Signal detection for cognitive radio using multiple antennas

Abstract

Cognitive radio (CR) is currently being seen as the future for wireless communications. Fundamental to the operation of a CR is the detection of presence of legacy communication. Energy detector (ED) is a simple and effective signal sensing method. However, the ED requires accurate estimation of the noise power. Any error in the estimation, which is common in practice, affects its performance. In this study we have looked at the detection problem using an antenna array. Under noise only case, the array covariance matrix has reasonably equal eigenvalues. On the other hand when a signal is present, the strongest eigenvalue is dissimilar to the rest. To measure the equality of the eigenvalues we use the sphericity test. We have found that under realistic assumptions about noise uncertainty and signal correlation the array based sphericity method of signal detection can perform better than the ED. Also, unlike the ED the sphericity test does not require any estimation of the noise power.