Performance Improvement of Downlink Beamforming Algorithms in Wireless Channels

Abstract

This dissertation is concerned with the use of current Smart Antenna Systems (SAS) for utilizing the spatial dimensions of the wireless channel to improve upon the capacity of wireless systems. This enhancement is achieved for downlink Multiple Input Single Output (MISO) wireless channels by augmenting link quality by the provision of beamforming, diversity and interference cancellation gain in SAS. One such system consists of a Transmit Beamformer (TB) at the base Station (BS) which provides beamforming gain, interference cancellation gain and, on provision of feedback from the Mobile Station (MS), gives beamforming gain even in channels undergoing independent fading at the antennas of BS. The beamforming gain of the TB is reduced in those fast fading channels that are uncorrelated across the antennas of the Transmit Antenna Array (TAA). Currently researchers are actively pursuing TB techniques based on element space beamforming. In an Element Space Transmit Beamformer (ES TB), each antenna weight of the TAA is individually adaptive to the varying channel conditions. This thesis, in contrast, presents an Adaptive Transmit Beamspace Beamformer (ATBBF) that uses beamspace beamforming to provide beamforming gain. An ATBBF consists of several orthogonal Transmit Beamformers (TBs) on the TAA. In beamspace beamforming a specific set of antenna weights determines an orthogonal direction of the TB. This weight vector (corresponding to each orthogonal beam) is further weighted by an adaptable weight. This weight, called the beamspace weight, thus adaptively weighs one orthogonal beam and is updated based on the changing channel conditions. An ATBBF deploys this set of orthogonal beams in space to transmit signals to a MS. The superposition of adaptively weighted orthogonal beams in space forms the beam pattern of an ATBBF directed in the desired direction. Simulations in single, double and triple multipath static wireless channels demonstrate the unique evolution and convergence of the ATBBF's beam pattern in relation to the convergence of its beamspace weights. The relationship between the magnitude and convergence time of a beamspace weight upon the multipath wireless channel is also elaborated. Furthermore, for the above channels, performance curves of both implementations of an ATBBF i.e. Full Dimension Adaptive Transmit Beamspace Beamformer (FD 14ATBBF) and Reduced Dimension Adaptive Transmit Beamspace Beamformer (RD ATBBF), along with that of an ES TB (having similar feedback mechanism) are simulated with uniform initial adaptable weights. Their analyses reveal that contrary to the RD ATBBF and ES TB, the FD ATBBF converges to optimum performance in all channels. The FD ATBBF also has a lesser convergence time than the TB in most double and triple multipath channels. The performance metric of an ATBBF is derived and analyzed in a dynamic channel undergoing Rayleigh fading independently at the antennas. A performance comparison between a FD ATBBF, RD ATBBF and an ES TB is made in terms of convergence and tracking of various slow and fast fading channels by simulations. Comparisons show that the FD ATBBF gives a performance equivalent to that of an ES TB and outperforms the RD ATBBF. Thus the FD ATBBF can provide beamforming gain and fading diversity similar to that of an ES TB. The performance of the FD ATBBF improves on increasing the number of antenna elements of the TAA from two to four i.e. from 2.91 dB in slow fading to 0.01 dB in fastest fading dynamic channel. Bit Error Rate (BER) studies for the above dynamic channels in the presence of noise are carried out to compare the BER curves of the FD ATBBF, ES TB and RD ATBBF with non beamforming techniques like Space Time Codes (STC). The studies confirm the performance analysis above by noting that the BER curves of the FD ATBBF and ES TB are similar in slow and fast fading channels and both outperform the RD ATBBF. The FD ATBBF also gives a gain of 2.2 dB and 4.8 dB over STC in slow fading for two and four antennas respectively. Furthermore, for four antennas, the FD ATBBF and RD ATBBF offer double data rate than the best performing STC technique, under constrained bandwidth requirements. In order to improve the performance of an ATBBF in fast fading channels for two antennas, a simple combination of an ATBBF with STC called the Space Time Coded Adaptive Transmit Beamspace Beamformer (STCATBBF) is proposed. This technique is made possible by utilizing the orthogonal beams of an ATBBF. The BER curves of the STCATBBF are simulated and compared with the FD ATBBF and STC in fast and slow flat fading channels. It is shown that in fast fading channels the STCATBBF gives a gain of 2.2 dB to 5.4 dB over the FD ATBBF, while in slow fading it outperforms STC .