DefinePK

Abstract

Vehicular Ad Hoc Networks (VANETs) defines a self-organizing network formed between vehicles in roadside infrastructure for facilitating real time decision making on road. However in real time trusted updates it can introduce a latency in high vehicular traffic conditions. To address these constrains, developed a blockchain based federated system with intrusion detection to secure the data flow of vehicular environment. Initially, Road Side Unit (RSU) and vehicle contains a Federated Learning Node (FLN), Edge Key Management Unit (EKMU), and access to the Blockchain-Backed Trust Ledger (BBTL). FLN gathers data, trains a local machine learning model and transfers encrypted data to an aggregator. The EKMU creates a lightweight cryptographic key pair and sets attaches a pseudonymized identity and a trust score on the blockchain. Truncated Polynomial Ring Unit (NTPRU) used in EKMU generates a lightweight cryptographic key pair. At the same time, the federated intrusion detection system uses the lightweight XGBoost which continuously monitors traffic patterns and the data points of behaviors while looking for abnormalities (anomalies), such as spoofing, replay attacks, or false-data injections. If found a threat, Federated Intrusion Detection System (FIDS) produces alerts and reports to BBTL. Finally, trust scores are constantly classified using DNN-19 and updated in BBTL. Then the Deep Reinforcement Learning Routing Engine (DRL-RE) can use these updated trust-based metrics to construct secure and adaptive routing decisions. The proposed strategy achieved recall of 97.91%, NPV of 97.62%, error of 2.240% and accuracy of 97.5% respectively. The proposed approach accomplishes secure Vehicular communication for employing a decentralized learning-based approach for enables a real-time threat detection for more resource resilience in intelligent transportation systems