Enhancing DLT Performance in Vehicular Networks via Connectivity-Aware Tip Selection and Reinforcement Learning

Abstract

The application of Distributed Ledger Technology in Intelligent Transportation Systems ensures a secure and decentralized mechanism for data sharing among vehicles and infrastructure. However, DAG based protocols such as the IOTA Tangle when applied to autonomous vehicular systems face significant challenges in ledger consistency, transaction confirmation, and convergence due to the highly dynamic and intermittently connected nature of vehicular networks. To address these limitations, this paper proposes a Connectivity Aware Intelligent Distributed Ledger Construction Model tailored for autonomous vehicular environments. The proposed framework integrates a connectivity aware tip selection mechanism with a reinforcement learning strategy based on sliding window bias thompson sampling to dynamically select optimal ledger construction actions under non-stationary network conditions. Vehicular connectivity is modeled using an alternating renewal process, and transaction arrivals by a nonstationary Poisson process. Extensive simulation results demonstrate that CA-IDLCM outperforms URTS, MCMC, and Biased-TS approaches by achieving confirmation rate of 95.8%, and maintaining the orphan fraction well below 2%, with tip counts stabilizing near 1.0. highlighting its robustness, adaptability, and suitability for next generation Intelligent Transportation Systems.