This thesis deals with channel estimation and equalization in frequency-selective Multiple Input-Multiple Output (MIMO) channels with 1-bit quantization at the receiver. Most current literature on quantized MIMO focuses on flat-fading channels, which motivates the topic for this thesis. The study intends to be as generic as possible, and therefore assumes no special structure of the channel. Furthermore, no joint processing or Channel State Information (CSI) is assumed at the transmitter, making the results valid also for multiple-access MIMO channels.
Three different communication schemes are explored: Orthogonal Frequency Division Multiplexing (OFDM), single-carrier with cyclic prefix (SC), and single carrier without cyclic prefix (NCP). Equivalent problem models are derived for the all the settings. A Cramér-Rao Lower Bound (CRLB) is derived for the estimation problem, as well as for the joint estimation of channel and data. The existing nonlinear algorithms Expectation Maximization (EM) and Generalized Approximate Message Passing (GAMP), as well as a linear estimator based on the Bussgang theorem, are adapted to the problems of estimation an detection. An optimal A-Posteriori Probability (APP) Soft Input-Soft Output (SISO) equalizer is designed, which has high complexity but is good as a performance benchmark. A turbo approach for Joint Channel and Data Estimation (JCD) is also explored.
Through simulations, all the methods and settings are compared. The advantage of SC over OFDM is shown. NCP turns out to be the best option at low SNR, but saturates early at higher SNR. The nonlinear algorithms exhibit better performance than the linear ones at the cost of higher computational complexity. An analysis of the required number of pilots for channel estimation is carried out. Higher order modulations are also explored, and PSK is compared to QAM for modulation orders of 8 and 16. The JCD scheme is shown to considerably improve the results for both estimation and detection.
«
This thesis deals with channel estimation and equalization in frequency-selective Multiple Input-Multiple Output (MIMO) channels with 1-bit quantization at the receiver. Most current literature on quantized MIMO focuses on flat-fading channels, which motivates the topic for this thesis. The study intends to be as generic as possible, and therefore assumes no special structure of the channel. Furthermore, no joint processing or Channel State Information (CSI) is assumed at the transmitter, making...
»