Algorithms And Architectures For Signal Detector And Channel Decoder In Wireless MIMO Receiver

With the fast development of communications technology,multiple-input multiple-output(MIMO)technology is employed as a key component in physical layer by emerging wire-less communications standards,such as LTE/LTE-Advanced(Long Term Evolution)and IEEE802.11n/ac.Equipped with multiple antennas at both transmitter and receiver,MIMO system is able to exploit spatial degree of freedom to increase the capacity without additional bandwidth and transmit power.The task of MIMO receiver is to recover the original data information cor-rectly.In order to achieve the optimal performance,the receiver has to adopt complex signal detection and channel decoding algorithms,which are big challenges for hardware architecture design.In this dissertation,signal detector and channel decoder of MIMO receiver are studied deeply at both algorithm level and architecture level.The contents are summarized as follows.First,multiple detection algorithms are required at the receiver to adapt to changing chan-nel conditions.This dissertation considers a flexible dual-mode detector,which can supports open-loop and closed-loop MIMO modes simultaneously.At the algorithm level,the proposed detector integrates K-Best detection and linear MMSE(Minimum Mean-Squared Error)detec-tion together.The MMSE-SQRD unit is reused by linear detection in the detector and thus the overall circuits overhead is largely saved.At the architecture level,the presented optimized CORDIC(Coordinate Rotation Digital Computer)systolic arrays ensure the pipelined operation of channel preprocessing and detection,which achieves high detection throughput.The imple-mentation results show that the proposed design can save 15.3%and 18.5%circuits overhead compared to LU decomposition and matrix inversion method.It also outperforms other detec-tors in terms of dual-mode support capability,antennas and modulation configurability and area consumption.Second,although breadth-first K-best detection is suitable for hardware implementation,its has high complexity when high order modulation and soft-output are required.Metric-first tree-search detection is a promising solution for MIMO receiver,because it can achieve opti-mal detection performance with optimal-efficient property.However,metric-first tree-search detection conducts tree traversal through visiting nodes one by one,leading to low processing throughput.In order to solve this issue,cross-layer parallel tree-search scheme and its hardware architecture are proposed.Equipped with a distributed stack at each level,nodes expansion can be processed in parallel.As a result,the detection throughput is improved with low circuits overhead.In addition,the proposed detector is shown to be optimal in terms of visiting all the valid nodes.The implementation results show that the presented design has higher hardware efficiency than other metric-first detectors.Third,turbo channel decoder in MIMO receiver is discussed.Based on the recursion prop-erty of QPP(Quadratic Permutation Polynomial)interleaver,a configurable interleaving net-work is proposed,which supports any 2~i(0<i?n)parallelism.Furthermore,a corresponding low complexity address generator is provided and this interleaver can be extended to support any contention-free interleaving operation.With this interleaving network,a corresponding parallel turbo decoder architecture is designed.The implementation results show that the proposed de-sign outperforms other designs in terms of multiple parallelism support,architecture efficiency and power efficiency.Finally,an improved iterative receiver is proposed by extending the soft-input soft-output tree-search detection algorithm to suppress co-channel interference.With the feedback soft information from channel decoder,this iterative receiver reduces the channel estimation and spatial covariance matrix estimation error.Moreover,the scaling operation for the a prior in-formation is proved to be effective in compensating the performance loss caused by Max-Log approximation.Simulation results show that the proposed iterative MIMO receiver can make full use of the a prior information to update the estimation,and improve the detection perfor-mance in the presence of co-channel interference.The research on signal detector and turbo decoder in this dissertation provides new insights and theoretical foundations for high throughput,high performance and low complexity MIMO receiver design and is very enlightening for practical design.