AI-Enhanced Signal Processing for Extended-Range Radar Detection in Counter-Fire Systems

This Phase?I SBIR effort aims to enhance the long-range detection capability of ground-based counterfire radars, such as the AN/TPQ-53, through advanced software-only signal processing techniques. Current radar systems experience a significant drop in probability of detection (Pd) beyond ~12 km, particularly for low-trajectory threats, limiting warning time and reducing counterfire effectiveness. Traditional solutions—such as increasing dwell time or narrowing search sectors—can improve signal-to-noise ratio (SNR) but at the cost of slower scan times and reduced area coverage. Our approach integrates three advanced signal processing methods to improve detection range while maintaining current radar scan rates: (1) Adaptive Clutter Filtering, which dynamically suppresses ground clutter, wind farm returns, and environmental interference to maintain a low false alarm rate (Pfa < 10??); (2) Multi-Frame Coherent Integration, which accumulates weak target returns over multiple radar frames to boost SNR and compensate for projectile motion and Doppler shifts; and (3) CNN-Based Detection, employing a lightweight neural network (DeepRadarNet) optimized for range-Doppler discrimination, shown in prior research to improve detection sensitivity by 0.5–1 dB SNR over conventional CFAR methods. The primary objective of Phase I is to demonstrate a ?15% Pd improvement at 30 km while keeping Pfa < 10??, all without modifying radar hardware. This will be accomplished through a feasibility study that evaluates at least one of these techniques in a MATLAB-based simulation environment, comparing performance against legacy CFAR and track-before-detect (TBD) through rigorous benchmarking. The study will define baseline performance, integrate and optimize processing techniques, and quantify improvements through standard metrics such as Pd vs. SNR curves, false alarm rates, and computational overhead. Phase I deliverables include a test plan, simulation results, and a Phase I Feasibility Report documenting performance improvements, risk mitigation strategies, and readiness for Phase II. The proposed methods will be assessed for integration into a System Integration Lab (SIL) during Phase II, with a transition roadmap supporting full-scale adoption in operational Army radar systems. By leveraging our team’s expertise—spanning CNN-based ATR development at Raytheon, SIGINT sensor fusion, and GPU-accelerated real-time DSP—we aim to extend Q-53 detection range and enhance counterfire mission effectiveness with a low-risk, high-impact software upgrade.