Abstract
In this thesis, the author proposes an optimal detector for the Thomson’s multitaper spectral estimation (MTSE) [1] in cognitive radios for spectrum sensing. The MTSE is used for spectral estimation and then the proposed detector can detect primary users and spectrum holes for secondary users in cognitive radios. This proposed detector is designed and optimized based on the Neyman-Pearson Theorem. The analytical detection performance (probability of detection) of detecting primary users (PUs) and spectrum holes in spectrum sensing is mathematically formulated, studied, and verified by simulations. The proposed detector for the MTSE is non-cooperative in spectrum sensing and can perform blind spectrum sensing without prior knowledge of the primary users. Compared to the conventional energy detector [2], the proposed detector is more reliable and robust for various multiple Slepian tapers. The proposed detector outperforms the cooperative autocorrelation-based detectors [3] in detection rate by 48%and with 3dB gain in SNR performance. Moreover, the minimumrequired observation size is smaller than the conventional energy detector by a reduction of 73.3%.