Optimizing Vehicular Networks with Variational Quantum Circuits-based Reinforcement Learning

TL;DR

The paper tackles the problem of jointly optimizing autonomous-vehicle kinematics and network decisions in vehicular networks to ensure safety and reliable connectivity under stochastic channels. It proposes a Variational Quantum Circuit (VQC) based Multi-Objective Reinforcement Learning (MORL) framework, formulating the task as a MOMDP and using a quantum Q-function approximator $Q(s,a;\theta)=\langle O_a\rangle_{s,\theta}$ with a 5-qubit, 3-layer circuit. Training relies on a Bellman-based loss $\mathcal{L}(\theta)$, experience replay, and an $\epsilon$-greedy policy, achieving faster convergence and higher rewards than DDQN. Results demonstrate substantial gains in convergence speed ($31.32\%$) and rewards ($\approx 18.64\%$), highlighting the potential of quantum agents for real-time, multi-objective decision-making in dynamic VNets integrating RF and THz base stations. This work provides a scalable, quantum-enhanced approach to jointly optimize handover decisions, data rates, and vehicle trajectories in complex wireless-vehicular environments.

Abstract

In vehicular networks (VNets), ensuring both road safety and dependable network connectivity is of utmost importance. Achieving this necessitates the creation of resilient and efficient decision-making policies that prioritize multiple objectives. In this paper, we develop a Variational Quantum Circuit (VQC)-based multi-objective reinforcement learning (MORL) framework to characterize efficient network selection and autonomous driving policies in a vehicular network (VNet). Numerical results showcase notable enhancements in both convergence rates and rewards when compared to conventional deep-Q networks (DQNs), validating the efficacy of the VQC-MORL solution.

Figures (2)

• Figure 1: Training performances (ego vehicle): (a) Total telecommunication reward (b) Total transport reward (c) Collision Rate
• Figure 2: Testing performance (ego vehicle): (a) Total telecommunication reward (b) Total transport reward (c) Total reward. The considered VQC architecture has 5 qubits and 3 layers.