Hybrid Bayes/GLRT Signal Detection
Abstract
Develop a new and simple scheme to detect short-duration underwater signals. Present methods tend to be overly complicated and/or too finely tuned to one type of signal. The proposed technique uses a Bayesian approach with "hyper"-parameters, meaning that the model can adapt itself to a wide range of possible signals and signal types. Obtain a new formulation of the GLRT that avoids enumeration and is computationally feasible by replacing intractable enumeration over possible signal characteristics with an a priori signal distribution, and by estimating the hyperparameters (of the prior distribution) jointly with other signal parameters, it is possible to It turns out that this estimation can be done very efficiently and neatly via the estimation-maximization (EM) algorithm. This approach relies on a coherent statistical model, and is easily and rationally extended in a number of different directions, such as using assumptions of energy contiguity in time and frequency. The objectives are to realize these extensions, and to compare their performance with existing transient detection algorithms.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1998
- Accession Number
- ADA552200
Entities
People
- Peter Willett
Organizations
- University of Connecticut