ABSTRACT
A novel attack-resilient and efficient QoS-based GPCR-ARE protocol approach is used for a secure VANET network. The intrusion detection system (IDS) with deep neural network (DNN) is planned for the improvement of vehicular security. The DNN system is trained with probability according to feature vectors, which are extracted from VANET system information and provide the probability of each class perceptive to normal and attack packets, and thus the sensor can recognize any attack vehicle. As a comparison to the traditional system of artificial neural network (ANN) applied to the IDS system, the improved technique assumes recent advances in ANN studies such as adjusting the initial parameters over the unsupervised pretraining of deep belief networks (DBN) as well as attack-resilient and efficient system, enhancing the detection accuracy. An experimental outcome, which is provided by the proposed methods, will be discussed.
