Sparse Signal Detection And Optimization In MIMO Wireless Communication System

Signals in nature are always dependent on a few parameters which can be regarded as the characteristic of sparsity. The sparsity can be exploited and utilized by appropriate measurement and representation. By utilizing the sparsity compressed sensing can acquire signals effectively by random sampling to reduce the sampling rate remarkably. In wireless communications there are lots of sparse phenomena. For example, in some wireless channel, there exits only a few main signal paths, which means that the channels have sparsity; in codebook based channel feedback scheme, only the index of the codeword is fed back, which means that there is sparsity in the feedback procedure; in user selection in downlink channel, the base station transmits to a small part of users simultaneously, which means active users have sparsity; in the signal detection in uplink channel, only a small part of users access to the base station simultaneously which means the transmitted signals can also have sparsity. In wirless communication, the communication efficiency and system performance can be improved by utilizing the sparsity to optimize the wireless communication technologies. Hence, it is vital to study the wireless communication technologies based on sparsity for future wireless system design.In the thesis, we will make use of the sparsity and sparse signal in channel estimation, channel information feedback, user scheduling in the downlink channel and user detection in uplink channel in the wireless communications, and apply compressed sensing into the optimization, detection and transmission of sparse signal. Effective wireless communication technologies based on compressed sensing are proposed to achieve better performance. The main contributions of the thesis are as follows:1. In MIMO channel estimation, the non-selective MIMO spatial channel model and its sparsity are analyzed, and a basis optimization algorithm is proposed to reduce the leakage effect and enhance the sparsity of spatial MIMO channel. We apply the basis optimization algorithm to channel estimation by compressed sensing in spatial MIMO channel, and verify the effectiveness and good performance with the proposed optimized basis.2. In MIMO channel feedback the channel representation based on sparse approximation is proposed, and the sparse representation vector is spited into several l-sparse vectors and compressed by compressed sensing independently. The compressed vectors are fed back to the base station and then recovered. We compare the digital feedback, analog feedback, hybrid feedback schemes and also the random vector quantization feedback. The simulations show that the sparse approximation based on Grassmanian codebook has better performance. In low SNR regime, the system rate with digital feedback outperforms it with analog feedback; in high SNR regime, analog feedback has better performance. The proposed algorithm can improve the system throughput, and the feedback resource consumption is related to the number of selected atoms for approximation in the codebook, and irrelated to the antenna number which is practical in the massive MIMO with large antenna number.3. By exploiting the sparsity in the active users in user scheduling in MIMO downlink channel, an efficient user selection and feedback algorithm based on compressed sensing is proposed. By user self-selection algorithm, only part of users feed back their channel information to the base station which can reduce the consumption of uplink feedback resources. Because compressed sensing can detect the active users, the signaling is reduced. Simulations show that analog feedback based on the compressed sensing outperforms the digital feedback with the same feedback users, but consumes less uplink feedback resources; the system throughput for digital feedback based on compressed sensing is very close to the digital feedback with all users, but occupies less feedback resources.4. For multi-user detection in MIMO uplink channel, we apply the compressed sensing to the sparse signal detection in MIMO communications, and directed dimension spread and diversity dimension spread algorithms are proposed. Because of making use of the sparsity, less short spreading code can be efficient for the user detection compared with the conventional CDMA system. In the diversity dimension spread algorithm, the length of the spreading code can be shortened further because the system dimension, such as multiple antennas, is used. In the sparse signal detection we propose a2-stage MMSE detection algorithm. Simulations show that when the active users are less or spreading code is long, the proposed algorithm outperforms MMSE, OMP, SP and CoSaMP.