E/E Architecture Synthesis: Challenges and Technologies

Abstract

In recent years, the electrical and/or electronic architecture of vehicles has been significantly evolving. The new generation of cars demands a considerable amount of computational power due to a large number of safety-critical applications and driver-assisted functionalities. Consequently, a high-performance computing unit is required to provide the demanded power and process these applications while, in this case, vehicle architecture moves toward a centralized architecture. Simultaneously, appropriate software architecture has to be defined to fulfill the needs of the main computing unit and functional safety requirements. However, the process of configuring and integrating critical applications into a vehicle central computer while meeting safety requirements and optimization objectives is a time-consuming, complicated, and error-prone process. In this paper, we firstly present the evolution of the vehicle architecture, past, present, and future, and its current bottlenecks and future key technologies. Then, challenges of software configuration and mapping for automotive systems are discussed. Accordingly, mapping techniques and optimization objectives for mapping tasks to multi-core processors using design space exploration method are studied. Moreover, the current technologies and frameworks regarding the vehicle architecture synthesis, model analysis with regard to software integration and configuration, and solving the mapping problem for automotive embedded systems are expressed. Finally, we propose four research questions as future works for this field of study.

Performance Evaluation of Zone-Based In-Vehicle Network Architecture for Autonomous Vehicles

“…Until 2019, vehicles used distributed IVN architecture with all ECUs connected to a central gateway [ 13 , 14 ]. This architecture is suitable as an IVN architecture for vehicles with a small number of ECUs and transferring and receiving small amounts of data.…”

“…In DIA, multiple domain controllers perform data collection, calculation, and control according to their functions, but in ZIA, a single high-performance computing unit performs all tasks. This unit works like a brain and also acts as a central gateway to pass data from one zone to another [ 13 ]. Autonomous vehicles require high-performance autonomous applications that collect data from multiple sensors, make situational decisions, and control the vehicle.…”

The Future Roadmap of In-Vehicle Network Processing: A HW-Centric (R-)evolution

“…In [13] authors focus also on Ethernet as a backbone of the IVN and propose an architecture and specific configuration for a particular use case, which, similarly to the proposal in [10] and [12], is also composed of a central node and several distributed nodes interconnected between them. In [14] authors give an overview on how all the new functionalities can be mapped to the IVN central processing unit, while providing the required level of reliability and performance. They also identify the current bottlenecks at software level in terms of handling the required complexity.…”

Hardware Acceleration of Data Distribution Service (DDS) for Automotive Communication and Computing

“…Now, each zonal gateway needs to manage different network technologies, different kinds of functionalities and traffic with different criticality levels. Authors in [25] review the challenges of current IVNs, focusing on the complexity of software configuration and mapping of functionalities to available resources. In [26], authors analyse the requirements of new ECUs in order to provide the capabilities demanded by autonomous and connected vehicles.…”

“…The recent exponential increase in complexity of automotive systems [5], due to the integration of novel functionalities like assisted or autonomous driving, has shown the limits of the original specifications, calling for a revolutionary shift in the design of the computing platforms and the related software stacks [6], [7]. Current luxury cars already contain more than 100 ECUs for a total of more than 100 million lines of code [8].…”