A Novel training-based MIMO channel estimation scheme for layered space-time systems in frequency selective wireless channels

摘要

New development in wireless technology using multiple antennas with appropriatespace-time processing has recently become the new frontier of wireless communicationsystems due to the potential for providing very high spectral efficiency and enormouscapacity improvement over the conventional wireless radio communications. The technicaladvances in using the multiple-input multiple-output (MIMO) wireless links present apromising breakthrough in resolving the bottleneck of current capacity limitation for futureintensive wireless networks. The MIMO wireless systems utilize multiple antennas at bothside of the transmitter and the receiver for enormous gains in spectral efficiency as well assystem capacity in terms of higher data throughput by exploiting the multipath diversity ina rich scattering environment. A number of MIMO systems have been proposed to permitvery high transmission rate, far exceeding the conventional communication technique. Inparticular, the Bell Laboratories layered space-time (BLAST) architecture has beenpresented that uses concept of spatial diversity and successive interference cancellationtechnique to improve the quality of signal reception over the flat-fading or the frequencyselective fading channel.However, in order to achieve the quoted capacity gains in MIMO systems, thechanneli nformation in terms of the multiple channeli mpulse responses(C IRs) and theirfading coefficients must be known or estimated, which requires the design of a suitablechannele stimator.T hus far, existing MIMO channele stimations chemesh aveb eenm ostlylimited to the flat-fading case or cater specifically for coded space-time systems such asspace-timeb lock code systems.I n this thesis,t he work is to considert he existing MIMOchannel estimation techniques (used in the flat fading condition) and extend them to caterfor a more realistic time-varying, frequency selective fading channel. The focus of thisthesis has been the design and development of suitable training-based MIMO channelestimation scheme as well as the formulation of a new pilot code to enable effectiveestimation for the frequency selective channel. The novel channel estimator is alsoincorporatedi nto the BLAST architecturet o allow the practical assessmenotf using nonidealizedchannel to be studied and analysed for the performance of the MIMO systems.The driver for this work has been the recognition of the importance of channel knowledgefor all the MIMO system to be used in practical application.