Parallel Searching-Based Sphere Detector for MIMO Downlink OFDM Systems

Abstract

Abstract-In this paper, implementation of a detector with parallel partial candidate-search algorithm is described. Two fully independent partial candidate search processes are simultaneously employed for two groups of transmit antennas based on QR decomposition (QRD) and QL decomposition (QLD) of a multiple-input multiple-output (MIMO) channel matrix. By using separate simultaneous candidate searching processes, the proposed implementation of QRD-QLD searching-based sphere detector provides a smaller latency and a lower computational complexity than the original QRD-M detector for similar error-rate performance in wireless communications systems employing four transmit and four receive antennas with 16-QAM or 64-QAM constellation size. It is shown that in coded MIMO orthogonal frequency division multiplexing (MIMO OFDM) systems, the detection latency and computational complexity of a receiver can be substantially reduced by using the proposed QRD-QLD detector implementation. The QRD-QLD-based sphere detector is also implemented using Field Programmable Gate Array (FPGA) and application specific integrated circuit (ASIC), and its hardware design complexity is compared with that of other sphere detectors reported in the literature.Index Terms-ASIC, bounded soft sphere detector, Field Programmable Gate Array (FPGA), implementation complexity, latency analysis, QL decomposition (QLD), QR decomposition (QRD), QRD-M, QRD-QLD-based parallel candidate search.

Block-Wise QR-Decomposition for the Layered and Hybrid Alamouti STBC MIMO Systems: Algorithms and Hardware Architectures

“…With the QRD of 1 , the ML detection becomes (5) Tree search detectors [17], [18], [20]- [22] for the SDM MIMO system can be straightforwardly applied to find the exactly or nearly ML solution. But, such detectors do not fully exploit the Alamouti structure possessed by the system.…”

“…has Alamouti sub-blocks defined in (3) and diagonal (22) Note that Alamouti STBC signals are transmitted by the lower part of the transmit antennas in Fig. 10; and, they are included in the lower part of the transmit signal vector in (20).…”

“…There exists one disadvantage possessed by our BCGR based approach, i.e., our approach can only compute the QRD of the smaller channel matrix for the SDM MIMO system. This disadvantage is revealed by the signal model given in (20)- (22). In contrast, the TACR based approach can compute the QRD of an arbitray matrix; but, it can not compute the QRD efficiently if the matrix is comprised of Alamouti sub-blocks.…”

“…The signal model for the layered Alamouti STBC system can be written in a linear form [6]- [14]. Thus, exactly or nearly optimal tree search detectors [17]- [22], which have been used extensively to detect SDM MIMO signals, can be applied straightforwardly to detect layered Alamouti STBC signals [13], [14]. It is well-known that the preprocessing QR-decomposition (QRD) [23]- [29] is inevitable to the tree search detectors.…”

Reduced-Complexity k-best Decoder for LTE Standard

“…The parallel merge algorithm was proposed in [9] to increase the throughput of the existing conventional K-best architectures. In literature, some researchers divide the tree search into several parts [10,11] and implement the K-best algorithm but still the complexity of the search does not decrease further.…”