摘要：In this paper, the performance of spectrum sensing is analysed while taking into account the effect of Primary User (PU) activities. A PU work period is defined and formalized to practically realize the PU activities when dynamically changing between ON and OFF transmission scenarios. In reality, it is important to consider such work period as the PU will only active for certain fraction of the total frame time. A new sensing model, namely, Constant False Alarm Rate-Dynamic Energy Detection (CFAR-DED) is introduced for dynamic-PU signal state scenario. The CRAF-DED model is then further categorized into three scenarios; Dynamic Threshold (DT), Two-Stage (TS), and Adaptive Two-Stage (ATS) detection algorithms. Closed-form expressions for the average detection probability have been mathematically derived when the PU is partially present within the observed period under AWGN and Rayleigh fading environments. Simulation results show that the proposed algorithms provide detection improvement as compared with conventional energy detection. In particular, ATS represents the most effective sensing algorithm of the CFAR-DED model. In addition, the results show that the probability of detection degrades severely when the PU work period and sensing time are reduced regardless of the method of detection used. Furthermore, it has been found that the detection performance of the proposed algorithms under Rayleigh fading with low-SNR is significantly deteriorated.
