The Subspace Based Blind Channel Estimation Algorithm In MIMO-FDE Systems

Multiple input multiple output (MIMO) technology can dramatically improve the channel capacity and achieve high efficiency of spectrum without increasing bandwidth and the transmission power. The frequency domain equalization (FDE) technology, which represented as orthogonal frequency division multiplexing (OFDM) and single carrier cyclic prefix (CPSC), can effectively combat multi-path fading and have low complexity to receiver design. By combining these two technologies, MIMO-FDE has been widely regarded as one of the most promising techniques for the next generation wireless communication systems. However, the successful performance of MIMO-FDE depends upon the availability of accurate channel state information (CSI) concerning the wireless communication channel.Channel estimation algorithms can be roughly sorted into two basic categories: pilot or training sequence aided algorithms and blind algorithms. The blind method only requires the transmission for a few pilot symbols to remove the estimation ambiguity and thus increases the transmission bandwidth efficiency. The so-called subspace-based blind channel algorithms are the most popular blind algorithm in the family of blind channel algorithms. By exploiting the second-order statistics of the received signals, the subspace-based blind channel algorithms can obtain the channel parameters blindly according to the orthogonality property between signal and noise subspaces. In this dissertation, we focus on the subspace-based blind channel method in MIMO-FDE systems. The main contents are as follows:1. A novel blind channel method based the signal eigen-vectors is proposed for MIMO-FDE systems, which the number of receiver antennas is greater than the transmitters. The proposed method transformed the channel estimation into a set of linear equations generated by the signal eigen-vectors in term of the structure of the system channel matrix, which fully uses the signal eigen-vectors and improves the accuracy of the channel estimation effectively.2. Aiming to decrease the number of the receiver blocks, we proposed a blind channel estimation method by shifting and exentending the cyclic prefix (CP) signals of received blocks and utilizing the property of block circulance of MIMO-FDE system channel matrix. The subspace-based blind channel algorithms need lots of observation signals to ensure getting a reliable the second-order statistics information of the observation signals. However, if the needed receiver signals are too much, the channel state may have significantly changed during one channel estimation because of the invariant property of wireless channel, which would deteriorate the system performance. This blind channel estimation method can effectively estimate the channel with fewer received blocks.3. The channel can be identified if and only if the channel matrix is full column rank. Thus it is necessary to introduce some redundancy at the transmitter to satisfy the condition of channel identifibility as the number of transmitter antenna is equal to the number of receiver antennas. A subspaced based blind channel method in the context of CPSC systems that exploits the repetition scheme between two consecutive blocks, which guarantee the channel can be estimated without using CP signals. Furthermore, as the repetition signals in each block is great than unity, the need received blocks will be significantly reduced by shifting the signals circulantly, which makes it is suitable for estimation of fast varying channel.4. A precoded channel estimation algorithm was proposed for MIMO-FDE systems. Due to the structure introduced by the precoding matrix, the channel can be estimated at the receiver based on simple cross-correlation operations. The precoded channel estimation algorithm can apply in the MIMO systems with the arbitrary number of the receiver and transmitter antennas. This method works with precoders to identify even channels with zeros on the unit circle and is insensitive to the over-estimated the channel order. It is concluded that the precoding scheme makes an SNR loss inevitably and there is a tradeoff between the accuracy of channel estimation and the decision error. The precoded blind channel estimation algorithm doesn't collect two or more received blocks to calculate the correlation matrix, thus it has lower complexity compared with orther blind channel estimation algorithms.