Enhancing security in vanets: Adaptive Bald Eagle Search Optimization based multi-agent deep Q neural network for Sybil attack detection

Currently, the use of Vehicular Ad-Hoc Networks (VANETs) has gained significant attention in toll management systems and traffic control. VANETs facilitate effective communication by connecting Roadside Units (RSUs) and vehicles. VANETs can ease decision-making for drivers, meanwhile, they carry some problems with security since they often modify topology. In VANET, the Sybil attack is a specific attack, that might generate traffic congestion and affect transportation safety services. Different Mechanisms have been implemented to discover various attacks in VANET, yet VANET meets diverse attacks. Therefore, this research article developed an effective Sybil attack detection model namely Adaptive Bald Eagle Search Optimization (ABESO) based Multi-agent-Deep Q Neural network (MA-DQN). The principal objective of the ABESO based DQN is to enhance the security level of VANET by identifying the Sybil Attacks. In this, clustering and effective cluster head selection are performed to discover the Sybil attacks. In the suggested ABESO based DQN algorithm, robust clustering is carried out, in which vehicle nodes of VANET are clustered through the utilization of the BIRCH clustering technique. Our proposed ABESO based DQN algorithm augments the overall network efficiency by effective cluster head selection. Taylor-based Waterwheel Plant (TWP) is exploited in the cluster head selection and diminishes the overhead in the network. In the proposed model, the MDQN-based approach selects the features and ABESO based DQN delivers an optimal output, i.e., it discovers normal and Sybil attacks. Experimental results are carried out on the basis of the sybil attack detection dataset that holds multiple data regarding attacks. The detection results affirm that the efficiency of the proposed ABESO based DQN approach is superior and outperformed previous methods.