Abstract Cognitive radio networks (CRN) enable secondary users (SUs) to opportunistically access underutilized licensed spectrum while protecting primary users (PUs) from interference. However, robust spectrum sensing under heterogeneous interference and noise uncertainty remains challenging. Conventional energy detection, matched filtering, and cyclostationary approaches either require extensive prior knowledge or degrade under interference, leading to excessive false alarms and service denial. We propose an adaptive frequency-domain constant false alarm rate (CFAR) spectrum sensing approach that dynamically sets frequency-specific thresholds based on neighboring spectral samples. This enhancement requires minimal prior knowledge. To address potential security risks from untrusted SUs exploiting this resilience, we introduce a centralized counter-access mechanism: an administrator-controlled jammer that defeats CFAR-equipped adversaries, preventing unauthorized access during emergencies. It safeguards primary communications by creating strategic frequency gaps that avoid known jammed PU channels. Using APCO Project 25 waveforms for PUs, Monte Carlo simulations evaluate cell-averaging (CA), greatest-of (GO), smallest-of (SO), order-statistics (OS), and censored CFAR variants across different channel models combined with multiple jamming scenarios. Results show OS and censored CFAR variants best stabilize false alarm rates and enhance detection in heterogeneous conditions, outperforming CA and GO/SO schemes. The comb-sweep jammer effectively denies service to CFAR-equipped SUs without impacting PUs, though at the expense of reduced spectrum availability. Sensitivity analysis confirms robustness across varied CFAR window configurations, precluding parameter-based evasion by unauthorized SUs. This work frames spectrum sensing and jamming in CRNs as a two-sided interaction between electronic protection for secondary users and administrative electronic attack for controlled denial.
Shams et al. (Mon,) studied this question.