Unified Framework for Studying Handover in O-RAN-Based V2X

Abstract

Cooperative and see-through driving require the high-throughput, low-latency data exchange provided by millimeter-wave Vehicle-to-Everything (V2X) systems. Yet, the reliance of these systems on narrow directional beams makes link stability highly vulnerable to vehicle mobility and frequent coverage transitions. While the open radio access network (O-RAN) architecture enables intelligent control of handovers via data-driven optimization, current simulation tools lack the end-to-end framework to implement these control loops and evaluate their performance under realistic traffic conditions. To address this limitation, this paper introduces a unified framework that tightly couples microscopic traffic dynamics with an executable, O-RAN-compliant handover control loop in ns-3. Within this ecosystem, we develop an enhanced interface that enables bidirectional communication between the simulated radio access network and a Near-Real-Time RAN Intelligent Controller. It natively supports key performance measurement and RAN control service models specifically extended for vehicular handover. Furthermore, we introduce a modular, offline reinforcement learning training and deployment workflow to optimize handover decisions safely without risky online exploration. By providing this validated, fully open-source toolchain, our work bridges the critical gap between architectural O-RAN development and practical, road-constrained V2X deployment scenarios.