Skip to Main Content
In cognitive radios, spectrum sensing is employed as a means to increase spectrum usage awareness. Spectrum sensing algorithms must satisfy the desired performance requirements, in terms of probability of detection and false alarm, with short sensing duration. In addition, low-complexity algorithms are sought, particularly in low-power cognitive radios. A mechanism to search available channels in the frequency domain is proposed in this paper. Rather than obtaining an accurate estimation of the spectrum within a certain frequency range, we focus here on the task of finding white spaces. The proposed technique is a steepest descent method, adapted to the problem of sensing white spaces. Based on an approximation on the right and left derivatives of a smoothed periodogram, the algorithm iteratively searches for a free channel by following the direction of steepest descent. The adaptation step is such that the algorithm produces large steps in flat spectrum regions and small steps in steep regions, in order not to miss narrow portions of the spectrum with free channels. Simulations are provided for the proposed algorithm in a scenario where IEEE 802.11g and IEEE 802.15.4 devices coexist. The performance and speed of convergence of the proposed algorithm are shown, providing a comparison with a system that performs random channel search.