Spectrum Sensing Using Cyclostationary Spectrum Density for Cognitive Radios

Abstract

Cognitive radios (CR) are proposed to alleviate the spectrum scarcity problems facing wireless service providers. In US, the FCC is considering spectrum regulation changes by allowing unlicensed operation in the TV broadcast bands provided that no harmful interference is caused. In this paper, we discuss the spectrum sensing aspects of cognitive radios. We particularly focus on the detection method based on cyclostationary spectrum density (CSD) estimation. The advantage of CSD is its relative robustness against noise uncertainty compared with energy detection methods. CSD estimation is a two dimensional transformation; therefore it is computationally complex. We transform the algorithm from the two dimensional space to a one dimensional case, therefore making the real time implementation more feasible. Through analysis and simulation, we identify the features with highest SNR to be used for CSD based detection. Based on the simulation results, we further propose dedicated hardware implementation architecture for CSD estimation using field programmable logic array (FPGA). Our implementation can achieve greater than 90% detection probability on BPSK signals with SNR of -18 dB, when the probability of false alarm is less than 10%.