Cognitive Radio: Spectrum Sensing under Unknown Signal and Noise Distributions

In Cognitive Radio networks, a proper sensing of vacant frequency band or spectrum hole is necessary for an unlicensed user in order to avoid interference with the licensees. Noise distribution plays a significant role in spectrum hole identification process. Noise is often assumed to be AWGN but as it appears from various sources, the distribution often deviates from its AWGN nature. If the deviation is significant, the total detection process becomes unreliable. In this paper, an attempt has been made to detect the spectrum holes considering a continuous distribution as noise distribution instead of any specific parametric distribution. This approach is known as non-parametric approach. The entire detection process is judged with two figures of merit, namely, probability of false alarm and probability of detection. Finally, some designing criteria are indicated based on these two parameters.