Blind space-time processing for wireless communication and sensor systems

High speed wireless communications has become a reality in the past few years. With the advance of technology driven by exploding demand for cellular communications, multimedia transmission and internet connection, we are entering a new era in which high-speed wireless communications is possible and available anytime, anywhere. Though high speed wireless communications is no longer an unachievable dream, it is far from people expectation, especially when we compare the achievable speed between wireline and wireless access. One of the most important factors that limit the advance of transmission speed of wireless communications is the hostile wireless channel fading and distortion. Due to its time-varying nature continuous probing symbols called training or pilot sequences have to be sent out in a timely manner in order to track the channel variation. An attractive way to reduce training overhead is to use blind equalization techniques. Blind equalization exploits the intrinsic property of the signal or channel, and removes the channel distortion without relying on known training symbols.; Space-time processing is a new way of increasing system performance and capacity and has received a lot of attention lately. By using either beamforming or diversity combining, the signal to interference and noise ratio (SINR) can be improved to obtain higher transmission rate and multiple access interference (MAI) can be suppressed to account for more users.; This thesis explores techniques for integrating these two dimensions. We study the use of output second-order statistics to blindly estimate channel responses for various space-time systems. We propose a new projection based blind algorithm for a single-user space-time system which has lower complexity and can be implemented adaptively. For a multiple user space-time system, we derive a two-stage algorithm which consists of blind equalization first followed by blind separation.; The above blind space-time techniques are then applied to specific wireless and sensor systems. We demonstrate the various forms of the cost matrix of a VOFDM system by exploring diversity in time, frequency and space domain. For MC-CDMA systems we first exploit its specific structure and show it has a MIMO structure even with only one receive antenna. We propose a blind frequency response estimation scheme that can compute all user channels in parallel and show that the roles of receive diversity and frequency diversity are similar and interchangeable with only slight difference in the cost matrix. We also propose a new blind beamforming algorithm which can estimate the channel responses and directions of arrival simultaneously. Blind acoustic channel estimation and source separation are also investigated.