CARVALHO, F. B. S.; CARVALHO, Fabrício Braga Soares de.
Résumé:
The cognitive radio concept evolves as a response to the scarcity of available frequency
bands for telecommunication services. Cognitive radio enables the monitoring of
the electromagnetic spectrum to identify spectrum holes; in other words, to verify i f a specific
spectrum band is temporarily available to be dynamically occupied by another user.
The improvement of the spectrum sensing, to enable a better sharing of the frequency spectrum
among different users, is a major challenge in wireless communications. Several new
techniques and algorithms are being proposed to deal with the spectrum occupancy in cognitive
channels. Nevertheless, some issues, such as fading and multipath are not properly
investigated. Fading and multipath effects are simplified when spectrum sensing, under such
conditions, is analyzed. In this thesis, the non-cooperative spectrum sensing problem (in
which a single detector performs the sensing) is evaluated. The contributions include the
analysis of fading effects on the spectrum sensing (based on energy detection and on statistical
tests) and the proposal of new spectrum sensing algorithms that consider fading effects
in cognitive channels. The first contribution is an evaluation of the performance of spectrum
sensing via energy detection in fading channels. The second contribution is based on statistic
tests to optimize the cognitive spectrum sensing. Skewness, Kurtosis and Jarque-Bera
statistical tests are considered when the cognitive channel is influenced by fading models as
Rayleigh, Nakagami-m, Rice and Lognormal. The third contribution refers to the proposal
of new sensing algorithms, based on Skewness, Kurtosis and Jarque-Bera statistical tests,
when the cognitive channel is subject to Rayleigh fading. The proposed algorithms lead to
superior performance when compared to the spectral detection based on energy detection or
on statistical tests.
