Distributed Spectrum Sensing for Cognitive Vehicular Networks using Quasi-Newton Optimization

Distributed cooperative spectrum sensing is an alternative sensing mechanism in which all the secondary users (SU) have individual decisions regarding spectrum availability. For vehicular communications, the decision of spectrum availability is shared by a voting mechanism among the SUs through energy detection (ED). However, the accuracy of the ED algorithm often depends on considered fading channel and detection threshold. Therefore, in this paper, we analyze the sensing performance of distributed cooperative vehicular network over $\alpha-\eta-K-\mu$ fading model, which is more suitable for vehicular communication in comparison with existing fading models. We choose the single lane and double lane urban scenario with dense traffic conditions. We first obtain the statistical knowledge of the received signal-to-noise ratio under the SU mobility then we derive the closed-form expressions for missed detection probability for dense traffic conditions with an entropy-based voting rule. We also compute the probability of incorrect detection as a function of both the probability of missed detection and the probability of false alarm. To minimize the likelihood of incorrect detection. Simulations were conducted to validate the accuracy of derived analytical expressions. Numerical results show that by optimizing the detection threshold, the obtained results of detection probability outperform the results of the sphericity test. Moreover, the probability of incorrect detection over $\alpha-\eta-K-\mu$ fading approaches to zero.