Blind signal separation in dynamic environments
Blind source separation (BSS) is a powerful statistics analysis tool capable of revealing hidden mechanisms and source signals from their combinations. It has a wide variety of practical applications in areas like image and speech processing, telecommunications, financial engineering, biomedical signal processing, and text document analysis, etc. The major advantage of BSS is that little knowledge is required about the mixing process of the source signals. We have been working on adaptive BSS algorithms for dynamic environments, i.e., the source signals' mixing process is varying with time rapidly. For example, in mobile cellular communication applications, the user may be constantly moving, and may experience “handover” between two service towers. The key to source separation in this scenario is to assure fast convergence of the adaptive algorithm. Recently, we developed a family of algorithms for both real and complex valued signals with superior convergence properties. Successful computer simulations were performed for interference suppression in wireless receivers. Currently, we are working on interference suppression techniques for receivers operating in both frequency-selective and time-varying channels, which represent the most challenging communication environment. This research is in collaboration with the Digital Signal Processing Lab at University of Central Florida.
Electrical,Computer, Software, & Systems Engineering
Dr. Yang performs research in adaptive signal processing, automatic target recognition and wireless communication systems. He is the president of the local chapter of the IEEE, and very engaged in both student projects and the development and assessment of novel teaching techniques.