificed as little as possible under the premise of ensuring that users participate in communication fairly.The simulation results show that the proposed algorithm can obtain good system performance under different numbers of transmit antennas and different numbers of users.The algorithm combined with fair scheduling is almost indistinguishable from the unfair ratio algorithm in terms of achievable sum rate,and it is of practical significance to successfully allow more users to participate in the communication.