V2X communication — AlpineTutor

Every sensor on the previous three topics shares one limit — it can only detect what it can see. V2X is the attempt to get past that, and its difficulties turn out to be much less about radio than you would expect.

A weathered roadside equipment cabinet bolted to the base of a traffic signal pole at a multi-lane junction, a slim vertical antenna mounted on the pole above it, conduits entering the cabinet from below, the signal heads on the gantry overhead showing red and green, painted turn arrows and lane markings on the carriageway and a few distant vehicles under flat overcast daylight. — A roadside unit on a signal gantry. A junction that broadcasts its own phase is useful to the first equipped vehicle that arrives.

What it adds

A radar sees the vehicle in front. It does not see the one four ahead that has just braked hard, or the ambulance approaching the junction from a road hidden behind buildings, or the black ice that a car reported two minutes ago on the bend you are about to enter.

None of that is a resolution problem. It is a line of sight problem, and no improvement in sensing solves it.

The naming is mechanical — the V is always the vehicle, the second letter is whatever it is talking to:

V2V — another vehicle
V2I — infrastructure, typically a signal or a roadside unit
V2P — a pedestrian or cyclist carrying a device
V2N — the wider network, for traffic and hazard information
V2X — all of it, when the counterpart does not matter

The heartbeat

A diagram of V2X message types and range. A junction scene shows a vehicle receiving a basic safety message from a vehicle hidden behind a building, a signal phase and timing message from a roadside unit on the traffic signal, and a hazard notification originating several vehicles ahead beyond line of sight, with the on-board sensor range drawn as a much smaller shaded region for comparison. A panel describes the basic safety message as a short broadcast of position, speed, heading and braking state repeated about ten times a second by every equipped vehicle, noting that nothing is requested and no connection is established. A latency panel shows a direct transmission taking a few milliseconds against a path routed through a network server taking several hundred, with a note that a vehicle covers about twenty-eight metres per second at motorway speed. A final panel explains the penetration problem, that benefit grows roughly with the square of the equipped fraction of the fleet so the earliest adopters gain very little, and that equipping roadside infrastructure breaks the deadlock because a single equipped junction serves every equipped vehicle passing it. — No connection is established and nothing is requested. Every equipped vehicle simply says what it is doing, and everyone listens.

The foundation is the basic safety message — and the surprising thing about it is that it is not an alert.

Every equipped vehicle broadcasts, roughly ten times a second, a short packet saying where it is, how fast it is going, which way it is heading, and whether it is braking. Nothing is requested. No connection is set up. Receivers simply listen and assemble their own picture from whatever they hear.

Alerts are then derived locally from that picture. A vehicle that notices three broadcasts ahead of it reporting hard braking can warn its driver before anything is visible.

Alongside it:

Message	From	Carries
Basic safety message	Every vehicle	Position, speed, heading, braking
Signal phase and timing	A junction	Current phase and seconds to change
Map data	A junction	Lane geometry, so the phase means something
Decentralised notification	Anything	A specific event — ice, a stopped vehicle, roadworks

Latency, and why the network is not in the path

At motorway speed a vehicle covers about 28 metres every second. A hundred milliseconds is nearly three metres, and a warning that arrives after the event it warns about is worse than no warning at all.

A message routed up to a server and back can easily cost several hundred milliseconds — fine for traffic information, useless for collision warning. So the safety messages are transmitted directly between vehicles, never through the network.

This has a second benefit that is easy to miss: direct transmission keeps working in a tunnel, in a car park, and on a road with no mast within twenty miles.

DSRC and C-V2X

A diagram comparing DSRC and cellular V2X. Two stacks are drawn side by side. The DSRC stack is built on IEEE 802.11p, described as Wi-Fi adapted for vehicles, with contention-based channel access where stations listen before transmitting, shown degrading as the number of transmitters rises. The cellular V2X stack is built on 3GPP sidelink, the device-to-device part of LTE and 5G, with scheduled resource selection, shown holding up better under load, and with an additional network mode alongside direct mode for non-urgent traffic. A panel notes that both occupy roughly the same band around five point nine gigahertz and serve the same purpose but cannot interoperate, so a radio built for one cannot hear the other. A lower panel covers security, showing a message signed with a short-lived pseudonym certificate drawn from a pool that rotates, satisfying two opposing requirements, that a receiver can verify the sender is a legitimate participant while nobody can track an individual vehicle across a journey, with revocation shown as the mechanism for removing a misbehaving sender. — Same band, same job, mutually deaf. Which is most of the reason deployment stalled.

Two families, occupying roughly the same band around 5.9 GHz, doing the same job, unable to hear each other.

DSRC derives from IEEE 802.11p — Wi-Fi, adapted for vehicles. Channel access is contention-based: a station listens, and transmits if the channel seems free. Simple and mature, but it degrades as the number of transmitters rises, which is precisely the busy junction where you want it most.

C-V2X derives from cellular sidelink, the device-to-device part of LTE and later 5G. Access is scheduled rather than contended, which holds up better under load. It has two modes:

Direct mode — vehicle to vehicle, no base station, the safety-critical path.
Network mode — through the cellular network, for the less urgent traffic.

Trust without traceability

Two requirements that pull against each other.

A receiver must be able to verify that a message came from a legitimate participant. Otherwise anyone with a transmitter can broadcast a phantom emergency brake and empty a motorway.

But a permanent signature would let anyone with a receiver follow a particular car all day — every junction it passed, every time it left home.

The resolution is rotating pseudonym certificates. Each vehicle holds a pool of short-lived certificates, signs with one for a while, then switches. Messages remain verifiable; the sequence of them cannot be stitched into a journey. A misbehaving participant can still be revoked, through a process deliberately made harder than ordinary verification.

Why so little of it is on the road

The technology has been ready for years. The obstacle is arithmetic.

The benefit to a vehicle depends on meeting other equipped vehicles, so it grows roughly with the square of the equipped fraction of the fleet. At 2% penetration, two equipped vehicles meet about one time in 2,500. The first buyers pay in full and receive almost nothing.

Two things break the deadlock. Regulation, by mandating fitment so penetration rises regardless of individual return. And infrastructure — a junction that broadcasts its phase is useful to the very first equipped vehicle that passes it, with no dependence on the fleet at all. That is why the roadside is frequently equipped before the cars are.

What to take away

V2X supplies what sensors structurally cannot — information from beyond line of sight.
The basic safety message is a heartbeat, not an alert. Warnings are derived locally by receivers.
Safety traffic goes directly, never through the network, because the latency budget is tens of milliseconds.
DSRC is 802.11p-derived and contention-based; C-V2X is sidelink-derived and scheduled. Same band, same job, mutually deaf.
Rotating pseudonym certificates make messages verifiable without making vehicles trackable.
A signed message is still a claim — sensors confirm before anything acts.
Benefit scales with the square of penetration, which is why infrastructure and regulation, not technology, decide the pace.

Last in this module, what makes all of this updatable after the vehicle has been sold — software-defined vehicles and cybersecurity.