Spectrum Sensing Issues for Cognitive Radio

In this thesis, first, we address the problem of energy efficiency for spectrum sensing. We propose a combined censoring and ”on/off” scheme as an energy efficient technique for cognitive radio networks. We maximize the lifetime of the spectrum sensing system by minimizing the energy consumption of the system subject to a specific detection performance constraint. We show that our proposed technique is more energy efficient than the traditional pure censoring and pure ”on/off” schemes. Second, we address the problem of wide-band spectrum sensing at low SNR. Energy detector is known to be sensitive to noise uncertainty, hence it is not reliable at low SNR. Here, we propose a two-stage spectrum sensing technique consisting of a coarse and a fine sensing stage. In the coarse sensing stage, we locate the possible empty channels within a frequency band using an energy detector and then in the fine sensing stage, we make a final decision about the emptiness of the reported channels from the coarse sensing stage using a cyclostationary detector. Our simulation results show that under a certain noise uncertainty, two-stage spectrum sensing outperforms both the energy detector and the cyclostationary detector in terms of the detection performance and mean detection time. However, increasing the noise uncertainty leads to a decreasing SNR range where our proposed technique performs better than the energy and cyclostationary detector. Spectrum Sensing Issues for Cognitive Radio