Cyclostationary Signal Processing in Digital Communication Systems

摘要

Our focus in this ARO research project is to investigate and study the application of cyclostationary signal processing in digital communication systems. We consider several important areas of application including channel equalization, co-channel interference rejection, and antenna beamforming. During this report period, we developed a number of new algorithm for the blind identification and equalization of multiple input multiple output systems. These methods are more robust and accurate than many existing methods. Blind separation of signals using both the higher order and second order (cyclostationary) statistics are studied. Several simple methods are proposed. We also developed a finite window decorrelator receiver for asynchronous CDMA systems. This new algorithm is near far resistant even when the processing window is rather short, overcoming the weakness of the conventional decorrelator that relies on large (almost infinite) window size. In another important study, we investigate the applicability of blind equalization algorithms in practical wireless cellular systems such as the GSM. Since the GSM transmission is nonlinear and is in burst mode, blind equalization algorithms must be adopted for this nonlinear modulation and must converge within each frame of data burst. Using a linearization method, we simplied the nonlinear GMSK signal into an equivalent linear QAM signal. A de-rotation scheme further allowed channel diversity be extracted without the need of additional downlink antennae. Successful blind equalization and semi-blind equalization results for GSM are established.