V2X does not lack radios, because cars can already communicate directly, use cellular infrastructure, operate in dedicated spectrum and rely on standards that have existed across several network generations;356 what remains much less glamorous is coordination.
Francesco Linsalata and colleagues approach the problem as an extension of something cellular networks already do for thousands of mobile terminals. Their architecture applies O-RAN ideas to V2X so a programmable controller can track link conditions, anticipate some beam failures and coordinate decisions that would otherwise be made separately by each vehicle.12
The result is compelling, although turning it into “problem solved” would erase most of the engineering still between a simulation and deployed infrastructure.
Radios everywhere
V2X means Vehicle-to-Everything. The umbrella covers vehicle-to-vehicle, vehicle-to-infrastructure and, in some specifications, communication involving bicycles or pedestrians.5 In Europe, the 5.9 GHz ITS ecosystem includes access technologies derived from IEEE 802.11 and Cellular-V2X.56
3GPP Release 16 also formalized multiple pieces of 5G V2X: system architecture, application-layer support, policies and NR sidelink radio work.34
The difficult question is therefore no longer how to invent the first protocol that lets two cars exchange a message, but what happens when hundreds of vehicles exchange messages in the same place while obstacles, directional beams, different technologies and moving decisions all compete for radio resources.
Decide together
O-RAN introduces programmable controllers into the radio access network, where specialized applications, xApps, can observe measurements and influence parameters on near-real-time timescales.2 Moro, Linsalata and colleagues transpose that idea into V2X by giving a more reliable control plane responsibility for following vehicles and coordinating higher-frequency V2V links.2
The intuition is straightforward: one car has a local view of its link, while a network controller can see several cars, several links and more context at once.
That matters at 28 GHz, the frequency used in the simulation described by IEEE Spectrum,1 because narrow beams can carry large amounts of data while another vehicle, a building or a turn in the road can quickly remove the path they depend on.
In Linsalata's description, the model combines urban geometry, road layout and live vehicle GPS positions into a network digital twin.1 The control layer can use that context to select beams or prepare a relay before the direct link disappears.
Avoid collisions
IEEE Spectrum describes a five-minute simulation over one square kilometre of urban area, using OpenStreetMap roads, SUMO traffic and roughly 50 to 70 vehicles per kilometre.1 Within that scenario, the authors report very high message-collision rates for an uncoordinated design, whereas O-RAN coordination drives the simulated rate close to zero;1 two caveats matter before treating that number as a road-ready result.
First, this is a simulation, rather than a city where vehicles from ten manufacturers have driven for months: the academic work proposes an architecture and numerical analysis without documenting a commercial deployment.2 Second, avoiding a radio collision leaves application interoperability untouched, because two cars can successfully transport a packet while disagreeing about the exact meaning of its contents.
Speak alike
The IEEE article then makes a more ambitious proposal: use the O-RAN control layer as an intermediary between vehicles from different ecosystems.1
Linsalata describes a familiar interoperability problem: proprietary firmware and different data formats can prevent one manufacturer's car from directly interpreting another manufacturer's sensor data.1 A common layer could normalize those inputs before they move further through the network.
The architecture is plausible on paper; the universal translator does not exist today as a deployed standard.
3GPP specifications already define standardized V2X architecture and application support.34 Standardizing a network layer, however, does not force manufacturers to expose every internal sensor or represent every proprietary data structure in the same way.
Interoperability therefore needs an industrial and regulatory decision in addition to radio engineering.
Mobile O-RAN
The academic paper itself treats full integration as unfinished architectural work,2 and the reason is visible in the object being coordinated: O-RAN was developed around radio elements that are mostly fixed in place, whereas a car moves quickly, changes environment and may act as terminal, telemetry source or relay.
The authors therefore propose extensions that allow the vehicle communication stack to become an element actually managed through the O-RAN control framework.2
That change has an interesting consequence: each car stops behaving like a tiny isolated network.
The network becomes a system trying to distribute decisions across several moving machines. Cellular architecture has already had to solve a related coordination problem for phones, even though the speed, safety and precision requirements of vehicle communication are much harsher.
Years left
IEEE Spectrum is explicit about the standards problem: the required vehicular extensions would still need adoption in the O-RAN ecosystem and recognition inside 3GPP V2X specifications.1 That convergence does not arrive through one Tuesday-night software update.
It requires stable interfaces, compatible vehicles, operators willing to expose the required control, safety guarantees and common semantics for shared data, which is why the article's provocative title can be turned around.
The problem of cars communicating badly is already solved in some of its components. Cellular systems know how to coordinate huge numbers of terminals, O-RAN makes part of that coordination programmable, 3GPP defines V2X architecture, and simulations show that broader network control can reduce radio conflicts.123
What remains unsolved is assembling those components into shared infrastructure where manufacturers, networks and software agree to speak closely enough alike.
The bottleneck is probably not the next miraculous radio. It is more ordinary and more difficult: making systems that can already talk finally understand what the others are saying.