Maximum-minimum energy based spectrum sensing under frequency selectivity for Cognitive Radios

Cognitive Radio (CR) technology which uses intelligent signal processing at the physical layer of a wireless system has been considered for coordinating better the spectral resources. In this study, we investigate spectrum sensing methods which utilize the frequency variability of the energy spectral density which is introduced by the primary transmissions. The variability is partly due to the transmitted spectrum shape and party introduced by the frequency selective multipath channel. The spectral variability is observed by dividing the sensing frequency band into relatively narrow subbands and comparing the subband energies. We compare the subband energy based methods against the eigenvalue based sensing method, which exploits the signal correlations introduced by the primary transmission. Eigenvalue based sensing is known to be robust against the inevitable noise uncertainty, which severely limits the usability of energy detection for primaries with low SNR. Our results demonstrate for the subband energy detection methods, similar robustness against noise uncertainty as the eigenvalue based methods have.