mmWave V2X Communications and Proof-of-Concept

Background
Automated driving vehicles are expected to be the killer application of 5G and the solution to traffic problems. For example, today’s traffic accidents are mainly caused by human failures, but automated driving vehicles are controlled by electronics instead of human, and thus are expected to effectively reduce traffic accidents. A great challenge is that automated driving vehicles must have full information of the environments without any blind spot, which often appears due to the limited LOS/FOV of onboard sensors and could result in false detections of on-road objects and lead to collision accidents. The cooperative perception is one of the most promising ways to address the challenge. Its key idea is to share the real-time sensor data among infrastructures and vehicles through wireless communications to eliminate the blind spots cooperatively.

Fig.1 Cooperative Perception

mmWave V2X for Cooperative Perception
The cooperative perception’s communication requirement is still an open question. Current vehicular communication standards and frequency bands are not insufficient for the huge amounts of data from onboard sensors, and the millimeter-wave communication is expected to be a strong candidate because of its wide bandwidth and large communication capacity. Fig. 2 is derived from the typical overtaking scenario, and shows the relation between required sensor data rate to detect the oncoming vehicle and the feasible channel capacity at each carrier frequency. It can be concluded that the millimeter-wave communication with cooperative perception is able to effectively increase the driving safety and allows vehicles to safely overtake at 51 km/h.

Fig.2 Channel Capacity vs Required Data Rate

Indoor Testbed Development
The hardware prototype as the Proof-of- Concept of mmWave vehicular communication for the cooperative perception and automated driving has been developed. The hardware structure is illustrated in Fig. 3. The moving robots (Kobuki) equipped with LiDAR sensors, mmWave APs, and onboard controllers act the automated driving vehicles, and they share the sensor data with other robots through the Fig. 15 Architecture of cooperative perception. mmWave data plane. All sensor data flows and link connections in the mmWave data plane are controlled by the SDN controller based on the real-time context information (e.g., position, orientation, channel status) of the robots through a Wi-Fi network which acts the control plane.

Fig.3 Indoor Testbed

Outdoor Testbed Development
For the details of the outdoor V2X Proof-of-Conception implementation, please refer to this demo.

Fig.4 Outdoor V2X PoC Implementation