Modern vehicles integrate an increasing amount of computational functionality, driven by the growing complexity of in-vehicle applications. At the same time, automotive system architectures are becoming more centralized, requiring powerful HPC platforms at the core. These platforms must deliver the performance needed for ADAS, AI, and autonomous driving, while also meeting stringent energy efficiency and safety requirements.The CeCaS project addresses these challenges across a wide range of topics and domains of expertise, including processor design in advanced FinFET technology, the transformation of the E/E architecture, and advanced packaging for automotive supercomputing platforms. Within CeCaS, our work focuses on application-specific accelerator design to enable efficient processing of compute-intensive workloads. In this paper, we present our contributions in this area, including the design of hardware accelerators for both conventional and neuromorphic AI workloads, the development and evaluation of representative AI benchmarks, and the use of virtual platforms for early design-space exploration and hardware/software co-design.

The European-funded HAL4SDV project aims to advance European solutions in software-defined vehicles by introducing a hardware abstraction layer positioned between executed software and execution units. HAL4SDV includes over 60 partners across 12 countries and receives funding within the Chips Joint Undertaking under Horizon Europe since April 2024 and is coordinated by TTTech Computertechnik. The proposed hardware abstraction layer includes safety-critical scheduling and platform deployment of software tasks, and is motivated by the requirement for abstracted hardware with unified interfaces in centralized automotive architectures.This work presents a correct-by-construction workflow which is developed by academic partners to schedule and deploy periodic software tasks onto diverse execution units. The workflow facilitates the execution of the same task stack on multiple unit architectures and consists of a task model and scheduling algorithm, which is followed by platform deployment for diverse hardware units, ensuring safe execution. In this multi-partner project, a bandwidth regulation unit for hardware accelerators and a RISC-V-based multicore system with tightly coupled memories are used as target platforms.A RoboRacer driving stack is chosen for evaluation, showing the viability of our workflow to schedule autonomous driving functions. To show generalization capability, synthetic task sets are additionally used to validate our deployment workflow.

Future automotive architectures are expected to transition from a network-centric to a domain-centered architecture featuring central compute units. Powerful domain controllers or smart sensors alleviate the load on these central units and communication systems. These controllers execute tasks with varying criticalities on heterogeneous multicore processors, and are ideally capable of dynamically balancing the computing load between the central unit and sensors. Here, Artificial Intelligence (AI) capabilities playa crucial role, as it is in high demand for such an automotive architecture. However, AI still requires specialized accelerators to improve their computation performance. Task-oriented distributed computing with criticalities up to ASIL-D necessitates the development and utilization of specialized methodologies, such as safety, through the isolation and abstraction of low-level hardware concepts. Meanwhile, online monitoring and diagnostics become vital features to detect errors during operation. The EMDRIVE architecture includes methods, components, and strategies to enhance the performance, safety, and security of such distributed computing platforms. The nationally funded EMDRIVE project connects its twelve partners from academia and industry and is currently in its intermediate stage.

The seamless and secure integration of subsystems is a pivotal requirement within contemporary automotive development, necessitating the application of design methodologies like the Vee model. While this approach includes dedicated verification steps for the included contexts and provides a high level of assurance that the system will operate correctly under specified conditions, formalizing specifications outside its operational design domain is per definition not included. Additionally, black-box systems like machine learning based functions prove difficulty to test by these traditional methodologies. In this project, we introduce and demonstrate a design workflow combining the Vee model design paradigm with continuous data-driven software engineering. Our workflow assists the continuous, safe and secure development and improvement of consumer vehicle functionality over the product lifecycle. This is achieved through the continuous monitoring of anomalies, as well as system states that deviate from the established design domain. The UNCOVER methodology consists of a continuous reduction in the amount of necessary monitored messages and presents a methodology throughout the entirety of the product lifecycle. We demonstrate our methodology through a simulation and show our automatic generation of monitoring components, and an automated preselection of identified safety or security incidents.

