Cooperative Spectrum Sensing with Imperfect Knowledge of Secondary Users’ Detection Capability

Abstract

Cooperative spectrum sensing is a key technology for cognitive radio networks (CRNs) where it can be used to identify idle spectrum holes. In a CRN, secondary users periodically monitor the spectrum to determine an idle band that can be used to transmit their information. Nevertheless, the fading and noisy environments limit the accuracy of individual secondary users decisions. Consequently, multiple secondary users may cooperate to improve the decision making process. Nevertheless, fusing the decisions of the individual secondary users reliably requires the knowledge of the detection and false alarm probabilities of individual users, Pd and Pfa, which have to be estimated at the cognitive radio base station (CRBS), and hence, may suffer from certain estimation errors. Therefore, the main aim of this work is to evaluate the impact of imperfect knowledge of Pd and Pfa on the fusion process. Towards this goal, extensive Monte Carlo simulation experiments are conducted where the estimation error is modeled as a truncated Gaussian random variable. In addition, adaptive detection threshold at the CRBS is considered to improve the system detection performance. The obtained results show that the global detection and false alarm probabilities may deteriorate significantly in the absence of accurate knowledge of Pd and Pfa. Moreover, using adaptive thresholds may be beneficial only when Pd and Pfa are known accurately. Otherwise, the performance may excessively deteriorated.