On Synchronization Algorithms For MIMO-OFDM Systems Over Frequency Selective Fading Channels

Both of high quality and high data rate will be required in the future wireless communication systems. Orthogonal Frequency Division Multiplexing (OFDM) is robust to frequency selective fading and can achieve high spectral efficiency. Meanwhile, Multi-Input Multi-Output (MIMO) technique can provide high throughput. The combination of OFDM and MIMO has been regarded as a promising solution for future wireless communication systems.Synchronization is the primary procedure for signal processing, which influences the performance of following baseband signal processing. OFDM system is sensitive to synchronization errors. Very small carrier frequency offset (CFO) may degrade system performance severely. Correct symbol timing and precise CFO estimation are necessary for OFDM systems. When introducing MIMO technique into OFDM system, synchronization error may lead to severe inter-antenna interference. Hence the synchronization is much more important in MIMO-OFDM systems.In this thesis, synchronization theories and algorithms for MIMO-OFDM systems are deeply analyzed and researched over frequency selective channels, and several new synchronization algorithms are proposed and compared with key conventional algorithms. The main contributions of this thesis can be described as follows:Firstly, the influences of symbol timing error and CFO on OFDM system performance are mathematically modeled and analyzed. The results are confirmed by computer simulations, which also provide theoretical foundations for synchronization algorithm designs in the following chapters.Secondly, synchronization algorithms based on the time domain correlation peak are analyzed. It is shown that over multi-path channels, the correlation length affects the performance of synchronization algorithms based on the time domain correlation peak significantly. The first arriving path can be effectively estimated by detecting the maximum value of the gradient of the correlation value, when the correlation length exceeds the largest multipath delay. Two improved synchronization algorithms based on the time domain correlation peak are proposed. The first one uses only one OFDM symbol as the training sequence and performs differential correlation at the receiver to eliminate the peak plateau of the timing metric. This algorithm can achieve considerable symbol timing accuracy with a steep roll-off timing metric trajectory. The second one is based on the correlation between the received signals and the copy of pilots at the receiver. Symbol timing is estimated depending on the ratio of multi-path power to noise power, which makes the timing metric peak more obvious. With multiple antennas, the number of multipaths increases linearly as the number of antennas increasing, which brings more available multipath diversity gains and can improve algorithm performance.Thirdly, the maximum-likelihood CFO estimation algorithms based on null-subcarrier are investigated, and it is shown that the maximum-likelihood CFO estimation algorithm based on equispaced null-subcarriers is equivalent to the CFO estimation algorithm utilizing the correlation of repeated training sequences. The further analysis indicates the necessary conditions of the channel zero numbers and locations, when the CFO estimation algorithm is identifiable over frequency-selective fading channels. Theoretical analysis shows that with equispaced null-subcarriers or data subcarriers, the CFO estimation are independent to channel zero locations.Fourthly, subspace-based blind synchronization algorithms are investigated, and an improved MUSIC (Multiple Signal Classification) algorithm is proposed for the uplink CFO estimation in IFDMA systems. Unlike conventional MUSIC algorithms which require multi-dimensional search over the parameter space, the proposed algorithm utilizes a special designed CFO estimation factor and makes the multi-user CFO estimations for the uplink in IFDMA systems be a one-dimensional estimation, and is much more efficient compared with conventional algorithms.Lastly, the synchronization problem in distributed MIMO-OFDM systems is investigated, and an improved synchronization algorithm is proposed. The proposed algorithm utilizes the good cross-correlation properties of Gold sequence to reduce the inter-antenna interference, and makes the estimations of the delay and CFO for different antennas more efficient. Compared with conventional synchronization algorithms for distributed MIMO-OFDM systems, the proposed algorithm offers more accurate timing estimation.The main target of the thesis is to provide new theories and algorithms for the synchronization problem for broadband wireless communication systems. The proposed algorithms have high efficiency and are more applicable compared with conventional algorithms.