Sensoriamento espectral cooperativo baseado em rádios cognitivos com múltiplas Antenas

Abstract

Multiantennas cognitive radios (CRs) have been used in spectral sensing to improve detection performance, but the majority of the works do not consider correlation, due to antenna proximity, in the received signals and antenna calibration errors. Therefore, this dissertation presents both performance analyzes using two blind statistical tests, the well-known GLRT (generalized likelihood ratio test) and the CFCPSC (Circular Folding Cooperative Power Spectrum Density Split Cancellation), in CCSS (centra lized cooperative spectrum sensing) with HD (hard decision) and SD (soft decision) assuming correlated received signals due to the proximity of the antennas and dynamic noise. It is also adopted frequency-selective fading according to the WRAN (wireless regional area network) standard IEEE 802.22 to give a more practical appeal to the analyses at hand. The scenarios in CCSS adopt different numbers of secondary users (SUs) and antennas per SU. At first, the performance analysis is carried out observing only the effect of the correlation of the received signal, considering the same noise power in each antenna of each SU. Then, in order to simulate a more complete scena rio, besides the correlation, different noise power in each antenna of each SU is also applied. The results showed great advantages in favor of CFCPSC.