Performance of Spectrum Sensing Schemes in Cognitive Radio for Static and Dynamic Primary Users in Additive Laplacian Noise

Abstract

Due to the evolution in wireless communications, the RF spectrum is overcrowded as it is scarce and expensive. However, it has been observed that the licensed spectrum remains underutilized. In the literature, various spectrum sensing schemes have been proposed for efficient spectrum utilization. In this case, the cognitive radio (CR) terminal senses the licensed spectrum of the primary user (PU). If the spectrum is found vacant, then the unlicensed user or secondary user (SU) can utilize the spectrum without interfering with the PU. The performance of the spectrum sensing techniques is presented and evaluated using receiver operating characteristics (ROC). The ROC for different spectrum sensing techniques is presented and compared with the literature assuming different scenarios such as coherent and non-coherent detections, different channel environments, PU with different waveforms, and various diversity schemes. A majority of the papers have assumed additive white Gaussian noise (AWGN) and static behavior of PU in the sensing interval. However, in a real-time scenario, these assumptions are difficult to follow. For example, in the multi-user environment, the interference from different sources such as multiple access interference (MAI) can be well approximated by additive Laplacian noise instead of AWGN. Further, due to the large density of PU and their frequent transitions, the static nature of PU in the sensing interval may not be followed. The dynamic behavior of the PU degrades the ROC performance of detection schemes, which assumed static behavior. In this thesis, we assume additive Laplacian noise (ALN) channel instead of AWGN channel. We assume static PU and use a modified correlation detector (MCD) with cooperative spectrum sensing (CSS) scheme. We compare our proposed correlation detection scheme with the other state-of-the art of the correlation detection schemes. Dynamic behavior of the PU, where PU may randomly change its states within sensing period, has not been analyzed with ALN in the literature. We assume the dynamic behavior of PU, where the Poisson process models the rate of arrival and departure of the PU in the prescribed spectrum sensing interval. In this case, we use a non-coherent scheme using improved absolute value cumulation detection (i-AVCD). Subsequently, we assume more than one transition of PU in the sensing interval. The two transitions of PU have been modeled using weighted samples, which are further based on Cumulative Sum (CuSum) for detection. In this case, we use the sample mean detection (SMD) and i-AVCD as detection schemes. The multiple transitions of PU are modeled using the Markov chain. Further, we assume PU with different modulation order M such as M-QAM and show the effect of M on the ROC performance. Finally, we consider attenuation in the i channel by Rayleigh distribution with diversity schemes such as Equal Gain Combining (EGC) and Soft limiting Polarity Coincidence Array (SL-PCA). In all the cases, we present ROC using simulations. We also derive analytical expressions of detection probability (PD) and false alarm probability (PF ). A close match between simulations and their analytical counterparts validates our analytical approach. Further, a significant improvement in performance of the detection schemes is achieved using the CSS and receive diversity schemes.