Abstract
Fundamental information-theoretic quantities, including conditional entropy of source location given received complex spectral values and per-iteration information gain (relative entropy) are applied as performance metrics to the optimization and real-time adaptation of receiver array spatial configurations for iterative (sequential) Bayesian localization of narrowband acoustic sources. Computational examples illustrate the application of these performance metrics to the adaptation of mobile-sensor spatial positioning and to the optimization of element positions for fixed vertical arrays in a shallow-water waveguide. Evolutionary search algorithms [Back and Schwefel, Evolutionary Computation 1(1), 1–23 (1993)] are investigated as a unified computational approach to both optimization problems. The optimized array spatial configurations are compared with configurations optimized with respect to traditional (energy-based) performance metrics, and the differences are interpreted. [Work supported by ONR.]
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call