Fast detection of spectral white spaces for cognitive radio networks

In military tactical communications, cognitive radio (CR) has been proposed as a potential technology. A CR network can support vital communications among troops and vehicles in an alien territory with unfamiliar or unknown radio environment that may be changing dynamically in time. We propose a new approach for dynamic detection of ¿white spaces¿ i.e. available radio channels, based on a novel concept of spectral sampling and variance estimation. We use FFT to obtain a set of samples of the short-term signal spectrum in a given band of frequencies. By analyzing spectral and statistical features of this set of samples, we are able to make fast and accurate determination of signal presence in a band.