Research On NOMA Multi-user Detection Technology Based On Perturbed Compressed Sensing

As one of the important application scenarios of the 5th Generation Mobile Communication Technology(mMTC),the scenario with large scale machine communication needs to have large scale access,low time delay and other technical requirements.Non-orthogonal Multiple Access(NOMA),as an important solution of mMTC makes more users can use wireless communication service simultaneously.In NOMA system,the grant-free technology serves the more users,reduces the delay of signaling transmission,and reduces the transfer application process of the users.However,the grant-free mechanism also causes one problem,where the base station needs to perform active user detection.In uplink grant-free NOMA system,as the number of active users is far less than the number of possible users,so the multi-user detection problem can be modeled as the sparse signal reconstruction problem of compressed sensing.The channel estimation error is often ignored in the design of existing sparse multi-user detection schemes,which results in performance loss in practical applications.In order to solve this problem,this paper studies sparse multi-user detection schemes with channel estimation error considered.The main work and contributions of this paper are as follows:Firstly,this paper studies the basic concepts and principles of non-orthogonal multiple access technology and the transmission model of grant-free NOMA.This paper studies the basic principle and the reconstruction algorithms of compressed sensing theory,and further studies the deformation model of compressed sensing theory includes perturbation compressed sensing and one-bit compressed sensing.Then,in this paper,a sparse multi-user detection model for grant-free NOMA system based on block sparsity is proposed.In this paper,the transmission model of grant-free NOMA system is transformed into compressed sensing reconstruction problem,and perturbed multi-user detection with channel estimation error is modeled as the perturbation in the measurement matrix.Based on this model,an iterative multi-user detection algorithm is designed to update the channel estimation at the same time as the user detection,thus reduce the performance loss caused by the channel estimation error.Experimental simulations verify that the proposed scheme has better detection accuracy than the existing scheme when considering channel estimation error.Lastly,this paper proposes a multi-user detection scheme based on one-bit compressed sensing.In order to reduce the complexity at the receiver,the single bit quantization of received signal is considered.The channel estimation error is also modeled as the perturbation of the compressed sensing measurement matrix.This paper extends the above iterative multi-user detection algorithm to the one-bit quantified model,and proposes a new iterative detection mechanism based on the gradient descent algorithm.Experimental simulations verify that the scheme proposed in this paper can effectively depress the error of channel estimation and improve the robustness of multi-user detection.The two multi-user detection schemes proposed in this paper make full use of the inherent sparse characteristics of received signals and design robust detection algorithms,which have practical application value.