How anomaly detection is helping OEMs make autonomous vehicles safer

0

AI-based anomaly detection is the last word in cybersecurity, and almost nowhere is it more important than in connected cars. As manufacturers venture further into the realm of connected, and eventually self-driving, vehicles, a solution that catches vulnerabilities before they can cause any damage becomes a vital security requirement.

Most vulnerabilities today are unknown until an attack takes place, as conventional cybersecurity measures can only detect and prevent known threats. For autonomous vehicles, that’s unacceptable; any threat that harms the operation of the vehicle’s steering, brakes, or any one of dozens of other critical systems, could mean tragedy.

To prevent those kinds of attacks, of course, you need to look under the hood of the vehicle before it hits the road – and install an effective security system during the manufacturing or assembly phase. One company that has been working with OEMs to do just that is Tel Aviv-based SafeRide Technologies. SafeRide offers a system called vSentry AI, a behavioral profiling and anomaly detection system that uses advanced machine learning and deep learning technology.

How does anomaly detection work? The security system – in this case vSentry AI – builds a profile of the vehicle’s operation, based on data collected by the OEM during the normal operation of the vehicle. If, during the course of a ride, data indicates that something anomalous is going on, the security system goes into action. In some cases the data is uploaded and analyzed in the cloud and an alert goes out to a security operations center (SOC) for further analysis. In other cases data can be analyzed locally in the car and it can alert the driver/controller that a hazardous situation may be developing – and enabling the driver and/or the controlling system to take action to avoid danger, such as pulling the vehicle to the side of the road and stopping it. With anomaly detection, the security system doesn’t look for threats, like malware, to act on – it looks at vehicle behavior.

With connected vehicles, however, there is a problem; the amount of data collected is tremendous. Large amounts of data could take large amounts of time to upload. SafeRide has worked to address this, too. The company offers a CAN optimizer, which, it says, “dramatically decreases the bandwidth needed to do so by providing 95-99% reduction in data size, with a typical lossless compression ratio more than 15 times better than other compression algorithms that are currently on the market.” According to SafeRide, the compression is a direct outgrowth of its AI technology. Thanks to its machine-learning technology, it can make do with sending only some of the data to the cloud for anomaly detection, enabling accurate and fast analysis of anomalies

That OEMs need this urgently is clear from the numbers. By 2025, 80% of new cars sold in Europe will be connected, while in the USA that figure will be 90%. In the case of trucks, about half will be equipped with advanced telematics. Like any connected device or system, once a vehicle is online, you’re riding on a wing and a prayer and in that sense, SafeRide’s anomaly detection is an answer to manufacturer’s security prayers.

Share.

About Author

Independently submitted opinions from our readers. Share your opinions by sending up to 500 words to john.thornton@ukimediaevents.com and you could earn US$100! Note: Payment will be made on publication on AutomotiveTestingTechnologyInternational.com. Only original contributions will be considered and the editor's decision is final.

Comments are closed.