Channel Estimation Of MIMO-OFDM System

MIMO-OFDM is regarded as one of the most important technologies of the 4th generation mobile telecommunication systems, which can not only effectively enhance the transmission rate and the capacity of the wireless communication system, but also combat multi-path fading and interference. The technique is also widely regarded as a most promising and efficient solution for the wideband multimedia operation, because of its high ability of anti-fading, high channel capacity and high bit rate. We can foresee that MIMO-OFDM will achieve greater attention in the present future.Channel estimation is one of the most important aspects in the mobile telecommunication systems; it has achieved its significant status for signal recover, transmitter-receiver strategy and resource allocation. Especially in MIMO-OFDM system, channel state information is required in OFDM systems for correlation detection and in MIMO-OFDM systems for space-time decoding, therefore, channel estimation's accuracy directly affects the overall performance of MIMO-OFDM systems.In this paper, we investigate several MIMO-OFDM channel estimation methods based on training sequences.Firstly, we introduced the definition of the training sequences and discuss the optimum training sequences in MIMO-OFDM system, we propose two training sequences: diagonal orthogonal pilot and phase-shift orthogonal pilot, and we analyze the channel estimation performance based on each scheme.Secondly, in order to reduce the system PAR and enhance the system's performance, we explore the channel estimation method based on a special training sequence whose IFFT is a chirp sequence with constant amplitude, inserting zeros in every odd sub-carrier index can increase the performance of the channel estimation for the systems in the performance of ISI, we propose use this sequence as one antenna's training symbols. In order to keep the orthogonal property of the training sequences in frequency domain, we propose two methods, one of them is introducing a constant phase rotating in frequency domain, the other is carrying the constant shift in the sequences, which all resulting a constant time shift in the time domain, then these several sequences don't interfere with each other. Simultaneously, in order to reduce the complexity of the algorithm and to eliminate the noise, we also propose real-time estimation algorithm for the delay spread of the channel.Lastly, we propose one kind of orthogonal expansion of the training sequences for the MIMO-OFDM system with 2N transmitting antennas. We expand the optimization training sequences of the SISO-OFDM system to be suitable for the MIMO-OFDM system, the optimized designed sequences, may transform the multi-antennas channel estimation question to the single-antenna channel estimation question. And we can simulate with the Matlab, we can see bit error code characteristic compared with different estimate algorithm. we investigate three channel estimation methods with training sequences, one is LS method, which is low complexity and easy to realize, but it is sensitive to the noise, System's performance degrades quickly when the SNR is low, the LMMSE method provides a better accuracy than the LS method, but the complexity of LMMSE is high relative, to reduce the complexity of LMMSE method, we propose a low rank LMMSE method based on SVD whose performance is close to the LMMSE when the SNR is low and the complexity of the low rank LMMSE is lower.