Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{208026,
        author = {Ms. Suvarna Sahebrao Patil and Dr. Shivleela Mudda},
        title = {Cognitive Radar for Predictive UAV Surveillance: A Review of Sensing, Tracking, Learning, Re-Acquisition, and Threat Assessment},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {4},
        pages = {83-98},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=208026},
        abstract = {Small unmanned aerial vehicles (UAVs) are difficult radar targets because low radar cross section, low-altitude clutter, multipath, rapid maneuvering, and intermittent visibility jointly degrade detection and track continuity. Cognitive radar is attractive in this setting because it closes the perception-action loop: the radar does not only estimate a target state, but can adapt sensing resources according to uncertainty, predicted motion, and mission risk. This paper presents a PRISMA-informed systematic review and critical synthesis of cognitive radar methods for predictive UAV surveillance through August 2026. The evidence is organized across six coupled functions: FMCW/mmWave sensing and micro-Doppler perception; uncertainty-aware single- and multi-target tracking; learning-based classification and trajectory prediction; adaptive radar resource management; autonomous target re-acquisition; and interpretable threat assessment. Recent 4-D radar studies demonstrate the value of joint detection-tracking and transformer-based trajectory modeling, while contemporary deep-reinforcement-learning work demonstrates increasingly capable time and task allocation. However, the literature remains fragmented: high classification accuracy is often reported on private or sensor-specific datasets, learned prediction is not consistently calibrated against tracking uncertainty, recovery after target loss is rarely treated as a primary metric, and threat scores lack a shared ontology. The review therefore proposes an integrated reference architecture and a benchmark framework spanning detection, tracking, recovery, risk, and embedded latency. The main conclusion is that publishable progress now depends less on adding isolated deep models and more on reproducible end-to-end evaluation under realistic clutter, occlusion, multi-target ambiguity, computational constraints, and explicit uncertainty.},
        keywords = {cognitive radar, UAV surveillance, FMCW radar, micro-Doppler, multi-target tracking, trajectory prediction, reinforcement learning, target re-acquisition, threat assessment, 4-D radar.},
        month = {September},
        }