Doctoral research at Birmingham City University on networks for connected and autonomous vehicles (CAVs). The work moves through four stages, each feeding the next.
An autonomous vehicle is only as safe as the network it talks to. I build a live digital copy of that network, fed by real Open RAN data, where failures and attacks can be found, rehearsed and fixed before they ever reach the road.
Measurements are collected from an Open RAN environment serving autonomous vehicles: signal quality, interference and throughput, reported from the radio units. The twin starts from observed behaviour instead of assumed values.
The wireless environment is reproduced in ns-3 and with MATLAB network ray tracing, which follows each signal path as it reflects off buildings or gets blocked. At 6G frequencies a single wall can decide whether a car stays connected. Try it: drag the car on the panel.
The digital twin runs on Microsoft Azure and follows the physical network in near real time. Because it mirrors the live system, what-if questions such as a failed cell or a surge in traffic can be answered without touching the real network.
Machine learning inside the twin watches for anomalies such as jamming. A response is tested in the twin first, then applied to the network, inside an architecture that is secure, efficient and aligned with 3GPP standards.
Open RAN splits the base station into units with open interfaces. Every layer becomes measurable, and every layer becomes a possible target. Select a component, or pick a threat to see where it strikes.