The dramatic growth of interest in wireless communication systems has motivated an attendant interest in signal processing techniques for data demodulation in multiple-access communication channels. Multi-user detection provides a useful theoretical basis for the study of optimum signal processing in such channels. One principle established through the study of multi-user detection is that multiple-access channels can be reduced to ambient-noise-limited channels with appropriate signal processing. The primary framework within which multi-user detection has been developed has been the Gaussian code-division multiple-access (CDMA) channel, in which the ambient background noise is white and Gaussian. Experimental evidence suggests that the ambient noise in many channels of interest in wireless applications (e.g., urban and indoor radio channels, and underwater acoustic-modem channels) is significantly non-Gaussian. In particular, it is known that such noise environments can be better modeled as being derived from more general linear processes, in which both Gaussian and impulsive components can arise. It is further well-established in the context of single-user communications that non-Gaussian noise can be quite detrimental to the performance of systems optimized for Gaussian noise, whereas it can be quite beneficial if appropriately modeled and ameliorated. Thus, as work in multi-user detection moves closer to practice, it is of interest to examine the multi-user detection problem in the context of more realistic ambient noise models. The proposed study will address this issue in depth. Specifically, three research problems arising in the area of multiple-access communications in non-Gaussian channels will be addressed - the analysis of extant multi-user detectors in non-Gaussian ambient noise; the design of optimum and near-optimum multi-user detectors for non Gaussian channels; and the design of waveform sets for non-orthogonal signaling through non-G aussian channels. The results of this study should be quite beneficial in practical communication systems, particularly in areas such as urban mobile telephony, indoor wireless communications, and tetherless underwater acoustic communications (which is of emerging interest in oceanographic instrumentation). For the foreseeable future, there will be considerable commercial and scientific interest in techniques that allow wireless networks to operate with better performance for more users/subscribers and with lower transmitter power requirements. This means that signal processing techniques that can significantly improve the performance of wireless systems in real channels, such as those proposed for study here, will be of significant interest to the commercial and scientific sectors.
