Subspace Signal Processing in Structured Noise
Doctoral thesis, Oct 1988-Dec 1990
COLORADO UNIV AT BOULDER DEPT OF ELECTRICAL AND COMPUTER ENGINEERING
Pagination or Media Count:
Some common types of noise can be dealt with by applying a linear model to the noise as well as to the signal. The noise obeys a low-rank linear model as structured noise and derive several signal processing methods based on a structured noise model. Whereas orthogonal projection operators play a key role in the solution of classical linear estimation and detection problems, the addition of a structured noise term to the model leads to oblique projection operators in the new solutions. Consider several subspace identification problems in the context of a structured noise model. Also consider parameter estimation with structured noise, where the signal and structured noise subspaces are known or have been identified from observed data. These results are applied to the decoding of complex number codes for detection and correction of impulse errors.