Adaptive Wavelet Inference, Prediction,and Feature Extraction for Tactical Data

The Navy has unprecedented capabilities for gathering sensory information through platforms such as autonomous vehicles. This information is exploited for a variety of purposes, including rapid retargeting, situational awareness, and surveillance. A critical need facing the Navy is software for automated browsing and processing of sensory data in order to make faster and more accurate decisions. Nonparametric statistical analysis is fundamental to automated processing in tactical warfare, where sensory data is frequently used to predict the outcome of a response. We propose to develop a new approach for nonparametric statistical inference and prediction using the adaptive wavelet transform. We will address several broad application areas in tactical and electronic warfare, including detection, classification, interference rejection, and noise suppression. Our research will emphasize general methods guided by statistical or empirical considerations. A range of noise and clutter conditions, including impulsive noise, is considered. The ultimate objective of our research is the development of a flexible and extensible software system. This system will support two primary applications: a computing environment for rapid prototyping of nonparametric statistical procedures for analysis of tactical data, and a data browser based on automated detection and classification tools.