Research On Blind Detection Algorithm Based On Factor Graph In Grant-free Multiple Access

With the rapid development of mobile communication technology,modern communication has achieved significant improvements in system capacity,transmission delay,and spectrum resource utilization.However,emerging services such as the Internet of Things pose higher challenges for the next generation of wireless communication systems,namely ensuring high reliability and low latency while accommodating massive communication devices.Due to the orthogonality constraints of existing Orthogonal Multiple Access(OMA)technologies,it is difficult to meet these demands.Therefore,Unscheduled Non-Orthogonal Multiple Access(NOMA)technology emerged.This method allows multiple users to send signals on the same time-frequency resources without the need to preallocate specific resources.It has become the focus of research in academia and industry.This paper focuses on the design of multi-user blind detection algorithms for the receiving end in grant-free uplink NOMA scenarios,aiming to accomplish the identification of active users,channel estimation,and user data detection.Based on the factor graph model as the theoretical foundation,and in combination with various advanced algorithms and detection techniques,the paper designs and improves the performance of different detection modules of the receiver.Firstly,based on compressive sensing theory,this paper proposes an enhanced approximate message passing algorithm for joint active user detection and channel estimation,which utilizes a smooth soft thresholding operator and convergence improvement.The proposed algorithm enhances the capability of active user identification and channel recovery.Secondly,based on the factor graph message passing scheme,the paper proposes a Bayesian blind detection algorithm with an improved convergence mechanism,accelerating the convergence of the iterative algorithm and effectively improving the system’s multi-user detection performance.Lastly,for the parallel message passing Bayesian blind detection algorithm,the paper introduces Serial Interference Cancellation(SIC)to adjust its scheduling strategy and detection process,improving the error performance while reducing the computational complexity of the iterative system.Simulation results show that the approximate message passing algorithm proposed in this paper,based on a smooth soft thresholding operator and convergence improvement,enhances the capability of active user identification compared to the conventional approximate message passing algorithm;the convergence improvement mechanism based on the factor graph message passing algorithm proposed in this paper has a significant performance gain compared to a single message passing algorithm;meanwhile,the introduction of SIC achieves further improvements in error performance while effectively reducing the detection complexity.