DEEP NEURAL BASED LIGHTWEIGHT ROUTING AND PROBABILITY PLOT CORRELATION FOR SECURE VANET COMMUNICATION

Abstract

Vehicular Ad-Hoc Networks (VANETs) are pivotal for controlling traffic flow and intensifying safety in Intelligent Transportation Systems (ITS) through effective transmission of data between nodes. In spite of that, VANETs pose notable issues, like, vulnerability to malicious activities and intricate routing issues that can compromise network security and performance. However there still remains room for improvement in the development of VANETs, considering, security weaknesses and latency in data transfer process. Most of the routing methods right now are not effective and secure, specifically when utilized in high-mobility environments. The requirement to design an intelligent and secure routing protocol that establishes low latency, high packet delivery ratio and resilience against malicious attacks is what motivates this work. A novel secure route based data transfer in VANET called, Deep Neural Lightweight Multi-objective and Ryan-Joiner Correlation (DNLM-RJC) is proposed. The DNLM-RJC method for secure route based data transfer in VANET is split into four layers, namely, one input layer, two hidden layers and finally one output layer. The network traffic patterns obtained are processed in the input layer. In the first hidden layer, the authentication mechanism of vehicle nodes is implemented by using Lightweight Multi-objective Secure Route-based Authentication model. This model enhances network security by preventing data manipulation from malicious nodes. After the node authentication is done, the second objective, secure data transfer is executed via second hidden layer. In order to perform effective and secure data transfer in VANET, Ryan-Joiner Correlation Coefficient is applied. The experiments are conducted on the Comprehensive Vehicular Communication Network Attack Dataset. We demonstrated the significance of the proposed DNLM-RJC method in comparison with studies employing secure route based data delivery in an extensive manner. Experimental results show that the proposed method can improve the authentication delay with improved attack detection rate and also reduces end to end delay significantly on different size samples.