On The Multiuser Detection And Codebook Design Of Sparse Code Multiple Access

With the increasing demand for wireless communication,3GPP has defined some key targets for the future 5G system,such as the large network capacity,high user density,ultra low latency transmission,etc.In nonorthogonal multiple access systems,different users' signals are nonorthgonally superposed and sent to the receiver.The receiver needs to decode the superposed signals to separate them,which is also called multiuser detection.Generally,the computation complexity in receiver side is largely increased but it increases the spectral efficiency.Since the nonorthogonal multiple access can better satisfy the targets for 5G,it has been listed as the key technique in 5G by the IMT-2020(5G)Promotion Group.Sparse code multiple access(SCMA)is a kind of nonorthogonal multiple access technique,which is proposed by Huawei Technologies Co.,Ltd.It combines the QAM mapper and spreading together,and uses codebooks to realize the two steps.It actually combines several constellations into a joint constellation with high dimension.The codewords are equivalent to the high dimensional points in the joint constellation.Because the combined constellations will help each other to detect the transmitted signal in the receiver side,SCMA can improve the spectral efficiency significantly.It has become one of the most promising nonorthogonal multiple access techniques.However,as a cutting-edge technique,SCMA definitely still needs to be improved.Until now,an important bottleneck is that although SCMA improves the network capacity,the detection in the receiver side has some problems.Because the signals from different users are no longer orthogonal,the receiver needs to use detection algorithms with high computational complexity to separate the signals.Actually,the complexity of most detection algorithms for SCMA will grow exponentially with respect to the number of users in the system,which brings inconvenience to the extend of communication systems.Besides,every user in SCMA systems has a codebook,in which every codeword is a high dimensional vector.It is still unsolved that how to design the codewords for each codebook for high detection accuracy and large network capacity.Most existing codebook design algorithms subjectively prefer heuristic principles to design the codebooks,and some utilize the channel capacity.But because there is neither analytical solution nor convenient approximation for the channel capacity of SCMA systems,some previous algorithms design codebooks by considering upper bounds for the capacity or giving the system a strong assumption.In this thesis,the author presents some researches for the SCMA multiuser detection and codebook design as follows.1.The second chapter proposes a novel low complexity multiuser detection algorithm,which makes the detector has computation complexity of polynomial order.Although the proposed algorithm leads to some accuracy loss in detection,it can be proved that the loss will be negligible if the interval in discretization is small.It is worth noting that when the real part in a SCMA system is independent of the image part,the two parts can be detected individually,which will further reduce the computation complexity.In particular,when the two parts are independent of each other,the proposed algorithm only needs near-linear complexity to update a message.2.The third chapter presents a new codebook design method for the previous detection algorithm,which can avoid the accuracy loss in many channel scenarios.Given the sampling points in discretization,the codebooks can be constructed by using greedy strategy and the constraints in factor graph.The simulation shows that in some common channels the proposed method will give us better codebooks than some classical design methods.It is worth noting that because the elements of the codebooks are all chosen from the sampling points,in some situations the codebooks can avoid the accuracy loss in the proposed detection algorithm.3.The fourth chapter shows some researches on the capacity and mutual information of SCMA systems.In the first section,the author proposes a series of analytical functions to approximate the mutual information of SCMA.The proposed analytical approximation has a simple form,and only needs a low computational complexity.Therefore we can analyze how to get a larger mutual information by adjusting the independent variables,i.e.,the codebooks.In order to prove its validity,the proposed design method is compared with a classical method.Numerical result shows that in the same scenarios the new method always can find codebooks with larger mutual information.In the second section,the author investigates the asymptotic property of the capacity.By constructing a kind of discrete random variables,this section proves that in SCMA systems,when the size of the support set increases,the mutual information can approach the channel capacity.In the communication systems,the multiple access is always the key part of the air interface.In the future 5G system,the nonorthogonal multiple access will be applied.In the thesis,the author analyzes the possible research topics and trends in SCMA system,and hope it can help someone who are interesting in SCMA.