Distributed computing capabilities at the network’s edge enable new use cases, e.g., smart factories, industrial internet of things, or autonomous mobility systems. While new applications evolve, managing the resources and being capable of integrating and adjusting the execution of tasks in a distributed edge infrastructure is of great importance. For this, applications need to be migrated between different nodes. These migrations must happen without interruption, allowing the system to meet service requirements while fully utilizing all available hardware resources. Therefore, applications should also be executable on all different compute nodes in a heterogeneous edge system without interfering with each other. To this end, applications should be granted only necessary permission, especially when un-trusted applications are integrated into the system. We, therefore, propose a migration method for isolated applications across heterogeneous compute nodes in service-oriented edge architectures. A service-oriented architecture is used to decouple applications, allowing for flexible scheduling. The migration method enables the fast migration of sandboxed applications based on WebAssembly by utilizing a two-stage migration approach. The concept can utilize multiple communication protocols for management and service communication. We have implemented a proof of concept based on the Zenoh1communication protocol. While the time required depends on the communication protocol and the memory size, we achieved migration delays of under 51 milliseconds for smaller applications. By providing a method for fast migration for applications across heterogeneous compute nodes, it is possible for distributed edge infrastructures to run applications independently from each other and adjust the execution node based on changes in the system’s environment. By integrating applications into our framework, they are executed isolated and with strict access control, allowing for easier reuse.

An ongoing trend in the domain of embedded computing systems is the consolidation of functionality on few, high-performance platforms. This development also impacts real-time systems, where a shift to parallel architectures enables meeting the increased throughput demands. On these multicore platforms, memory contention is a central concern regarding time-predictability. Additionally, isolation between tasks is required to limit the impact of faults during run time. We propose a scratchpad memory-based approach to predictable execution that integrates runtime-based isolation mechanisms with an LET-based task model. In order to mitigate interference between cores, each core executes from a local memory, while the data transfers between the local memories and the shared main memory are incorporated into a global, static schedule. Our task execution model is based on the Logical Execution Time (LET) paradigm, which we extend to include explicitly scheduled data transfers similar to the Predictable Execution Model (PREM). The implementation and evaluation of our approach is ongoing and will be evaluated on a custom RISe- 'v, based multicore platform. This novel approach allows for consolidating hard real-time tasks with high demands for functional safety onto a single platform.

Complex edge systems are often structured with service-oriented architectures. Different communications stacks such as MQTT, DDS, or Zenoh are used, hindering reuse of service implementations across systems. One emerging solution for deploying such services is WebAssembly, which enables platform-independent, secure, and low-overhead execution. We propose a concept for integrating Web-Assembly modules into microservice-based architectures using a specialized runtime. This runtime manages the communication between the WebAssembly module and other parts of the system. The runtime for integration of WebAssembly addresses the challenge of reusing service implementations across systems with different communication protocols. At the same time, this provides isolated, safe and secure execution. Both capabilities are central to service-oriented edge systems.

Autonomous and distributed Industrial Internet of Things (IIoT) systems are increasingly developed and deployed. They have an enormous demand for resilience and availability. At the same time, they are in a constantly changing system environment. The underlying edge computing infrastructure is characterized by ever increasing processing power and connectivity as well as a high degree of decentralization. To reduce downtime and long redesign loops, self-adaptation capabilities are needed. Automatic reallocation of the executed tasks to the compute nodes is a possible self-adaptation measure. However, the reallocation should be compliant with the different demands, constraints and specifications of the design. At the same time, a major challenge is that the allocation decision should be fast enough to be calculated at runtime. This paper therefore proposes an allocation method that uses demands in the form of policies to compute automatic reallocation at runtime. The integration of the allocation method into runtime is enabled by combining constraint programming, step-wise multi-criteria solution approaches, and resource management at multiple levels. The policy-based allocation method is tested and evaluated in the context of a smart factory site for the function offloading of automated guided vehicles (AGVs) and driverless micromobiles. Our results show that the allocation method is capable of recalculating the allocation during runtime in milliseconds while maintaining design conformity. This enables the system to react to changes in the environment, thereby reducing the downtime of decentralized Industrial Internet of Things systems and increasing availability.

Innovations and novel applications in the area of the Industrial Internet of Things (IIoT) are driven by the technical possibilities of digitalization and edge computing. This leads to rapid advancements and enormous time pressure in the development and operation of new functionalities. Edge computing systems with self-x functionalities are able to react independently to changes in operation and thus mitigate this time pressure problem. The autonomous response during the operation of the self-x system must nevertheless remain compliant with the original system design requirements. A distributed edge computing system has complex requirements in different components and at different levels of the system. This leads to a major challenge when describing these requirements and constraints in such a way that they can be automatically checked and fulfilled during operation. This paper proposes a model-based description of policies that is used as a basis for reallocation of services during operation. The approach was tested and evaluated using an IIoT use case of a camera-based monitoring system for smart construction sites. Our results show that, based on the policy description, it is possible to automatically compute the reallocation when changes occur in the system, without any intervention from the developer. With this self-x capability, the system can remain in operation longer. Overall, this helps to reduce time pressure in the development, deployment and maintenance of new innovations and applications in the field of the Industrial Internet of Things.