Search Results

This paper explains why software for efficient model-based development is needed to improve the efficiency of automakers and suppliers when implementing solutions with next generation automotive embedded systems. The resulting synergies are an important contribution for the automotive industry to develop safer, smarter, and more eco-friendly cars. To achieve this, it requires implementations of algorithms for machine learning, deep learning and model predictive control within embedded environments. The algorithms’ performance requirements often exceed the capabilities of traditional embedded systems with a homogeneous multicore architecture and, therefore, additional computing resources are introduced. The resulting embedded systems with heterogeneous computing architectures enable a next level of safe and secure real-time performance for innovative use cases in automotive applications such as domain controllers, e-mobility, and advanced driver assistance systems (ADAS).

The current automotive electronic and electrical (EE) architecture has reached a scalability limit and in order to adapt to the new and upcoming requirements, novel automotive EE architectures are currently being investigated to support: a) an Ethernet backbone, b) consolidation of hardware capabilities leading to a centralized architecture from an existing distributed architecture, c) optimization of wiring to reduce cost, and d) adaptation of service-oriented software architectures. These requirements lead to the development of Zonal EE architectures as a possible solution that require appropriate adaptation of used security mechanisms and the corresponding utilized hardware trust anchors. 1 The current architecture approaches (ECU internal and in-vehicle networking) are being pushed to their limits, simultaneously, the current embedded security solutions also seem to reveal their limitations due to an increase in connectivity.