Performance of Eigenvalue Based Spectrum Sensing In Full-Duplex Cognitive Radio Networks

Spectrum sensing is a key component in any cognitive radio network. Recently full-duplex communication, i.e., the ability to transmit and receive at the same time at the same frequency, has become feasible. Residual self interference is inevitable even after applying self interference cancellation techniques in radio frequency (RF) and baseband domains. In this paper, we study the performance of popular eigenvalue based spectrum sensing techniques under residual self interference. Moreover, we investigate their performance when exploiting the correlation coefficient matrix, rather than the covariance matrix, to estimate the decision statistics. Finally, we propose three new correlation coefficient matrix based algorithms that outperform existing techniques.