Research On Iterative Detection In MIMO Wireless Communication Systems

Multiple-input multiple-output (MIMO) system employs multiple antennas at thetransmitter and the receiver, which significantly improves frequency spectrum andefficiency and becomes key technique for the future high-speed broadband wirelesscommunications. It is a major challenge to design practical signaling and signalprocessing schemes capable of supporting data rates close to MIMO capacity limit.An efficient approach is the use of bit interleaved code modulation (BICM) at thetransmitter in conjunction with the iterative receiver. The thesis focuses on theapplication of the iterative technique in the MIMO and multiple-input multiple-output-orthogonal frequency division multiplexing (MIMO-OFDM) systems. In this thesis,we will study the design reduced-complexity and improved-performance soft-in soft-out (SISO) detector, the convergence features of iterative receiver and the iterativechannel estimation in the MIMO-OFDM system..First, we study the iterative receiver in the MIMO systems. Since the optimalSISO detector has exponentially complexity, many research interests have been paidto reduce the complexity of optimal SISO detector. We present the popular SISOdetectors, analyzing their performance and comparing their complexity. Then wepropose an extrinsic information update-based SISO detector. Compared withconventional SISO detector, which only uses the extrinsic information of channeldecoder, the novel SISO detector uses the posteriori information of channel decoderand extrinsic information of SISO detector of present iteration. It speeds up theconvergence and reduces the complexity.Second, we give some discussion on the iterative receiver in MIMO-OFDMsystems. We derive the SISO detector based on linear minimum variance un-biased(LMVU). We propose a channel-clustered-based SISO detector, which clusters thechannel into groups according to the correlation coefficient of time and frequency. Ata cluster, we replace the instantaneous covariance matrix of transmitted symbol vectorwith its average in frequency domain. Then same LMVU weight is used in the cluster.The detector greatly reduces the complexity with little loss in performance.In the case of soft interference cancellation–linear minimum square error (SIC-LMMSE)-based iterative receiver, since the weight filter has to be updated on aniteration fashion with matrix inversion, the computational complexity becomescomparatively high. Hence the soft interference cancellation (SIC)-matched filter (MF)-based receiver is proposed to reduce the complexity at a cost of performancedegradation. Since it ignores the existence of residual interference at each iteration, theSIC-MF-based soft detector performs worse when a large number of transmit antennasis used. To mitigate this problem, we scale the a priori information fed to the SISOdetector. The corresponding algorithm is referred to as the soft interferencecancellation-scaled matched filter (SIC-SMF)-based soft-detector. We also optimizethe scaling factor by using extrinsic information transfer (EXIT) chart. The SIC-SMF-based soft detector significantly improves the system performance with nearly thesame complexity as SIC-MF-based soft detector.In the OFDM systems, the length of cyclic prefix (CP) is longer than themaximum delay of channel. We add the received data that is not disturbed by inter-symbol interference (ISI) with received data of OFDM symbol and then get anequivalent signal model. By using this new model, we design the iterative receiver.The convergence speed is faster and performance is improved.Third, we study the convergence behavior of iterative receiver. In the imperfectchannel model, we derive the effective conditional Gaussian channel model and giveerror free feedback (EFF) bound of bit interleaved code modulation wit iterativedecoding (BICM-ID). In the MIMO system, we derive the EFF bound under receivecorrelated channel. We also analyze the exchange of extrinsic information between thedetector and decoder by using the EXIT chart and obtain the meaning result.Finally, we study the iterative channel estimation in the MIMO-OFDM system.We improve the channel estimation accuracy by using the posteriori information afterchannel decoder. We propose a soft interference cancellation–least square (SIC-LS)based iterative channel estimation scheme. We estimate time domain channel by LScriterion by using the symbol after SIC. Since soft symbol matrix is not unity, LSalgorithm needs matrix inversion. Hence complexity is still high. We propose areduced-complexity SIC-LS iterative channel estimation scheme. Simulation resultsshow that its performance is close to that of SIC-LS-based iterative channel estimation.