Optimizing Spectrum Sensing using Average Slope Detection and Machine Learning Techniques

---

Abstract

The cognitive radio network represents a pivotal advancement for 5G applications, countering the limitations posed by spectrum scarcity. Spectrum sensing is vital for identifying vacant spectrum bands in a network framework that comprises of Primary User (PU) and Secondary Users (SU’s). The traditional spectrum sensing scheme like Energy detection is highly sensitive to uncertainties in noise with limited sensing accuracy. To overcome this limitation a novel method, Average Slope Detection (ASD) with Cooperative Spectrum Sensing is proposed. The Cooperative Spectrum Sensing network is simulated in MATLAB The classification of noise and PU is done be integrating Machine Learning algorithm, K-Nearest Neighbor which shows an improved accuracy by 18.34% over existing methods for a SNR -20dB.

---

Keywords

Cognitive Radio (CR); Average Slope Detection (ASD); Spectrum scarcity; Cooperative Spectrum Sensing (CSS); MATLAB; K-Nearest Neighbor