**1.3 Cybersecurity for autonomous driving system**

With the increase in connectivity and communication between vehicles, traffic management systems and other elements of the transport infrastructure, the attack surface and potential vulnerabilities are also increasing. One of the main challenges in implementing cybersecurity in autonomous systems is that security mechanisms such as encryption, authentication and integrity checks require time and computing resources. These additional requirements can potentially impact the real-time capabilities of the systems by increasing latency and slowing the response time of autonomous vehicles. However, with careful planning and innovative solutions, it is possible to achieve both cybersecurity and real-time performance without compromising the safety and reliability of autonomous driving systems.

The importance of cybersecurity has been acknowledged by lawmakers, leading to the introduction of UNECE Regulations R.155 and R.156 [4, 5]. These regulations establish requirements for the cybersecurity of vehicles and their systems, and require the automotive industry to take appropriate security measures to ensure the cyber resilience of their vehicles.

The combination of cybersecurity and real-time capability requires close collaboration between the various disciplines involved in the development of autonomous driving systems, such as vehicle engineering, software development and IT security. An appropriate IT infrastructure that provides both cybersecurity QoS and real-time IoT capabilities is crucial for the safety and reliability of autonomous vehicles. To achieve this, the following concepts and ideas are presented, which enable an efficient combination of cybersecurity and real-time capability to ensure the safety and functionality of autonomous driving systems.

#### **1.4 Multisensory input information**

In addition to vehicle data, a variety of external sensor-generated data, serverbased environmental data and even satellite-based positioning information are used as input variables in the ADS. External sensor-generated data includes Car2Car communication. This ensures that the speed and distance of autonomous road users in the vicinity are monitored.

The real-time requirements in the immediate vicinity of autonomous vehicles are obviously higher than those in the superimposed environments, from which, for example, spatial or environmental data are obtained. Decentralization (edge computing) in the IoT network allows the next action decision to be made as close as possible to the distributed sensors. This decision is then made available to higher-level intelligent instances for further coordination and regulation of the overall process. As a result, there are multiple levels of interaction in the IoT network. During the software development process, it is important to consider the transitions between the different interaction levels.

In Section 6, we will consider velocity and position control. It is important to note that the time to acquire data, calculate the next action and provide instructions must be at least twice as fast as the process speed or constant to control the current process in real-time [6]. In addition, certain safety requirements for the ADS can only be ensured by guaranteeing real-time conditions in the information chain. It is obvious that there is some interplay between cybersecurity and safety in terms of real-time requirements. However, this issue is not addressed in this article.

We assume that both cascaded and cross-layer control loops are likely to become necessary to meet the varying requirements of the different layers of the hierarchical model, e.g. hardware, operating system, software, Car2Car, server and cloud. In order to calculate the continuous autonomous driving speed for all collision-free positions, the information chains require the processing of multisensory input information, resulting in a MIMO (multiple input, multiple output) system [7, 8]. We consider the multiple antenna approach on the transport level as given. We also want to evaluate the driving behavior of the AD, and for this purpose we introduce the term "ADS driveability" in Section 7. We want to encourage an objective evaluation of the driving experience of an AD, as this can ultimately be a decisive factor in the competitive use of ADS services.
