Research On Intelligent Multi-domain Resource Allocation Algorithms For 6G

With the explosive growth of user service demand,how to use the limited wireless resources to carry more wireless services has become a challenging problem in the research of the next generation wireless network.Reasonable allocation of limited communication resources can effectively improve wireless communication transmission capacity.However,traditional resource allocation algorithms have great computational complexity and cannot meet the requirements of the sixth generation of mobile communications(6G).Introducing machine learning(ML)is an effective way to solve this problem.By training historical data and extracting system features to improve communication performance,ML technology can effectively alleviate the problem of high complexity.This thesis aims to explore the application of ML technology in communication resource allocation.Aiming at full-duplex spatial modulation(FD-SM)system,two transmit antenna selection(TAS)algorithms and a power allocation(PA)algorithm are proposed in this thesis.By combining them,an intelligent federated resource allocation(IFRA)algorithm is proposed.Firstly,the channel capacity of the FD-SM system and its upper and lower bounds are derived.In addition,two new TAS methods based on support vector machine(SVM)and deep neural network(DNN)are proposed for the influence of residual self-interference on FD-SM system performance.In the design of this paper,a new feature extraction method based on principal component analysis is proposed to help improve the training of the proposed classifier.Simulation results show that the proposed intelligent TAS scheme can achieve nearly optimal performance while significantly reducing complexity.This thesis also studies the intelligent PA algorithm and IFRA algorithm of the FD-SM system.Specifically,with the aid of a PA matrix codebook,this paper proposes a PA scheme based on the minimum bit error rate(BER).A PA algorithm based DNN for the FD-SM system is proposed.Furthermore,combining TAS and PA algorithms,the IFRA algorithm is proposed in this paper,which shows much better BER performance than TAS or PA algorithm.In addition,precod-aided spatial modulation(PSM),a variant of spatial modulation technology,is applied to untrusted relay(UR)system to improve its security.It's called the PSM-UR system.Firstly,the secrecy rate of the PSM-UR system is deduced.In addition,in order to improve the security of PSM-UR system,an intelligent receive antenna selection(RAS)algorithm based DNN is proposed.Simulation results show that the proposed DNN-RAS algorithm can guarantee the secure communication of the PSM-UR system with low complexity.