Researches On Compressed-sensing-based Active User Detection And Channel Estimation

With the development of the fifth generation(5G)communication system in recent years,large-scale distributed-antenna(LSDA)systems such as cloud radio access network and fog radio access network,and massive machine-type communication(mMTC)systems work as emerging network architectures and technologies,and have been advocated by both researchers and enterprises.The active user detection(AUD)and channel estimation(CE)problems are one of the basic problems in wireless communication researches.In classical AUD and CE methods,orthogonality between identification pilots belonging to different users is required,and thus the pilot length needs to scale with the total user within the system.Because the number of user devices in LSDA and mMTC is much larger than the traditional cellular network,solving AUD and CE problems by classical methods in systems like LSDA and mMTC is very expensive or even impossible.In order to reduce the required pilot length,this dissertation studies the AUD and CE problems in wireless communication systems by leveraging compressed sensing(CS)technologies,and proposes a novel cross-layer protocol based on CS for random access in mMTC.Firstly,this dissertation studies the AUD and CE problems in one single cell using CS.The basic CS theories and the model of the AUD and CE problem in one cell are introduced.Because the processing time of AUD and CE should be strictly less than the channel coherence time,this dissertation proposes low-complexity methods to accelerate the solving procedure.Based on the assumption that active users are sparse,the target problem can be solved effectively using pilots much shorter than the total user number.Secondly,the AUD and CE problems in LSDA systems equipped with wired fronthaul links are studied.Based on CS technologies,this dissertation proposes a novel penalty function to exploit the sparsity of user activity and spatial sparsity of user signals existing in LSDA systems.Different from existing works,the proposed method does not need any prior information of channels,but can still provide a good performance.It is worth mentioning that the proposed penalty function can even work without the sparsity of user activity.Besides,three low-complexity methods that can be deployed in both centralized and distributed manners are proposed to solve the penalty function.These methods have different performances and computational complexities,and thus can provide differentiated services.Thirdly,this dissertation studies the AUD and CE problems in LSDA systems which are equipped with wireless fronthaul links.A two-stage CS process is proposed to exploit the sparsity of user activity and spatial sparsity of user signals,respectively.A corresponding two-stage low-complexity method and its distributed deployment are proposed to accelerate the solving procedure.Based on the sparsity of user activity,the proposed method can detect active users and estimate the channels from active users to remote radio heads effectively.At last,this dissertation proposes a cross-layer random access protocol in the mMTC scenario,which is an important application of the AUD and CE methods described above.Different from traditional protocols,AUD and CE in the proposed protocol are implemented by CS in isolation,which makes it possible that multiple collided packets can be recovered in one single time slot.Hence,the system capacity is enlarged greatly.Because interaction for synchronization could be unnecessary between BS and inactive UEs in the proposed protocol,the energy efficiency of UEs can be enhanced greatly.This dissertation provides theoretical and empirically results on data recovery,choice of system parameters,transmission delay and dynamics of active users in the proposed protocol.