|Blind multidimensional matched filtering for wireless channelswith multiple antennas
The hardware complexity is one of the most crippling handicaps of implementing a blind channel estimation scheme in a wireless receiver. The state-of-the-art blind channel estimation methods either need matrix decomposition methods, which are implementation inefficient and expensive, or higher order cumulant functions of the received data, which eventually leads to the need for long data records for accurate estimation of higher order statistics of the received signal. The Blind Matched Filter (BMF) receiver on the other hand makes use of an adaptive algorithm, i.e. the Constant Modulus Algorithm (CMA), to estimate the channel without the need for an implementation inefficient approach as stated above. The BMF method assumes that the receiver utilizes channel matched filtering before the equalization of the channel. For receivers, employing multiple antennas, it is well known that the optimal detection strategy is to pass the received signals through a multidimensional matched filter. For this reason a signal transmission strategy similar to that of the BMF receiver would definitely benefit communication systems employing multiple antennas.
In this thesis four novel approaches have been proposed, all implementing multidimensional matched filtering blindly in their architecture, for wireless communication channels with multiple antennas. Two of these approaches are for channels utilizing multiple antennas only for signal reception. These two approaches differ from each other in the channel equalizer structure that they employ. The first approach is the blind implementation of the correlation-based Decision Feedback Equalizer (DFE). The second approach equalizes the Single Input Multiple Output (SIMO) channel and obtains its estimate by making use of the multichannel DFE. The multidimensional matched filter equivalents were established through the use of an adaptive filter as well as the channel equalization being performed blindly for both novel receivers. It has been shown in this thesis that the equalization performance of the proposed methods are close to that of the matched filter bound, and to support this finding, a comparison between the matched filter estimation error performances of the proposed techniques is also provided. A short discussion to improve the convergence speed of the proposed approaches is also included.
The other two novel receivers are for the channels utilizing multiple antennas in their transmitter sides, i.e. Multiple Input Single Output (MISO) and Multiple Input Multiple Output (MIMO) channels. Different from the state-of-the-art blind techniques in the literature proposed for the MISO channels, in the novel MISO receiver reported in this thesis the channel estimation and channel equalization are implemented all in one receiver without any need for an extra step to perform any one of these two. The blind receiver does also perform the decoding of a coded data stream, deploying a space-time block coding scheme, to realize the spatial diversity in the channel. Step by step derivation of the novel receiver is shown in this thesis alongside the simulation section, where the performance of the novel MISO receiver is illustrated and vindicated.
The MIMO counterpart of this receiver was also explained as the fourth novel receiver. Different from the other three, this receiver further benefits from spatial multiplexing. As it was mentioned before, the BMF receiver is advantageous over the state-of-the-art blind schemes due the simplicity in its implementation. However, the realization of the BMF receiver requires the noise variance to be estimated and the equalizer parameters to be calculated in state-space with costly matrix operations. In this thesis, a novel architecture is proposed to simplify a potential hardware implementation of the BMF receiver. The novel approach transforms the BMF receiver into a computationally efficient all-adaptive format which replaces the matrix operations. Furthermore, the novel design does not have any need for any extra step to estimate the noise variance. Comparative channel equalization and channel identification performance analysis of the conventional and the novel all-adaptive BMF receiver is reported and evaluated through exhaustive simulations in this thesis.
|University of Westminster
|Digital Object Identifier (DOI)