Demonstrator for dependable edge-based cyber-physical systems

• Simon József Nagy Budapest University of Technology and Economics
• Richárd Szabó Budapest University of Technology and Economics
• Máté Levente Vajda Budapest University of Technology and Economics
• András Vörös Budapest University of Technology and Economics

Resumo

Dependable cyber-physical systems (CPS) are increasingly used in various application fields, such as urban mobility, smart city, industrial IoT and telecommunication. Beside functional requirements, dependable CPS systems have to meet several extra-functional requirements such as reliability, availability, fault-tolerance and performance. The complexity of modern CPS systems significantly increased since the extensive use of distributed services, redundant architectures and advanced safety mechanisms. In addition, several new technologies have emerged in the edge, such as embedded GPU-s, AI acceleration and virtualisation tools, which enhance the extra-functional properties, such as latency and performance of dependable CPS systems. Because of the increased complexity and the cutting edge technologies, evaluating the extra-functional requirements becomes difficult for modern CPS systems. Consequently, several new analysis techniques have also been developed. We developed an open-source demonstrator for dependable edge-based CPS systems in the field of smart city and urban mobility. With the demonstrator, the researchers can compare and evaluate different technologies, safety mechanisms and analysis techniques. The demonstrator consists of several emerging technologies such as hardware accelerators, load-balance mechanisms, containerisation and container deployment tools. The architecture of the demonstrator was developed following the edge computing paradigm and model-driven engineering approach. The demonstrator contains distributed redundant and fault-tolerant services. We also developed a hardware-in-the-loop (HIL) test environment to simulate various environmental scenarios and evaluate extra-functional properties.