Space-time processing algorithms for smart antennas in wireless communications networks

Antenna arrays sample propagating waves simultaneously at different places. Space-time processing refers to digital signal processing techniques that employ antenna arrays to increase both the capacity and performance of wireless systems. Blind channel equalization/estimation algorithms based on single input multiple output data models (one user transmitting to an antenna array) or multiple input multiple output data models (multiple users transmitting to an antenna array) form a specific class of space-time algorithms that generalize results both from array signal processing and classical equalization. They estimate the channels between one or more transmitters and a receiver using the received signals only and without relying on training sequences. This thesis is concerned with the development of deterministic subspace algorithms for blind channel equalization/estimation and focuses on two aspects of such algorithms. First, subspace algorithms are mostly based on computationally intensive matrix decompositions, while data rates in wireless systems are high. Therefore we have special attention for computational complexity. We show that complexity can be decreased via the use of adaptive signal processing techniques and via specific multi-user coding schemes. The second theme is the robustness of subspace algorithms. We demonstrate that coding provides robustness against order detection problems and additionally develop algorithms that are robust against an unknown spatial color of the additive noise.