Stochastic Resonance-Aided Energy Detection for RF-Powered Cognitive Radio Networks

Abstract

Conventional stochastic resonance (SR) techniques often face challenges with higher-frequency signals and parameter optimization for real-time applications, as observed in practical orthogonal frequency-division multiplexing (OFDM) systems that are vulnerable to noise uncertainty (NU). In this study, we present a novel SR-aided energy detection (ED) method that incorporates multi-taper spectrum estimation technique to improve spectrum estimation precision and Gauss-Seidel-like iteration method to accurately adjust the SR parameters for real-time adaptation. This combined strategy enhances weak signal detection, prevents signal distortion, and increases robustness against fluctuating noise conditions. Results from 5,000 Monte Carlo simulations showed that, at 0 dB NU, SR-aided ED attained 90% detection probability at -11 dB, outperforming conventional ED with an SNR gain of 12.5 dB. At 3 dB NU, the conventional ED accuracy degraded by 5.5 dB, resulting in a false alarm probability of 77%, while SR-aided ED demonstrated robustness to NU. At 10 dB NU, ED failed to distinguish the differences between noise and signal power, giving rise to 99% false alarm probability. In contrast, despite a 6 dB degradation, the developed SR-aided ED approach still guarantees a 1% false alarm probability. In clipping-prone systems, conventional ED is vulnerable to signal clipping. Conversely, SR-aided ED remains unaffected.