An additional step to eliminate the interference is to look at the radar sensors in traffic as multiple units of the same sensing system. In this vision, the radar sensors work together to achieve the same shared goal. (NXP)

Mitigating radar-to-radar interference

An effective radar interference mitigation strategy should have the right balance between complexity and capability of dealing with the interferers.

Sensors are a fundamental building block of highly-automated and autonomous-driving vehicles. And each different sensor – radar, lidar, and camera systems, has strengths and weaknesses. Radars are typically more robust than cameras to adverse weather conditions, more cost-effective than lidar, and provide a better velocity and distance resolution. However, typical drawbacks can be lower angular or cross-range resolution and radar-to-radar interference.

There are advanced solutions that further increase the radar angular resolution, from sparse arrays to super-resolution algorithms. Solutions also exist to improve the radar's sensitivity to distinguish small targets from strong targets nearby, such as a pedestrian next to a large truck or a building. As more and more vehicles are fitted with radar systems, interference is of growing concern.

Radar interference is a well-recognized challenge. As they all use the same allocated frequency spectrum, multiple radars may transmit at the same time and frequency. If they also share a common visible path, they will interfere with each other.

The MOre Safety for All by Radar Interference Mitigation (MOSRIM) report is possibly the first project to assess the severity of the radar interference problem. A National Highway Traffic Safety Administration (NHTSA) study builds on that and adds different traffic scenarios to determine when and how often interference happens.

A common traffic scenario is shown in Figure 1: Vehicles that have long-range, front-facing radar with a relatively narrow field of view (FoV) approach each other. Interference happens as the FoV of the two sensors overlap, and the radars are active simultaneously and in the same frequency spectrum.

Radar interference techniques 
There are three classes of techniques available to mitigate radar interference without affecting overall system performance, as shown in Figure 2. The first class of techniques avoids the saturation of the front end, which happens when a radar sensor is exposed to a strong interferer. The second manages digital interference by recognizing and removing the interference in the digital domain. The third class avoids interference by adapting the radar waveform to reduce the probability of interfering with other radars.

These techniques attempt to mitigate the adverse effects of interference before or after it happens. As they are implemented to every sensor in their respective radar system independently, they do not provide a robust strategy to avoid interference in a structured manner.

A possible approach could be to statically allocate resources to radar applications. For instance, front and rear radars could use non-overlapping parts of the spectrum to avoid the worst-case interference scenario of a front-looking radar illuminating the rear-looking radar of the car in front. Using different polarizations for different applications might mitigate this worst-case scenario as well, but proper considerations on the influence of polarization on the FoV of the antenna and on the propagation should be considered.

For more advanced solutions, the radar sensor could borrow channel access rules developed by the telecommunication industries, such as TDMA, FDMA, CDMA, or OFDMA. Alternatively, the solution could take a more randomly-determined approach, using protocols such as ALOHA, CSMA, or CSMA-CA. This is usually deployed at the medium access control (MAC) layer of the communications stack.

Following the former approach would require a form of centralized coordination, which means that every radar sensor should communicate to a central unit, e.g., the telecommunication infrastructure, what it would like to do. It should receive the time and frequency slot back from the infrastructure that it can safely use to sense the environment and limit the interference.

The latter approach is distributed, whereby each radar sensor follows the same transmission protocol, which ensures fairness and a given performance. The ALOHA protocol is the most straightforward approach. In it, each unit transmits when it wants to, which is basically what the current radar sensors do.

ALOHA simplicity 
One of the first improvements of the ALOHA protocol in communications is the slotted-ALOHA. If, for instance, a collision of two transmitted signals occurs, it happens for the entire transmission frame; there are no possible partial overlaps of messages. This simple trick already doubles the efficiency of the transmission in communications.

As compared to communications, radar sensors do not need to communicate with each other to sense the environment. One could, therefore, only standardize the way they access the channel, leaving the full freedom to deploy any radar waveform. For instance, going back to the ALOHA and slotted-ALOHA example, one could think of organizing the wireless resources in time and frequency blocks; for example, 20 ms by 250 MHz. This technique would allow any radar sensor to use an integer number of time/frequency slots. The time synchronization could come from the GPS signal, which is already available in the car.

If agreed by all parties, this simple measure could allow more sensors to act in the same environment while giving the industry the full freedom to differentiate in terms of radar waveforms.

More complex and efficient multiple access schemes could also be envisioned. For instance, a form of sensing of the available time and frequency resources and a mechanism of randomization in accessing the channel, similar to the CSMA-CA protocol used in Wi-Fi networks. Further exploration is required to identify the proper resource allocation protocol for radar applications. It should be based on the MAC protocols but with significant adaptations to consider the different kinds of traffic, priority settings, and quality of service targets.

Whatever the solution will be, an agreed way to access the shared resources will surely increase the maximum number of radar sensors co-existing in the same environment.

Cooperative sensing 
An additional step can further be taken to eliminate the interference. Instead of looking at the radar sensors as individual units, one could look at the radar sensors as multiple units of the same sensing system. In this vision, the radar sensors work together to achieve the same shared goal. They will need to be equipped with a communication link to all other sensors to coordinate access to the shared channel and share information, acting as a larger ecosystem.

An example of how this would work in a scenario would be when an application needs to have a good image of the environment. It could query the radar sensing system to build that image; however, the information might come from a radar sensor mounted on a different car or road infrastructure. Such cooperative sensing would avoid the interference a priori because no unit is seen as an interferer, but they all work together.

The proper radar interference mitigation strategy does not need to be the most efficient. Having the right balance between complexity and capability of dealing with the interferers strongly depends on the foreseen scenarios. Combinations of randomization, detection and avoidance of interference might be powerful enough to support a significant market penetration of radar sensor systems.

There will be more elaborate techniques to deal with radar-to-radar interference. Eventually, there will be some form of agreement within the radar sensor community to share the sensing resources effectively. Ultimately, there will be a standardized way to access the communications channel while, at the same time, maintain the possibility to have differentiating sensing performance.

For further information, see: https://www.nxp.com/applications/automotive/adas-and-highly-automated-driving/automotive-radar-systems:RADAR-SYSTEMS

Continue reading »
X