Research On Detection Algorithms Of Communication Systems With Multiple Antennas

The next-generation mobile communication systems can provide high data rate and high quality data transmission for subscribers . There are the three key techniques of the physical layer: MIMO,OFDM and channel coding. MIMO technique significantly improves spectral and power efficiencies at no additonal cost of frequency-spectral. Compared with single carrier systems, OFDM systems have higher spectral efficiency and can effectively combat frequency-selective channel fading in broadband systems. Channel coding technique is a indispensable means to achieve near system capacity communications. We study the detection and channel estimation algorithms of narrowband and broadband MIMO systems. The main contributions are as follows:1. IC-MMSE iterative detector is a low-complexity detection algorithm based on MMSE filtering and decoding feedback information, which makes a good compromise between complexity and performance, and is therefore very suitable to V-BLAST systems with large modulation constellations and large number of antennas. We propose a modified IC-MMSE iterative detector with estimated channel state information. Theoretical analyse proves that the modified detector can acquire smaller MMSE filtering output mean-square error. Numerical results show that compared with the traditional IC-MMSE detector (i.e., the IC-MMSE detector that neglect channel estimation error), considerable signal-to-noise gains can be obtained without additional complexity.2. Low complexity algorithm for near optimum detection of V-BLAST systems is currently an important research area. We propose a novel near optimum detection algorithm with low and fixed complexity, namely the MG-C detector. Its property of near optimum performance comes from the utilization of a multiple output decision feedback strategy based on MMSE filtering. Whereas its property of low compexity comes from its adopting Cholesky factorization for calculating dominant symbol combinations and LLR values. Numerical results show that the algorithm can obtain near optimum performance with lower complexity.3. We draw a thorough research on existent and newly proposed V-BLAST detection algorithms, and propose two equivalence theorems. Firstly, among the wellknown suboptimum detection algorithms, the PDA algorithm is equivalent to the MMSE-ISDIC algorithm in terms of performance, where the later has been widely applied in detection of CDMA systems. Secondly, for near optimum detection algorithms, the newly proposed (on June 2007) orthogonal projection based K-best algorithm is equivalent to the K-best algorithms based on QR and Cholesky factorization. The proposal of two equivalence theorems connects the researches of the above two aspects, and can avoid duplicate works of researchers, or research results with the same performance but different names.4. The detection algorithm for MIMO-OFDM systems is currently one of the research focuses. We study the detection and channel decoding algorithms for BICM MIMO-OFDM systems over frequency- and time-selective (i.e., doubly selective) fading channels, and propose an iterative detection algorithm that approaches the performance of MAP detector with ideal ICI cancellation, namely the quasi-MAP-ISDIC detector. The new detection algorithm combines MAP and ISDIC detection techniques, and adopts the idea of group optimum filtering processing to reduce the complexity of MAP detection. Based on the different roles the detection module regards the symbols as, i.e., the symbols for MAP detection or interferences, the Turbo decoding module respectively provides extrinsic information and soft decision togather with residual interference variance for the detection module. With system-level iteration, the algorithm can approach near MAP detection with ideal ICI cancellation in performance with lower complexity.5. We study channel estimation algorithms for MIMO-OFDM systems over doubly selective fading channels. We adopt frequency-domain subcarrier cluster as pilot symbols, this pilot mode takes the distribution property of ICI into consideration and is therefore suitable to channel estimation of MIMO-OFDM in doubly selective channels. With respect to a newly proposed polynomial approximation based LS estimator, we propose a restriction theorem on its parameter selection, and show the limitation of the algorithm in scenario where the wireless channel has a larger normalized Doppler frequency. Besides, we propose an MMSE channel estimator based on optimum linear preprocessing and polynomial approximation, namely the OLP-MMSE-POLY estimation algorithm. The proposed algorithm reduces the signal modeling error introduces by ICI and the noise enhancement effect of the LS estimator with a larger polynomial order, and therefore acquires significantly improvement in estimation performance.