The Research Of Active User Detection And Channel Estimation In Uplink C-RAN

In recent year,devices connected to the communication network is growing quickly,since the rapid development of new wireless communication technologies.To ensure the connection of massive communication equipment,high data transmission efficiency and ultra-low transmission delay,China Mobile Research Institute has proposed a promising network architecture called‘cloud radio access network?C-RAN?',which is able to satisfy the requirements of explosive real-time data service.If C-RAN system adopts the traditional Grant-based access technology,it will lead to a lot of signaling overhead and reduction of the transmission efficiency due to the large-scale user connection characteristics of C-RAN system.Therefore,this article adopts the latest Grant-Free access technology in the 5G communication which does not need to apply for wireless transmission resources from the base stations?BSs?in advance.However,the BSs don't know which users are transmitting data and the channel state information?CSI?based on the Grant-Free access technology.The BSs can't recover the data transmitted by the active users without CSI.Therefore,the BSs should detect the active users and estimate channel firstly.This article studies the joint algorithm design of active user detection?AUD?and channel estimation?CE?in C-RAN uplink system based on the Grant-Free access technology.According to the existing research,active users are sparse in C-RAN system,inspiring the use of compressed sensing?CS?theory to solve the problem of AUD and CE.On the basis of the efficient and parallel alternating direction multiplier method?ADMM?,this paper proposes Double-ADMM?D-ADMM?detection algorithm in single antenna system and Block-ADMM?B-ADMM?detection algorithm in multi-input multi-output?MIMO?system,respectively.With respect to solve the issue of AUD and CE,this paper studies the single antenna system and MIMO system.More details are shown as following:1.D-ADMM algorithm design in single antenna system:The joint detection problem is a standard compressed sensing block sparse problem in single antenna system.This paper proposes an efficient joint detection algorithm called D-ADMM algorithm by decomposing the original problem into several sub-problems and introducing ADMM.Firstly,the original problem is decomposed into several sub-problems for parallel solution,which is a standard CS problem.We release the7)₀norm to the7)₁ norm to solve the constraint problem of7)₀ norm of each sub-problem.Secondly,we introduce regularization parameters to transform each sub-problem into unconstrained convex optimization problem.So it is not necessary to acquire the prior knowledge of sparsity.Finally,each sub-problem is decomposed into several simple local problems by ADMM and solved alternately and iteratively.The advantage of the D-ADMM algorithm does not need the prior knowledge of sparsity.Compared with the standard convex optimization reconstruction algorithm,the D-ADMM algorithm can reach the same signal recovery performance approximately,but the calculation time is reduced by two orders of magnitude.2.B-ADMM algorithm design in MIMO system:The signal recovery problem is defined as a standard CS block sparse problem in MIMO system.This paper proposes efficient B-ADMM algorithm by transforming the block sparse problem into a convex optimization problem and introduces ADMM.Firstly,this paper transforms the original problem into a convex problem by releasing the non-convex constraint problem of7)_2,0norm to the convex constrained optimization problem of7)_2,1 norm.Secondly,the regularization parameter is introduced to transform the convex problem into an unconstrained problem.Finally,the paper introduces ADMM to recover the original single.The advantage of B-ADMM algorithm does not need the prior knowledge of sparsity.Compared with the standard convex solver,the B-ADMM algorithm can reach the same signal recovery performance,and the calculation time is reduced by an order of magnitude which does not increase with the increase of pilot length linearly.