Treffer: Eigensystem based techniques for blind channel estimation and equalization

Channel estimation and equalization form an integral part of modern communi-cation systems. Proper equalizer design can counteract channel distortions such as intersymbol interference (ISI), additive noise, and multiuser interference. The performance of the equalizer, on the other hand, relies on how well the channels are estimated. These two problems are thus intertwined. Traditional channel estimation techniques rely on the transmission of train-ing symbols that enables the receiver to estimate the channel state information (CSI) before actual data are sent. This method consumes a significant portion of the valuable bandwidth, especially in wireless environment where bandwidth and spectrum are scarce. Moreover, this is simply not possible in some applica-tions. The channel estimation problem can be solved by using blind estimation techniques where the CSI can be identified by using only the received signal. In this thesis, we investigate novel eigensystem based methods to estimate and/or equalize finite impulse response (FIR) single-input single-output (SISO), and multi-input multi-output (MIMO), and space-time systems. We propose a minimal redundancy block based space-time precoder-equalizer system using second-order statistics (SOS) that can blindly equalized the channel without re-quiring the amount of transmit redundancy to be greater than or equal to the channel order. Using the precoder-equalizer system, we can relax the require-ment of having more receive antennas than transmit antennas. This will lower the cost of the receiver which is crucial in the case of downlink communications. Besides the economic advantage, the space-time precoder-equalizer system can also reduce the chance of encountering non-equalizable channels. We then propose a blind channel estimation algorithm that exploits higher-order statistics (HOS) of the received signal to blindly estimate FIR SISO chan-nels. Using HOS techniques, we can alleviate the problem of Gaussian distributed noise; thereby improving the mean squares error ...