Efficient Receiver Design For Grant-free Wireless Communication Systems

The increasing demand of the Internet of Things poses challenging requirements for future wireless communications,such as massive connectivity and high spectral efficiency.In the grant-free access systems,user terminals can spontaneously transmit data to the base station without initiating a scheduling request,thereby reducing signaling overhead,transmission delay,and power consumption.Due to those advantages,grant-free access has become one of the key technologies in massive Machine Type Communications(m MTC)scenarios.Since the base station does not have the prior active state information of potential users,active user identification is necessary for the base station in the grant-free access systems.We focus on the efficient receiver design which contains active user identification,channel estimation,and signal detection in the uplink grant-free access systems.Firstly,we investigate the active user identification and channel estimation in the uplink Multiple Input Multiple Output(MIMO)grant-free system.In m MTC scenario,only a small fraction of potential devices are active.According to the sparse characteristics of active devices,we can build the active user identification and channel estimation as a Multiple Measurement Vector(MMV)problem.Block Orthogonal Matching Pursuit(BOMP)algorithm extended from Orthogonal Matching Pursuit(OMP)is proposed to solve this problem.On the other hand,we establish the constrained Bethe free energy minimization problem from the variational Bayesian inference perspective,where the constraint set is designed according to the characteristics of random variables.We then derive Hybrid Message Passing(HMP)using the method of Lagrange multipliers and investigate HMP based active user identification and channel estimation in rich scattering channel and spatial correlated channel model.The simulation compares the performance of BOMP and HMP based active user identification and channel estimation algorithms in terms of channel estimation mean square error,active user miss detection rate,and active user detection error rate,respectively.Simulation results show that the channel estimation performance of HMP is better than which of BOMP and nearly optimal.Secondly,to fully explore the uncertainty of channel estimation,we study the joint active user identification,channel estimation,and signal detection method in the uplink grant-free massive MIMO communication systems.Non-Orthogonal Multiple Access(NOMA)has been proposed for the grant-free transmission systems due to higher spectrum efficiency,where multiple users occupy the same time-frequency resource block and the base station distinguishes users by user-specific codebooks.We can model the joint channel estimation and signal detection as the bilinear estimation problem,where matrices to be estimated show the common column sparsity.On the one hand,we investigate the receiver design in the grant-free Sparse Code Multiple Access(SCMA)system.We construct the factor graph based on the joint probability distribution of all random variables and apply message passing rules to different parts of the factor graph.The joint identification,estimation,and detection algorithm is then derived based on message passing.On the other hand,we develop the joint identification,estimation,and detection algorithm for unified NOMA structure with the variational Bayesian method.By solving the Bethe free energy minimization problem,we derive the low-complexity joint identification,estimation,and detection algorithm named Bi-VMP-GAMP.Simulation results show that joint active user identification,channel estimation,and signal detection receiver performs significantly better than the two-stage algorithm in terms of channel estimation mean square error and bit error rate.Since timing offsets are inevitable due to channel environment differences between users,we investigate the identification and time delay estimation in the uplink grant-free access Orthogonal Frequency Division Multiplexing(OFDM)system.Establishing a general received signal model considering both the comb-type pilot and block type pilot,we formulate the identification and estimation as the block sparse vector estimation problem.We introduce discretized auxiliary variables to describe the channel's block sparsity characteristics and propose the active user identification as well as channel and delay estimation algorithm based on Belief Propagation(BP)and Generalized Approximate Message Passing(GAMP).Besides,The autocorrelation algorithm and the Modified Orthogonal Matching Pursuit(M-OMP)algorithm are derived.Simulation results show that the proposed methods can identificate users' active states and estimate channel accurately.