Improved autocorrelation-based sensing using correlation distribution information

Abstract

Accurate and efficient spectrum sensing is a critical component of cognitive radio, which is a technology poised to improve dynamic resource management in future wireless networks. Autocorrelation exploitation has been shown to provide improvements in sensing performance over simple methods like energy detection. A new upper bound on the performance of autocorrelation-based detectors based on an NP test under the assumption of correlation distribution information (CDI) is presented, where random parameters of the signal autocorrelation are not known, but their distribution is assumed to be known. Not only does this reveal how close existing ad-hoc autocorrelation-based detectors are to optimal performance, but also it indicates how much performance improvement over energy detection is theoretically possible.