RF imperfections compensation for wavelet-based multi-band spectrum sensing

Cognitive radio (CR) techniques are used widely to improve the utilization of spectrum resources by allocating spectrum holes dynamically. The spectrum sensing is one of the key CR techniques, especially for wideband cognitive radio. It has been the main emphasis of a lot of researches in CR communications. The wavelet transform detection method can divide the wideband spectrum, and then comparing the estimated energy of the each band with the threshold to determine whether the presence of a primary users (PU) signal or not. However, for the direct conversion receivers, because of its structure and hardware defects, there are a variety of RF front-end damages, especially in-phase/quadrature (IQ) imbalance. These problems not only result in the mirror-frequency cross-talk but also cause the errors to divide the wideband spectrum, finally considerably degrading the sensing performance. Thus in this paper a simple fast and effective solution was proposed, the Fourier series coefficients based compensation (FC) method. The position of the residual image bands can be locked by the Fourier series coefficients so that it can completely eliminate the residual image bands, and make the wideband detection getting a better result. Extensive computer simulations with two type of signals and various imbalance conditions are provided. It demonstrated that the proposed algorithm can eliminate the image channel cross-talk efficiently, yielding detection and false alarm probabilities essentially identical to those of an ideal RF receiver.