Universal Signal Matching for RF Threat Classification Navy SBIR FY2009.2 Sol No.: Navy SBIR FY2009.2 Topic No.: N092-113 Topic Title: Universal Signal Matching for RF Threat Classification Proposal No.: N092-113-0190 Firm: Research Associates of Syracuse 111 Dart Circle Rome, New York 13441 Contact: Brian Bush Phone: (315) 339-4800 Web Site: www.ras.com Abstract: This effort investigates and assesses the feasibility of new robust dynamic methods to classify threats from received RF signals for application across a variety of sensors and platforms using new information that can now be obtained from modern digital EW receivers. Mathematical and statistically based techniques including covariance functions, autocorrelation and kurtosis to automatically characterize additional emitter characteristics proposed will be justified. Classification includes classic parameters (RF, PW, PRI) and new automatic statistical processes for scan, PRI and RF Agile typing and characterization. New descriptors for Waveform Function (e.g. Track, Search) and Type (e.g. Pulse Doppler, FMCW) will be developed to automatically assess waveform intent for improved situation awareness, support EA, and improve ID. Intentional Modulation on Pulse (IMOP) Type and Characteristics are incorporated into emitter track / correlation using FPGA based IMOP results from a prior Phase II/III SBIR. The processes will be integrated into existing multi-hypothesis Bayesian belief network enabled tracking, classification and identification MATLAB processes (C/C++ for real-time). Established metrics of effectiveness are used to characterize performance. Proof of concept MATLAB code will be demonstrated with signals from RAS' suite of synthetic signals, real-world modern stressing radar digitized data (unclassified), and PRI and Scan Pattern generation tools. Benefits: Key findings developed during this effort are applicable to any ES, EA or ELINT signal application where signals must be detected, classified, indentified and /or recorded. They can be applied in a multitude of other DoD efforts such as COMINT or MASINT systems. Key benefits of the proposed approach include: 1) Provides capability for automatic description of threat waveforms based in observed parameters 2) Utilization of statistically based automatic processes accounting for the measurement uncertainty known to exist, especially at low SNR; results in robust quantifiable performance at low SNR and with small numbers of pulses 3) Incorporates "all" that is known or can be measured about the emitter to obtain the complete classification of the waveform / emitter for more accurate identification 4) Leverages and expands upon significant research and development RAS performed for AFRL and NRL 5) Utilizes results from the NAVEA ES-PFEP Phase II SBIR (and III for NAVAIR PMA-265) to incorporate complex IMOP characteristics 6) Proven, field tested multi-hypothesis Bayesian belief net with real-time capability 7) Modular processes that will able to be easily modified or enhanced 8) Well defined established metrics are used to determine performance and can be translated into Key Performance Parameters (KPP) as might be used for operational system requirements The processing enhancements proposed herein have several benefits and numerous military and commercial applications. Potential DoD applications include EA, ES systems such as the ALR-67(v3), the ALQ-218, and the to be developed Next Generation Jammer (NGS) for Super Hornet, Hornet, and Growler platforms. Other relevant candidate NAVY applications for insertion of the technology developed on this SBIR are the PMW-180 Ship's Signals Exploitation Equipment (SSEE),and the NAVSEA Surface EW Improvement Program Block II (SEWIP) now undergoing contractor evaluation and down-selection. This program would benefit from some of the concepts proposed herein. Potential applications in the private sector could include wireless waveform characterization and communication waveform fidelity collection, classification and library generation for mapping spectral interference for adaptive spectrum allocation. Automatic statistical based processes could be valuable in monitoring and recording key medical phenomena such as EKG or brain wave activity. Other signal classification applications include speech recognition, image processing and wideband or LPI signal characterization. Certain Software modules such as RF Agile or PRI typing (aka waveform sequence) may have some application in RF test equipment and automated test procedures. These will be explored in more detail in Phase I and Phase II. Return