Robust spectrum sensing in cognitive radio

Spectrum sensing is a critical step in cognitive radio to learn the radio environment. Despite its long history, in recent years spectrum sensing has attracted substantial interests from the cognitive radio community as well as other areas. Although there have been many methods, most of them need some idealistic assumptions and are hardly applicable in real cognitive radio practice. In this paper, we first discuss the possible hostile environment facing the spectrum sensing in cognitive radio. We then investigate some methods which could survive under such hostile environment. These methods include eigenvalue/covariance based detections, cooperative sensing, and cyclostationary detections. The robustness of these methods are discussed in detail through theoretic analysis and simulations.