This Small Business Innovation Research Phase I project will conduct research on the methodology and application of time-frequency analysis, and develop a broadly applicable time-frequency toolbox providing advanced time-frequency signal processing techniques to the non-expert practitioner. The goal of the program is to make time-frequency analysis readily accessible to a wide range of users and non-specialists so that it becomes a standard signal-processing tool. In most real-world applications, nonstationary signals are the norm. Everyday applications such as communications, speech, sonar, radar, manufacturing, and biomedical signal processing routinely process nonstationary data or must function in the process of nonstationary interference. Standard methods of processing nonstationary signals, however, suffer from a trade-off between time and frequency resolution, resulting in degradations in accuracy and performance. Powerful modem techniques for processing nonstationary signals, known as "time-frequency representations," overcome the shortcomings of the traditional methods. Time frequency analysis has been successfully applied to a broad range of applications. Due to limitations in existing software and methodology, however, these methods have not seen widespread use by non-specialists. Moreover, a comprehensive collection of software tools for computing time-frequency representations and incorporating the results into standard analyses or applications is not currently available. This program should provide a software toolkit offering a complete methodology for time-frequency analysis. The toolkit will be commercialized as an S-PLUS module and a MathCad extension pack, and will be marketed to the installed base as well as other scientists and engineers. Instructional material (books and hypermedia) will also and the technology licensed to other vertical markets (e.g., communications).
