The Design And Optimization Of Noma Codebook

In recent years,mobile communication technology has achieved rapid development,with significant improvements in system capacity,spectrum utilization,communication delay and so on.But at the same time,the popularity of the "Internet of Things" and the explosive growth of mobile Internet business volume also pose new challenges to the mobile communication technology.On the one hand,the number of mobile device accesses is rapidly expanding,so greater system capacity is required in future mobile communication networks.On the other hand,spectrum resources are limited,so the new-generation mobile communication technologies need to use spectrum resources more efficiently.The non-orthogonal multiple access(NOMA)technology allows user signals to jointly occupy communication resources in a non-orthogonal manner,which can effectively improve spectrum utilization and system capacity.Since user information is superimposed on the same resource element and sent after being mapped to codebook,it is inevitable that there is interference between users during decoding at the receiver.We need to design and optimize a suitable codebook to help distinguish different user signals,which plays a vital role in non-orthogonal multiple-access systems.Therefore,this thesis focuses on designing and optimizing user codebooks in non-orthogonal multiple access to further improve system performance.First of all,we construct a unified model and propose a universal codebook design,incorporating multiple optimizable signal domains into a mapping matrix.Then,the system is theoretically evaluated from the aspects of system capacity and decoding error rate,which are also used as targets to optimize the codebook.Secondly,we propose to apply a new loss function and use batch gradient descent algorithm to jointly optimize the universal codebook from multiple domains.The traditional methods of optimizing codebooks are generally heuristic search algorithms such as genetic algorithms,but the optimization complexity is still too high when the system is densely connected or joint optimization is performed.The simulation results demonstrate that our proposed method can find a better codebook with higher error performance.Finally,we innovatively propose to apply the constellation rearrangement to the non-orthogonal multiple access system to make it to be a new optimizable domain.Traditional codebook optimization is mainly performed in the power and phase signal domains,therefore the optimized codebook is sensitive to the estimation accuracy of channel state information.The newly proposed scheme can solve this problem and further improve the performance of the system after being optimized by an improved simulated annealing algorithm.