The trend toward autonomous driving, coupled with public concerns about the safety of driverless vehicles, has made cybersecurity a top concern for automotive original equipment manufacturers (Oems). The integrity of vehicle systems and control of the vehicle must be ensured to ensure the safety of drivers, passengers and pedestrians. Cybersecurity is essential for things like networked car subsystems, but it’s also critical for image sensors used in advanced driver assistance systems (ADAS) and driver monitoring.
Image sensors act as the eyes of the car, supporting ADAS functions such as lane departure warning, pedestrian detection, and emergency braking. They help car systems assess their surroundings and monitor driver behavior. In the future, they will also assist in identifying and verifying the identity of car users and monitoring their vital signs in order to control the car via an onboard computer if the driver is incapacitated. Therefore, the image sensor must remain in normal use, especially in the extreme conditions that the car may encounter.
Network security threat
Automotive image sensors are primarily affected by four types of cybersecurity threats: forgery, tampering, bypassing, and eavesdropping (especially for in-car applications).
Due to the current supply shortage in the automotive semiconductor industry, counterfeit products are on the rise. While installing non-genuine parts may not be malicious, it can affect system performance. Because non-genuine parts use different boot processes, protocols, firmware, and software, the slightest consequence is that the ADAS system simply doesn’t work. In the worst case scenario, the system uses non-conforming components with severely degraded performance, resulting in compromised system safety functions.
Automatic Emergency braking (AEB) systems operate on the premise that their image sensors have well-defined characteristics (such as high dynamic range and low light performance) and are calibrated to these specifications (such as exposure control and frames per second). Counterfeit sensors may look identical to the real thing, but their performance and characteristics are quite different. For example, counterfeit cameras may use the same sensor, but have not been tested to ensure that the final component meets performance requirements, which may manifest as failure under high intensity work. That is, it works under normal conditions, but degrades or simply fails under other conditions, such as hot, sunny days or cold winter nights. Some sophisticated counterfeits may mimic real sensors to support initialization operations or simple device health checks, but their performance in dynamic range or frame rate is greatly compromised. Because AEB systems are optimized using genuine components, the performance degradation of counterfeit alternatives can also affect the performance of the system, with potentially catastrophic consequences. For example, an object or pedestrian that could have been detected at a distance in front of the car, leaving a reaction time of a few seconds, may now be detected only within a few meters, not enough time to avoid a collision (Figure 1).