+********************************************************************** * * * Einladung * * * * Informatik-Oberseminar * * * +********************************************************************** Zeit: Freitag, 15. Dezember 2023, 9.00 Uhr Ort: Raum 2202 (HBau, 2. Stock), Ahornstr. 55 Referent: Alexandru Kampmann M.Sc. RWTH Lehrstuhl Informatik 11 Thema: A Dynamic Service-oriented Software Architecture for the Automotive Domain Abstract: Today, almost all foreseeable areas of innovation in the automotive domain are software-driven. This development has led to the coining of the term Software-defined Vehicle (SDV), where the critical differentiator between competing OEMs becomes the software and not necessarily the hardware. Critical to this vision is the ability of the OEM to continuously provide software updates throughout the vehicle's lifetime. Implementing SDVs requires novel architectures in almost all technical domains. For example, it is foreseeable that centralized high-performance computers communicating via real-time Ethernet will replace the multitude of distributed, customized ECUs in today's architectures. Most importantly, SDVs require novel, modular, and updatable software architectures to enable short development cycles and rapid time-to-market of software innovations. This thesis presents ASOA, a software architecture for the automotive domain. While today's architectures are rigidly integrated at an early stage of development, ASOA follows the idea of dynamic runtime integration. The integration of software services in ASOA is dynamically established late in the development cycle - at the execution time of the system. Individual services are agnostic to the overall system, as an Orchestrator acts as the architecture controller that bundles system-specific knowledge. Continuous addition and modification of the software architecture are vital aspects of SDV vision. This agility is difficult to achieve with traditional architectures, where modifications to large parts of the systems are quickly required to perform architectural adaptions. In ASOA, these adaptations affect the Orchestrator and not necessarily individual services, which can be kept free of the details about the system they are part of. ASOA distinguishes between functional orchestration and resource orchestration. Functional orchestration controls operating modes, data flow, and active services, allowing for dynamic architecture reconfiguration. In addition, functional orchestration supports the ad-hoc creation of redundant service structures and fault-tolerant execution of the Orchestrator. Resource orchestration provides methods for assigning compute cluster resources to the software services using multi-criteria optimization. It enables a reduction of the compute cluster power consumption and of the execution times of causality chains distributed across multiple services and ECU. ASOA also comes with a tool for system engineering that captures various views of the system architecture. The portable implementation of ASOA enables transparent, API-identical execution of services on both Linux-based high-performance computers and resource-constrained microcontrollers with minimal operating systems. The implementation also enforces explicit dependencies between the consumption of computation time and data communication, both essential factors influencing the timing behavior of the software. Knowledge of these dependencies is critical to uncovering causality chains and functional dependencies. Various critical functions are implemented in full-scale vehicle prototypes using ASOA in the project UNICARagil. Es laden ein: die Dozentinnen und Dozenten der Informatik