2 Time Synchronization for Radar With IEEE 802.1as

Why Radar Needs a Shared Time Base

Sensor fusion without temporal aliasing: Radar and camera each have limits. A camera is sensitive to darkness, glare, and weather, while radar has lower angular resolution and cannot read signs or lane markings. Fusing the two overcomes the limits of each, but only if both measurements describe the same instant. Consider a vehicle traveling at 30m/s. If a camera captures a frame at t = 0 and the radar captures the frame at t = 10ms, the vehicle has moved 0.3m between the two measurements, and the fused result places the same object in two different positions. A shared time base lets the central processor timestamp each measurement and compensate for the difference, or lets the system trigger the sensors at the same moment in the first place.

Aligned frame start across radar sensors: Radar sensors that run from individual clocks drift apart by the parts-per-million (ppm) difference between those clocks, so the frames slide against each other over time. When the PHY in each sensor provides a frame trigger derived from the IEEE 802.1AS time, every radar starts the chirp sequence at the same network time. This action is useful whenever several sensors are meant to work as one system, the central ECU can combine the data without needing to compensate for timing differences. This function is essential for distributed aperture radar.

Distributed aperture radar: In a satellite architecture the central ECU combines data from several sensor heads. A further step is distributed aperture radar (DAR), in which two or more smaller radar heads operate together as one radar with a much larger virtual aperture, giving a finer angular resolution than any single head can. This requires the heads to be synchronized in time with nanosecond-level accuracy, which is well beyond what software-based synchronization can deliver. The quality of the Ethernet time base is therefore a direct performance factor for the radar system. See ZEN-3P-DARSW for more information on distributed aperture radar.

How IEEE 802.1as Works

IEEE 802.1AS (gPTP) is one of the Time-Sensitive Networking (TSN) standards. This standard lets every node in an Ethernet network share a common wall clock. One node acts as the leader and provides the reference time. Each follower node exchanges timestamped messages with the leader to correct the offset between the two clocks, correct the frequency drift of the local clock, and measure the path delay of the link. The achievable accuracy depends on how precisely the timestamps are captured. For more information on the gPTP algorithm and the timestamp handshake, see Time Sensitive Networking With Ethernet PHY.

Timestamping Location

A packet can be timestamped in the PTP software stack, in the MAC of the processor, or in the PHY. Any variable delay between the wire and the point where the timestamp is taken becomes a timing error, so the closer to the cable the timestamp is taken, the more accurate the synchronization. The software, start-of-frame detect, and hardware methods are described in How to Implement IEEE 1588 Time Stamping in an Ethernet Transceiver. For ADAS sensors such as radar, the PHY is the best place to take the timestamp, since the highest level of accuracy is achieved. A single-digit nanosecond level of accuracy can be achieved with this method, which is essential for applications such as distributed aperture radar.

Figure 2-1 PTP Timestamping Location