Japan.co.jp

An autonomous vehicle does not merely need a connection to the internet. It can generate enormous amounts of perception data: camera video, LiDAR point clouds, high-definition three-dimensional maps and information from roadside sensors. If some of that data can move quickly between vehicles and infrastructure—and if the same radio link can help determine where the vehicle is—the road network itself starts to become part of the sensing system.

A Kyoto University team led by Professor Hiroshi Harada, Associate Professor Yusuke Koda and doctoral student Hiroaki Endo reported on October 1 that it had demonstrated that idea on an actual road. A moving vehicle transmitted uplink data at up to 10.3 Gbit/s in the 60 GHz band, while the same communications waveform was used to estimate vehicle speed and position. The experiment used 1,840 MHz of bandwidth, roughly five times the 400 MHz maximum channel bandwidth allocated to current 5G in Japan.^[1]

The breakthrough is not just the headline speed

The more unusual part of the experiment is that the radio signal did two jobs. The researchers estimated Doppler frequency from the received communications waveform, converted it into vehicle speed and then integrated that speed from a reference position to estimate location. Maximum speed error was on the order of 5 km/h, while average position error was about 3 meters.^[2]

This is an example of integrated sensing and communications, or ISAC. Instead of treating a base station only as a device that moves bits, ISAC uses properties of the radio wave—such as Doppler shift, delay and reflections—to infer position, motion or the presence of objects.

An 18 km/h car on a real road

The test took place on a road extending roughly 330 meters in a straight line from an intersection. The transmitter-equipped vehicle traveled at about 18 km/h. Its antenna was 1.5 meters high, while the receiver sat approximately 5.8 meters above ground inside a building with a view of the road and intersection. Standard-gain horn antennas with 25 dBi gain were used at both ends.^[2]

One detail makes the field result particularly noteworthy: the receiving antenna did not track the moving car. Its beam was fixed on a point about 100 meters from the intersection center. The transmitting antenna was oriented toward the intersection in the horizontal plane, but the infrastructure-side beam was not continuously steered to follow the vehicle.

The researchers defined a link as usable when 5G block error rate, or BLER, stayed at or below 0.1. With QPSK at MCS 7, the system delivered 3.4 Gbit/s to about 250 meters from the intersection. With 64QAM at MCS 17, it delivered 10.3 Gbit/s out to roughly 150 meters.^[2]

Why use 1.84 GHz of spectrum?

One of the most fundamental ways to increase wireless capacity is to use more bandwidth. In rough terms, a wider radio channel is like adding more lanes to a road. Japan’s current 5G allocations use channel bandwidths up to 400 MHz. The Kyoto test used 1,840 MHz—about five times as much.

The researchers operated around 60 GHz within the 57–71 GHz frequency range associated with the n263 band considered for 5G. At these frequencies, channels approaching 2 GHz wide become possible. The team built its testbed around 5G OFDMA waveforms, expanding subcarrier spacing from the 120 kHz maximum commonly used in current 5G to 960 kHz to support the ultra-wide signal.^[2]

That does not mean a normal 5G smartphone has suddenly become a 10-gigabit device. This was a research testbed using specialized software-defined radio equipment, much wider spectrum and experimental parameters. The 10.3 Gbit/s figure is a physical-layer data rate under the test conditions, not an everyday application throughput or a promised commercial service speed.

The attraction—and difficulty—of 60 GHz

The 60 GHz band offers enormous bandwidth, but high frequencies come with harsher propagation. Signals attenuate more quickly and are more easily blocked by buildings, vehicles, people and vegetation. They do not bend around obstacles as effectively as lower-frequency signals. Practical systems therefore depend heavily on antenna gain, beamforming, dense infrastructure or multiple links.

Vehicle mobility adds another layer. A moving car produces Doppler shift and a fast-changing radio channel. The narrower and more directional the beam, the harder it becomes to maintain the link as the endpoint moves. Kyoto’s team implemented synchronization and demodulation processing in software-defined radio specifically to keep the ultra-wide 60 GHz signal usable in motion.^[2]

Why positioning matters when GNSS already exists

Satellite navigation is powerful, but it is not equally reliable everywhere. In dense urban canyons, signals can be blocked or reflected by buildings, degrading accuracy. Kyoto University explicitly cited this limitation as one reason to explore positioning from communications infrastructure.^[2]

A radio-derived position estimate would not necessarily replace GNSS. Its value may be redundancy. Future vehicles could fuse satellite positioning with inertial sensors, cameras, LiDAR and infrastructure-based radio measurements. A communications network that already exists for V2X could contribute another independent estimate without requiring a separate positioning system at every site.

