Reseach On Compressed Sensing Based Non-Orthogonal Multiple Access

Grant-free non-orthogonal multiple access(NOMA)has recently gained significant attention for reducing signaling overhead in machine-type communications(MTC).In this context,compressed sensing(CS)has been identified as a good candidate for joint activity and data detection due to the inherent sparsity nature of user activity.This paper augments activity and data detection for frame based multi-user uplink scenarios where users are(in)active for the duration of a frame,namely frame-wise joint sparsity model.However,signals are recovered on the basis of a time slot in most of state-of-the-art algorithms,so that the frame-wise joint sparsity isn't not fully exploited.Under the circumstances,a threshold aided block sparsity adaptive subspace pursuit(TA-BSASP)is proposed.The frame-wise joint sparsity is firstly transferred into block sparsity and then block sparsity recovery algorithm is adopted to make sure that signals are reconstructed in a whole frame each time.The proposed TA-BSASP enjoys a performance gain compared to state-of-the-art algorithms.Besides that,stardard restricted isometry property(RIP)concept is normlised to block RIP concept and based on the latter,theoretical analysis in convergence of block sparsity recovery algorithm is given in the view of that such analysis is seldomly done in the current algorithms.It should be noted that the activity of users is normally unknow in multi-user detections.It means sparity is not a priori condition in CS.Considering that either sparsity or noise power should be known in current research,cross validation aided block sparsity adaptive subspace pursuit(CVA-BSASP)is developed.Cross validation which is stemed from statistics or machine learning is featured as the method of overfitting judgement in CVA-BSASP.No other priori conditions is needed in the algorithm and thus CVA-BSASP is highly practical.Furthermore,the computational complexity of the proposed algorithms are derived.Superior performance of the proposed algorithms is demonstrated by numerical experiments.