Hybrid frequency-domain CFAR detectors for cognitive radio interference resilience and the dual-use security paradox

Abstract

Abstract Cognitive radio networks (CRNs) enable secondary users (SUs) to opportunistically access underutilized licensed spectrum while protecting primary users (PUs). Robust spectrum sensing under heterogeneous interference and heavy jamming remains challenging: conventional approaches either require extensive prior knowledge or degrade significantly under interference, while basic constant false-alarm rate (CFAR) variants remain vulnerable to structured jamming that contaminates only part of the reference window, causing fixed-rule estimators to either include contaminated samples or discard clean ones. Hybrid CFAR architectures address this by first analyzing reference-window contamination patterns and then adaptively selecting or combining estimation strategies. We adapt eleven such hybrid architectures—originally developed in the radar literature—to frequency-domain spectrum sensing in contested CRNs, and evaluate them across four families (adaptive algorithm selection, order-statistic fusion, intelligent censoring, and weighted processing) against classical CR-CFAR baselines via Monte Carlo simulations using APCO Project 25 as the PU waveform and an OFDMA-based SU, under barrage and swept-FM jamming. Results show that First-Order Difference CFAR achieves the best detection performance through global sorting and statistical jump detection ( $$P_{\text{d}} \approx 1$$ at SNR  $$= 3$$  dB under 50% sweep coverage), followed by Smallest-Of and Order-Statistic Smallest-Of variants. However, this resilience creates a dual-use security paradox: unauthorized devices employing these techniques can resist administrative spectrum control. We address this through an algorithm-aware comb-sweep countermeasure that exploits CFAR reference-window dependencies, inducing near-unity false alarms on vacant channels within a 4 dB JSR denial plateau ( $$-5$$ to $$-1$$  dB) while preserving PU detection, demonstrating that algorithm-aware enforcement can restore spectrum governance in contested environments.