UC San Diego

In this dissertation, we consider the performance analysis of interference suppression techniques for multiple antenna systems. In the first part of the dissertation, the performance of multicode direct sequence spread spectrum Multiple-Input Multiple-Output (MIMO) systems in the presence of frequency selective fading is evaluated. We derive the asymptotic distribution of the multi-antenna interference when the processing gain is sufficiently large. The probability of error is derived for the conventional RAKE receiver, and its performance is compared for various system configurations. We consider system tradeoffs for both fixed rate and fixed diversity. For a fixed total data rate, we demonstrate the advantage of decreasing the number of transmit antennas while increasing the number of codes, and for a fixed total diversity order, we demonstrate the advantage of decreasing the number of RAKE taps while increasing the number of receive antennas. In the second part of this dissertation, interference suppression is achieved by precoding at the transmitter. The performance of a pre- BLAST-DFE technique with Tomlinson-Harashima precoding and decentralized receivers, operating over Multiple-Input Single-Output (MISO) frequency selective fading channels, is evaluated. First, we derive the probability of symbol error for Tomlinson-Harashima precoding operating over Single-Input Single-Output (SISO) channels with Intersymbol Interference (ISI). We then generalize this result to MISO frequency selective fading channels with decentralized receivers by using the QR decomposition technique. The effect of optimal ordering of the decentralized receivers, which minimizes the total transmit energy, is investigated. We obtain a closed-form expression for the probability density function of the squared diagonal elements of the upper-right triangular matrix belonging to the optimal QR decomposition when two transmit antennas and two receivers are used. We also provide simulations to corroborate the analytical results. Finally, we consider the effects of channel estimation errors on the performance of a system employing Tomlinson- Harashima precoding and the QR decomposition, operating over MISO frequency-flat fading channels with decentralized receivers. The QR decomposition of the channel matrix is employed to arrive at an equivalent channel where successive interference cancellation at the transmitter can be used to remove the effect of the multiuser interference