Apr 21, 2017 by Kai Richter
With discussion around the subject of automated driving gathering pace, the development of Advanced Driver Assistance Systems (ADAS) looks set to become the ‘next big thing’ in automotive electronics. The development of ADAS introduces new challenges and significantly higher levels of design complexity, making the adoption of effective timing design, analysis and verification tools essential in their design.
“Even in terms of where automotive electronics is today, ADAS is a disruptive technology that will fundamentally challenge the way automotive E/E and software systems have been built and integrated in the past,” said Dr. Kai Richter, Engineering Director at Zoreza Global. “ADAS adds enormously to the software cross-dependencies in vehicles, which, in order to execute all the required new functions, will require additional levels of computing performance and network communications far exceeding that of established vehicle E/E systems. To make the step-change needed to turn concepts into reality, the integration of effective timing design, analysis and verification tools into the overall design process will be crucial.”
With the development of ADAS, comes the introduction of new types of sensors, new object and trajectory recognition functions and new virtual reality video projection technology among others. These will also generate data for the automated driving functions to predict and determine how the vehicle should act in terms of the primary control functions, such as accelerating, steering, braking etc. All these control functions must also cope with new safety and reliability requirements. To facilitate this, in-vehicle computing power will need to be dramatically increased with topologies changing from heavily distributed controllers to fewer, high-performance computing centers that are connected directly to dozens of sensors and actuators spread over the vehicle.
Architecturally, Ethernet will come to dominate on-board and in-vehicle communications, providing the conduit for new functionality as well as existing control signals that currently use CAN and FlexRay.
In fact, the newly developed generation of ADAS hardware platforms look like large E/E systems already. Several automotive and general purpose cores, graphics and other special co-processors run parallel software that communicates through Ethernet within a single box. The current reference implementations include a mix of time-triggered and event-triggered components and a host of possible safety hazards. To simplify portability, provide interoperability between vehicle domains and application protection, new concepts in software integration, such as virtualization and partitioned scheduling, will be required.
“It is no simple task to efficiently utilize these platforms,” continued Dr. Kai Richter. “Also, they have to be utilized optimally as the software usually grows faster than the platform developers can imagine. Timing analysis is the key enabler for this optimization and, to assist in the process, we provide customers with consultancy services for requirements collection and the development of optimized system architectures for the highly integrated systems found in the ADAS domain today. They also value our ongoing commitment to ensuring that our tools are kept fully up-to-date for the requirements and new technologies being introduced.”
The Symtavision Tools adds worst-case timing analysis for Ethernet communication in several flavors including AVB and time-triggered Ethernet, and also covers the ADAS software technology for both AUTOSAR and non-AUTOSAR operating systems, with timing analysis for partitioned and hierarchical schedulers (with or without hypervisor). In addition, the end-to-end timing analysis supports heterogeneous mixes of both event and time-triggered applications and architectures, as well as statistical simulation and worst-case analysis, and compatibility with the existing OSEK, CAN, LIN, and FlexRay modules.
Webinar 23rd May 2017 - From ADAS Functions to Scalable Systems
Find out more on Automotive IQ
