Signal Processing and Interpretation Using Multilevel Signal Abstractions.
MASSACHUSETTS INST OF TECH CAMBRIDGE DEPT OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCE
Pagination or Media Count:
This thesis advances multilevel signal abstractions as a useful conceptual framework in signal processing. More specifically, multilevel abstractions for one-dimensional harmonic spectra are developed and demonstrated to be useful in combining heuristics with algorithmic techniques in a variety of signal processing problems associated with real helicopter data, such as adjusting spectral estimation parameters and tracking fundamental frequency and power of helicopter spectra. Furthermore, multilevel abstractions for two-dimensional wavenumber spectra are used as the basis for a diagnosis system that searches for inappropriate parameter settings in an array processing system for direction determination. The diagnosis system illustrates the use of multilevel single abstractions in automating reasoning about complex signal processing systems.
- Theoretical Mathematics