How V2X moved from safety messages to shared perception

The early vision of vehicle-to-everything communication focused on small, urgent messages: speed, heading, traffic-light state, emergency braking, roadworks and collision warnings. Japan has considered the 5.9 GHz band for such V2X uses, while research has also explored 28 GHz and other millimeter-wave 5G bands.^[2]

Autonomous driving changes the scale of the data. A roadside camera might see a pedestrian hidden behind a truck. LiDAR at an intersection can create a three-dimensional point cloud of vehicles and cyclists. A high-definition map may need rapid local updates. Sharing that perception requires far more capacity than a traditional safety beacon.

The same Harada laboratory demonstrated another step in this progression in June 2026, using sub-terahertz waves to transmit 1.7 Gbit/s over more than 300 meters to a moving vehicle with a 920 MHz-wide, 5G-standard-based waveform. That experiment emphasized longer range. The October 60 GHz work doubled the bandwidth to 1.84 GHz, pushed the data rate above 10 Gbit/s at shorter range and added simultaneous positioning.^[3]

Why point clouds need this kind of link

LiDAR represents the environment as large sets of three-dimensional points. Higher resolution produces more points and more data. Multiple high-resolution cameras add another large stream. Autonomous vehicles typically process much of this locally, and they do not need to upload every raw sensor bit to the cloud.

But infrastructure sharing creates a different opportunity. A vehicle cannot see through a bus, around a sharp corner or beyond the range of its own sensors. A roadside unit or another vehicle may see exactly what it cannot. Ultra-high-speed V2X can make it possible to exchange richer pieces of that environment model quickly enough to be useful.

In that sense, future V2X is less about “giving the car internet” and more about turning a road corridor into a cooperative sensor network. The importance of a 10 Gbit/s uplink is that the vehicle itself can become a high-capacity source of environmental information, not merely a recipient.

ISAC turns communications infrastructure into a sensor

Integrated sensing and communications also has an infrastructure argument. Deploying separate radio, radar and positioning systems requires separate spectrum, antennas, power, sites and maintenance. If one network can carry data and extract sensing information from the same waveform, some of those resources may be shared.

Standards work is beginning to reflect that direction. In 2026, 3GPP created Release 20 technical work addressing integrated sensing and communications in 5G-Advanced system and charging contexts. That does not mean a commercial 6G ISAC standard is complete; it does show that the concept has moved from an academic niche into formal global standardization activity.^[4]

The next problem is not headline speed—it is messy reality

The test vehicle traveled at about 18 km/h. That is relevant to city streets, but highway speeds create larger Doppler effects and faster handovers. The field test also involved a single vehicle and controlled equipment. A commercial road has trucks blocking radio paths, multiple cars competing for capacity, rain, snow, pedestrians and transitions between infrastructure nodes.

The approximately 3-meter average positioning error is useful but not enough by itself for lane-level autonomous control. Its likely value is as an additional measurement to fuse with better-localized sensors and positioning systems. Future work will need to test multiple base stations, beam tracking, higher speeds, more severe blockage and multi-user operation.

Then come the non-radio problems: spectrum policy, roadside-site density, fiber or wireless backhaul, equipment cost, interoperability, cybersecurity and responsibility when infrastructure-supplied perception is wrong.

A fundamental demonstration, not a finished network

Kyoto University describes the result as a fundamental demonstration. Parts of the research were supported through commissioned projects of Japan’s National Institute of Information and Communications Technology and a Ministry of Internal Affairs and Communications program aimed at advanced 5G radio-link technology and international standardization through Japan-U.S. industry-academia cooperation. The team plans to present the work at the IEICE Short Range Wireless Communications research meeting on November 19–20.^[2]

Still, the distinction between a laboratory simulation and a car actually moving down a road matters. More than 10 Gbit/s was transmitted from that car while the same waveform produced meter-scale position information. The receiver beam did not have to chase the vehicle throughout the run.

If 6G is understood only as “faster downloads on a phone,” this kind of experiment looks excessive. A more useful picture is a network that communicates, measures and helps build a live digital model of the physical world at the same time. Kyoto’s 60 GHz test is an early but unusually concrete glimpse of that possibility.