Decision-Driven Time-Adaptive Spectrum Sensing in Cognitive Radio Networks

In cognitive radio systems based on periodic sensing and transmission, a secondary user (SU) senses the activity of a primary user (PU) in the sensing phase and then makes use of the detected spectrum opportunity in the transmission phase. Under the constraint that the PU should be sufficiently protected, it is challenging for the SU to accurately sense and utilize spectrum opportunities in low signal to noise ratio (SNR) environments. However, in existing spectrum sensing schemes, the time-frequency resource blocks available for data transmissions are always wasted once a false alarm event happens in the sensing period where the SU is doing nothing but waiting for another sensing opportunity. By reutilizing these initially wasted time-frequency resource blocks for spectrum sensing, the SU can get more accurate information on the activities of the PU, which enables the SU to make more efficient use of the spectrum opportunities left by the PU. In this paper, a decision-driven time-adaptive spectrum sensing scheme is proposed to improve both spectral efficiency and energy efficiency based on improved resource usage. Specifically, if the PU is detected to be absent in the sensing phase of a frame, the SU transmits data in the transmission phase of the frame; otherwise, the SU will sense the spectrum for a duration of one frame period, where the frame period consists of the transmission phase of the current frame and the sensing phase of the next frame. When both signal and noise are Gaussian, the optimal maximum likelihood ratio detector is derived. Furthermore, the performance upper-bounds of spectrum utilization and secondary throughput are obtained. Finally, both simulation and theoretical results show that the developed scheme can improve spectrum utilization, secondary throughput and energy efficiency effectively in extremely low SNR environments.