Research On Massive Unsourced Random Access Technology Based On LDPC Codes

As the demand for user and device communication is rapidly climbing,mobile communication technology is advancing and evolving.The emergence,deployment and application of massive amounts of smart devices have placed higher demands on mobile communication systems.Completing random access is a prerequisite for devices to send messages in the network.However,traditional mobile communication systems use grant-based random access schemes,which are no longer applicable to massive machine-type access scenarios due to the limited resource pool of preamble sequences and the collision of a large number of users.Therefore,in this thesis,a massive unsourced random access scheme based on LDPC codes is designed for the massive unsourced access scenario with strong device communication episodic,low activity,short data packets,and high latency tolerance.First,this thesis introduces the environment and significance of the massive random access technology.Then,according to the different random access technologies,the communication scenarios are divided into traditional mobile communication access scenarios and massive random access scenarios.Subsequently,the research status of massive random access technology at home and abroad is investigated.The advantages and disadvantages of each scheme are listed for different access scenarios,and the reasons for the scheme design based on random access without user identification are given in this thesis.Then,this thesis introduces the LDPC code compilation technology.First,the basic principle of LDPC codes and Tanner diagram are introduced,and then the quasi-cyclic LDPC code structure,coding algorithm and rate matching technique are introduced.Finally,three types of LDPC decoding algorithms,namely,logarithmic belief propagation decoding algorithm,minsum decoding algorithm and layered decoding algorithm,are described and their performance is compared and analyzed by simulation.Then,this thesis investigates the user detection technique for massive unsourced random access.First,the mathematical model and reconstruction algorithm of compressed sensing are briefly introduced,and then the basic model of massive unsourced random access user detection is described.Then,this thesis gives a compressed sensing model for embedded information bit random access channels based on On-off random access.Finally,the thesis uses simulations to analyze and verify the effects of the degree of compression of the transmit vector,active user sparsity,channel noise and estimation margin on the model performance.Subsequently,this thesis investigates the joint decoding technique for massive unsourced random access.First,the sparse interleaving multiple access technique transmitting and receiving scheme is briefly introduced,and a multi-user joint confidence propagation decoding model is given.Then,the functional nodes are introduced,the Tanner graph structure of the joint decoder is described,and the joint decoder algorithm under Gaussian white noise channel is given by combining with the log domain confidence propagation algorithm.Then,the bit collision analysis method is induced by the statistical bit interleaving graph.Finally,the performance of the joint decoder with different number of users and different transmission sequence lengths is studied and analyzed by simulation.Finally,this thesis addresses the technical shortcomings of user detection techniques and joint decoders in massive unsourced random access scenarios,and investigates a joint decoder based on multi-packet reception of LDPC codes.The information sent by the active user is divided into two parts: the preamble information and the load information.Among them,the front information is sent and received by the embedded information bit random access channel compressed sensing model,and the load information is sent and received by the multi-packet LDPC code joint codec.Finally,through simulations with different degree distribution functions and different code rates,the thesis demonstrates that in a massive unsourced random access scenario,when the number of active users is less than 100,the scheme proposed in this thesis can accomplish information transmission with less than 0.05 per-user error probability around3 dB bit signal-to-noise ratio.At the same time,this thesis is able to complete massive unsourced random access with a minimum signal-to-noise ratio of 1～1.75 dB lower than that of the conventional sparse interleaved multiple access scheme.