Hierarchical Evolutionary Optimization with Predictive Modeling for Stable Delay-Constrained Routing in Vehicular Networks

TL;DR

This work addresses delay-constrained, stable routing in highly dynamic VANETs by introducing a hierarchical optimization framework based on MOEA/D, augmented with LSTM-based mobility prediction and incremental route adjustments. A formal two-objective routing model is developed, minimizing $T(P(t))$ (delay) while maximizing $S(P(t))$ (stability), and is solved via MOEA/D with dynamic adaptation. The framework integrates an LSTM predictor to forecast node movements and uses incremental adjustments to rapidly react to topology changes, yielding a diverse Pareto front of routing solutions. Empirical results in an emergency-response scenario demonstrate low delay and high stability, with a reported average travel time of $2.13$ s and stability score of $0.72$, illustrating practical benefits for real-time, reliability-critical VANET applications.

Abstract

Vehicular Ad Hoc Networks (VANETs) are a cornerstone of intelligent transportation systems, facilitating real-time communication between vehicles and infrastructure. However, the dynamic nature of VANETs introduces significant challenges in routing, especially in minimizing communication delay while ensuring route stability. This paper proposes a hierarchical evolutionary optimization framework for delay-constrained routing in vehicular networks. Leveraging multi-objective optimization, the framework balances delay and stability objectives and incorporates adaptive mechanisms like incremental route adjustments and LSTM-based predictive modeling. Simulation results confirm that the proposed framework maintains low delay and high stability, adapting effectively to frequent topology changes in dynamic vehicular environments.

Figures (2)

• Figure 1: Training error of the LSTM model
• Figure 2: Pareto Sets with Travel Times and Stability Score