Novel Algorithms for Blind Sensing of Signals With Variable Bandwidths in Cognitive Radio

In a cognitive radio, an interface is designed for use by secondary users (SUs) on a noninterfering basis in spectrum allocated to the primary user (PU). However, to prevent interference to the PU, the SU cognitive radio needs to detect the presence of the PU at very low signal-to-noise ratio (SNR), due to multipath and shadowing on its way to the SU. Wideband spectrum sensing is considered here. The main contribution of this article is a blind algorithm for blindly sensing PU signals, of varying bandwidths, at very low SNRs. A novel oblique projection (involving the oversampled received signal) is used. The effect of PU signal bandwidth on the performance of the blind spectrum sensing algorithms is investigated in this article, which has not been done earlier. Unlike existing blind spectrum sensing algorithms, whose performance depends on the bandwidth of the PU signal, the oblique projection-based novel method adapts itself to the varying bandwidths of the PU signal to maintain the constant level of superior performance for all bandwidths of PU signals; this is solely due to the oblique projection sensing method, which also exploits the correlation between the PU signals on the different receiver oversampled branches.