An authentication approach in SDN-VANET architecture with Rider-Sea Lion optimized neural network for intrusion detection

Abstract

Vehicular Ad-hoc Network (VANET) offers expedient services in intellectual transportation models but is susceptible to several attacks. The Intrusion detection systems (IDSs) help to prevent security risks by discovering irregular network behaviours under local sub-networks inspite of complete VANET. An optimization-driven model is devised for detecting intrusion in SDN-VANET. The network comprises several entities, like the On-Board Unit (OBU), SDN, Road-Side Unit (RSU), and Authentication Server. The proposed approach involves five phases, namely registration, key generation, data encoding, authentication, and decoding. The authentication of OBU is done with the data packets. The detection of intrusion is performed using the optimized model. The selection of features is done from log files considering the Renyi entropy. The detection of intrusion is done using the Rider-based neural network (RideNN), trained with designed Rider-based Sea Lion Optimization (RBSLO), and is produced by blending the Rider optimization algorithm (ROA) and Sea Lion Optimization (SLnO). The proposed RBSLO-based RideNN offered to improve the achievement with the elevated precision of 92.5%, the recall of 95.4%, the F-measures of 94%, and the smallest computation time of 135.654 s.