Fast and reliable detection of incumbent users in cognitive radios

Abstract

Fast and reliable Spectrum Sensing (SS) plays a crucial role in the cognitive radio (CR) technology in order to prevent unwanted interference to the primary users (PU) and to reliably and quickly detect the white spaces in the spectrum for opportunistic access by the secondary users (SU). Spectrum Sensing must often be performed in the absence of information such as PU signaling scheme, noise level and channel fading coefficients. While these parameters can be estimated in the SU, estimation errors significantly deteriorates the performance of SS techniques. In this thesis, we introduce and evaluate the performance of two novel blind spectrum sensing algorithms which do not rely on knowledge of these parameters. The first is a SS technique for signaling schemes which introduce controlled intersymbol interference in the transmitter. The second is for cases when the receiver of the SU is equipped with a multiantenna system. This approach exploits the path correlation among the signals received at different antennas. Next we analyze the performance of Spectrum Monitoring (SM), an new technique which allows the SU to detect the presence of the PU using its own receiver statistics. In contrast to SS, with SM, the SU does not need to interrupt its own transmission in order to detect the presence of the PU. We carefully construct the decision statistics for SM and evaluate its performance. The performance of a hybrid SM/SS system shows a significant improvement over SS alone.