The use of time frequency analysis for spectrum sensing in cognitive radios

Spectrum sensing is being widely researched as the key tool used by the cognitive radios to gain awareness of their surrounding radio environment. Non-parametric spectrum sensing techniques such as energy detection and cyclostationary feature detection are preferred over the parametric techniques because any practical visualisation of cognitive radios would necessitate minimum assistance by the primary user for its detection. These non-parametric spectrum sensing techniques have been thoroughly analysed for their performance and generally offer a compromise between accuracy and the sensing time taken to detect the presence of the primary user. The time windowing used in these techniques results in smearing of spectrum in the frequency domain. Hence a large time window (sensing time) is required to obtain accurate results in the frequency domain. This paper suggests a novel non-parametric spectrum sensing approach utilising time frequency analysis. Use of time frequency analysis results in simultaneous improvement in temporal and spectral resolution thus giving better results using smaller time windows. The approach is demonstrated using Wigner - Ville time frequency distribution. The approach is rigorously investigated under varying conditions of prevailing noise and for different primary users (Frequency Hopped Spread Spectrum and Orthogonal Frequency Divisional Multiplexing). The performance is compared to other non-parametric techniques (energy detection and cyclostationary feature detection) using the commonly used comparison metrics such as probability of false detection, probability of loss of detection and receiver operating curves using different sensing times. The comparison shows that time frequency analysis based spectrum sensing gives much improved results over the other well-researched approaches in terms of reduced probabilities of false detection and missed detections and smaller sensing time required to sense the presence of a primary user.