I. Introduction
In order to validate and homologate the radiated immunity of any automotive system, car manufacturers need to comply with standards and regulations such as ISO 11451 and ECE R10 [1]. The immunity tests at vehicle level allow to validate a complete vehicle indoor (in a semi-anechoic chamber) on a roller bench, by activating the complex functions that are operated by several embedded electronic components (modules) [2], [3]. Automotive functions, currently, are divided into two types. First of all, functions that are based principally on one sensor, such as Advanced Driver-Assistance Systems (ADAS) functions (e.g. a Line Departure Warning function) [4]. For these simple ADAS functions, it is easy to perform the radiated immunity tests, by stimulating the sensor with a known scenario and monitoring how the function reacts [5]. We can also find very complex ADAS functions based on very sophisticated systems, requiring multiple sensors, such as the Adaptive Cruise Control function, which is made of a radar and a camera [6], [7]. To validate such complex functions, one needs to synchronize the stimulations of the sensors and one needs to monitor how the system reacts during the test [2], [3], [5]. In the coming years, one will also have to validate autonomous functions. These autonomous functions are based on multiple sensors using artificial intelligence (AI) that command the functions and verify the consistency of data provided by the sensors. To validate autonomous functions and complex ADAS functions, the current trend for the moment, is to re-create outdoor scenarios indoor, by developing a test bench made of several sensor stimulators. Such test bench must be able to synchronize all the stimulations between each other [5], [3]. However, this solution is limited because synchronization of different stimulations of sensors cannot be done for the moment in current test facilities, and some sensors are difficult to stimulate with physical signals (e.g. LIDARS meant to map the surrounding environment). Furthermore, if the embedded software detects incoherencies in the data and information provided by the sensors and external RF links, these functions automatically disable themselves and one cannot continue the validation process. This issue will become more prevalent as self-monitoring algorithms will become more and more efficient in order to detect malfunctions and cyber-attacks.