Modelamento esparso e técnicas de decomposição material aplicados à detecção do espectro.

Abstract

Spectrum sensing is the first step for enabling opportunistic frequency utilization in Cognitive Radios (CR). Most of the previous proposed techniques for spectrum sensing usually perform poorly in low SNR scenarios, and those having good performance normally have high complexity. In this paper, it is presented a method based on sparse modeling for spectrum sensing and free channel identification for the open digital television system spectrum in Brazil, considering the effects of additive white Gaussian noise and signal attenuation. The detection is enhanced through the use of matrix decomposition techniques and the results show that the proposed method can identify channel occupancy with high precision, even in critical SNR scenarios.