Quick-Look Signal Classification and Geolocation Based on Coherent Signal Dispersion with Machine Learning
Abstract
Approved for Public ReleaseABSTRACTThe proposed effort addresses the topic of Discovering and Denying Non-Cooperative, Reconfigurable, Agile, Ubiquitous Communication Links. The specific objectives of the proposed research are to develop and investigate methods to: detect, characterize, and classify noncooperative sources based on several quick-look signal features; employ machine learning methods for enhancing signal classification; and utilize a novel single-receiver approaches to geolocate the RF sources. The specific RF signal characterization techniques to be considered, termed STatim AGnoscis (STAG), refers to a capability for immediate signal recognition. The approach is based on a multi-channel wideband receiver architecture to detect, quickly fingerprint, and classify a significant fraction of signals within the RF environment. STAG methods are potentially applicable to sources with single antenna systems, beamsteering systems, and MIMO systems. The technology is tolerant of waveform agilities and has capability to associate frequency hopping sources across the frequency hopping band. For development and demonstration purposes, a receiver system comprising four 1-GHz instantaneous bandwidth receivers will be utilized to illustrate the efficacy of the approaches. The four receive channels shall be utilized in different ways to help demonstrate various features of the approach. A successful research program will lead to an effective and efficient means for signal classification, enumeration of RF signals in the environment, and geolocation to support a wide range of DoD applications, including EW operations.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 05, 2021
- Source ID
- N000142112464
Entities
People
- Thomas G. Pratt
Organizations
- Office of Naval Research
- United States Navy
- University of Notre Dame