Today's automotive electrical system architecture is heavily distributed. Gone are the days of centralized, single-box solutions — today's mainstream architectural choice is a multitude of electronic modules interconnected with multiple vehicle networks. True, the positive business value of the early J1850 experience has helped our industry converge toward today's common off-the-shelf networking solutions such as CAN, LIN and FlexRay, but the transition toward distributed-ness is not finished. Upcoming x-by-wire subsystems will also be distributed. Even automotive propulsion systems are transitioning from early centralized engine control into distributed architectural components.
But this more than 10-year automotive industry transition into the distributed domain has been no easy task. All of this distributed-ness has involved a substantial industry effort, and the success of these distributed architectures has employed significant resources for both OEMs and suppliers. For the automotive electronics engineering community, the advancement of this distributed technology requires several things, especially tools, education and standardization.
In the distributed development environment, the engineer's toolbox must include the right tools. Early OEM-developed bus analyzers that simply collected the flow of network data bits and bytes have been replaced by powerful, programmable tool solutions; tools that measure, play back, simulate, calibrate or even flash program. No longer managed by OEMs alone, an entire tool industry has been spawned. Whether simply analyzing or simulating a complex distributed function, today's ability to interact directly with vehicle network information in engineering units or in graphical form is a significant step forward. Even though the use of Vector's CANoe as a “remaining network simulation” tool is one of the prime choices by many OEMs, many beginning engineers lack the knowledge of distributed systems necessary to fully use such tools. While future tool development will continue to provide value, as an industry, we also need to focus on education to increase the level of knowledge.
As some OEMs look for such talent, the number of engineers who are knowledgeable in developing distributed embedded systems is rather small. While some universities actively teach control systems that are also distributed, the distributed engineering textbooks remain unwritten and the current expertise is essentially limited to the automotive industry. Vector's “CAN Goes to College” was created to help the academic community. As a part of its first use in a class project effort, Kettering University's Juan Pimentel, produced a remarkable safety-critical network solution called FlexCAN. This innovation is an event-based application in the midst of a near-complete mindset that time-triggered control is the only choice. More cooperation between the auto industry and the academic community will be necessary to expand the knowledge base.
Standardization is another major contributor to the success of going distributed. Although today each OEM uses its own homegrown solution, the level of commonality is actually quite high. While each OEM may use its own internal terminology, the essential distributed concepts are nearly the same across all car companies. But standardization is more than common terminology; industrywide common requirements must be documented. In the midst of those engineers who continue to believe that their particular individual contribution represents some competitive advantage for their company, there are others who recognize that the technology that surrounds these distributed embedded systems has little to no competitive relevance. They see that a single common solution is of much higher importance, and this is exactly the business case for AUTOSAR. AUTOSAR is the global convergence toward an industry standard solution shared across all OEMs. There is little doubt that the AUTOSAR agenda will be a success — it is not a question of if, but only a question of when.
In my opinion — it is tools, education and standardization that will provide the biggest benefit to our industry's continued evolution toward distributed-ness.
ABOUT THE AUTHOR
Bruce Emaus is the president of Vector CANtech (the U.S. subsidiary of Vector Informatik) and also the chairman of the SAE Embedded Software Standards Committee.