Optimization of Spectrum Sensing Schemes in Cognitive Sensor Networks

The currently dominant spectrum allocation policy is reported to be inefficient. Cognitive radio, therefore, has been proposed in the literature to improve the spectrum usage efficiency. This dissertation proposes the optimization of spectrum sensing schemes in cognitive sensor networks. The modeling of the spectrum occupancy is a prerequisite for cognitive radio analysis. We describe the radio spectrum occupancy as a continuous- time Markov chain, and mathematically define the model by deriving the transition rate matrix and the probability state vector. The dissertation addresses an important aspect of spectrum sensing that has been often overlooked in the literature. While the cognitive radio is supposed to be aware of its surroundings, existing work does not consider the characteristics of unlicensed users for finding the optimum sensing period. In this work, we propose an application- specific method that finds the optimal sensing period according to the characteristics of both secondary and primary networks. According to the unlicensed user’s state in the Markov chain, two optimization problems are formulated to derive the optimum sensing periods. The secondary network’s throughput and power consumption are also studied and the corresponding parameters are derived. By numerical and simulation analyses, it is elaborated that the proposed method increases the secondary network’s throughput by up to 11% and significantly decreases the power consumption of the secondary network by as low as 33% of the non-hybrid approach. In addition, we study cooperative spectrum sensing in cognitive sensor networks and address two important issues. First, an optimization problem is solved to obtain the minimum required number of cognitive users. Second, we define a metric for sensing ac- curacy and propose a novel energy-aware secondary user selection method that identifies the most eligible cognitive users through a probability-based approach. The network’s lifetime is compared at several sensing accuracy thresholds and the trade-off between sensing accuracy and network lifetime is studied. Finally, the effects of several fusion rules on the proposed method are studied through simulation and numerical analyses. It is discussed that the Majority rule has the best performance among the examined rules. i

Optimization of Spectrum Sensing Schemes in Cognitive Sensor Networks

The currently dominant spectrum allocation policy is reported to be inefficient. Cognitive radio, therefore, has been proposed in the literature to improve the spectrum usage efficiency. This dissertation proposes the optimization of spectrum sensing schemes in cognitive sensor networks. The modeling of the spectrum occupancy is a prerequisite for cognitive radio analysis. We describe the radio spectrum occupancy as a continuous- time Markov chain, and mathematically define the model by deriving the transition rate matrix and the probability state vector. The dissertation addresses an important aspect of spectrum sensing that has been often overlooked in the literature. While the cognitive radio is supposed to be aware of its surroundings, existing work does not consider the characteristics of unlicensed users for finding the optimum sensing period. In this work, we propose an application- specific method that finds the optimal sensing period according to the characteristics of both secondary and primary networks. According to the unlicensed user’s state in the Markov chain, two optimization problems are formulated to derive the optimum sensing periods. The secondary network’s throughput and power consumption are also studied and the corresponding parameters are derived. By numerical and simulation analyses, it is elaborated that the proposed method increases the secondary network’s throughput by up to 11% and significantly decreases the power consumption of the secondary network by as low as 33% of the non-hybrid approach. In addition, we study cooperative spectrum sensing in cognitive sensor networks and address two important issues. First, an optimization problem is solved to obtain the minimum required number of cognitive users. Second, we define a metric for sensing ac- curacy and propose a novel energy-aware secondary user selection method that identifies the most eligible cognitive users through a probability-based approach. The network’s lifetime is compared at several sensing accuracy thresholds and the trade-off between sensing accuracy and network lifetime is studied. Finally, the effects of several fusion rules on the proposed method are studied through simulation and numerical analyses. It is discussed that the Majority rule has the best performance among the examined rules. i