Research On Grant-free Access In NOMA System

Traditional networks use orthogonal multiple access and only allocate radio resources to a single user device.With the development of science and technology,the arrival of 5G era,the explosive growth of mobile Internet and Internet of things business,in order to support large-scale connection,and improve the overall spectrum efficiency and throughput,several non-orthogonal multiple access(NOMA)schemes have been developed.In NOMA,multiple user devices share a radio resource block.Grant-free NOMA combines the functions of noma with the simplicity of the classical random-access principle.UE can send its data randomly on the shared resource block without any complicated handshake and expand its information on the uniquely allocated expansion sequence,thus improving the throughput and spectrum efficiency.In the Grant-free NOMA system,BS must run the advanced multi-user detection(MUD)algorithm,which not only finds the set of all active ues sending information bits,but also estimates their channels and decodes their data.Because the channel information at BS is often out of date,UE transmission is sporadic and intermittent,so active UE detection,channel estimation and data detection need to be carried out at the same time.In order to develop high performance MUD method,people have made great efforts.Most of the existing algorithms need prior information such as the number of active user devices,which is not practical in practice.Moreover,the collision problem caused by the same pilot sequence selected by different devices should also be focused on.The main work of this paper is as follows:First of all,we apply the sin spread sequence to the grant-free NOMA system.This turns the user detection problem into a frequency estimation problem,allowing us to use non-iterative,low complexity and accurate algorithms.The proposed method does not need the number of active user devices.Simulation results show that in most practical cases,the performance of the proposed method is better than the existing algorithms,and the computational cost is almost comparable.Secondly,we consider a framework of uplink grant-free NOMA,where simultaneous interpreting random pilot sequences and their data are transmitted by devices.In this case,the probability of two or more devices selecting the same pilot sequence is always non-zero.The receiver can only estimate their total power;however,it cannot distinguish the two devices in case of collision.In other literatures,the time slot with collision is abandoned directly.This paper suggests that these colliding devices should be regarded as interference signals at nodes.In this way,as long as some constraints are met,the device without collision can still decode successfully.For this framework,we prove that the total interference power can be well approximated by Poisson point process.We use this approximation and some properties of ordered statistics as a basis to derive simplified analytical performance bounds.