A Spectrum Sensing Technique for Cognitive Radios in the Presence of Harmonic Images

Harmonic downmixing is an important effect that must be taken into account when performing sensitive spectrum sensing using direct-conversion receivers. When the local oscillator waveform contains harmonics of the fundamental frequency, the quadrature mixer in the receiver will downconvert RF signals found at these harmonics, termed harmonic images, in addition to the RF signals around the fundamental frequency. The harmonic images will be detected by power spectral density based sensing algorithms and will cause certain parts of the desired spectrum to be mistakenly flagged as occupied. Although harmonic downmixing is important to consider, it is an often negelected effect. This paper presents a harmonic rejection spectral sensing technique, that exploits two quadrature mixers. The mixers work with different LO frequencies, which decorrelate the harmonic images so that cross-correlation of their outputs renders an improved spectral estimate. In addition to rejecting the harmonic images, spurious signals entering the receiver through the analog baseband inputs was also be rejected. The frequency resolution of the spectral estimate is scalable and the rejection of the harmonic images increases 15 dB per 1000-fold increase of the correlation time. The complexity of the algorithm is analyzed and its performance is shown by means of simulations. The effect of I/Q imbalance is also taken into account.