In the run-up to its annual Technology Forum event, Freescale semiconductor has announced a new microcontroller family aiming at automotive applications. Featuring Ethernet connectivity and video compression, the MPC5604E is intended to make parking assistant systems more affordable.
In next-generation automotive environments, data traffic over internal data bus systems will significantly increase. In the discussion about future data bus architectures for in-vehicle use, chip maker Freescale bets on Ethernet - at least in specific application segments such as driver assistant systems that make use of image-processing. "We bring Ethernet into the driving car", said Stephan Lehman, Director Global Automotive Marketing for Freescale.
The statement referred to the fact that in the current vehicle generation Ethernet is used exclusively for diagnosis purposes when the car is in the workshop. Freescale however plans to use the non-deterministic Ethernet system in moving vehicles - in application fields hitherto not claimed by competing network architectures such as MOST or FlexRay. According to Lehman, Freescale plans to employ Ethernet in the first place to transport compressed video signals from the camera to driver assistant systems which process these data. Examples could be parking assistant systems which generate a top-view image based on data provided by several rear-, front- and side view cameras. In such applications, bandwidth would be an important factor, but the network system does not necessarily need to show a deterministic behavior. Thus, Lehmann expects a "largely standard 10/100 Mbit/s Ethernet without real-time abilities" for this type of application. The only modification to the standard protocol is that the application adds hardware-based AVB compliant time stamping.
In other automotive applications, Ethernet could complement and over time displace the MOST technology, Lehmann said. Ethernet as a well established industry-standard technology could offer significant cost benefits over MOST with its costly licence fees. "Theoretically, Ethernet could even be used as backbone technology", he said, "but this is certainly many years away."
In order to give designers of image-processing driver assistance systems a processor powerful enough for this kind of application, Freescale has designed another member of its Qorivva MCU family which is based on Freescale's 32-bit Power architecture. Sitting next to multiple cameras placed around the car, the MPC5604E directly supports Ethernet connectivity and video data compression. The device transmits compressed high-resolution video data over Ethernet to a processing unit which generates a 360° view around the vehicle. The resulting top-view image is displayed at the dashboard as part of a parking assistant system. Again, Ethernet could help to cut costs: While for video data transmission, typically four or five low-voltage differential signaling cables at cost of about $10 per cable are required, the transport of these data via Ethernet could lead to a lower overall number of cables and thus save money, Lehmann explained.
The Ethernet support can also be regarded as a step towards more open, scalable architectures in the automotive context. Another example for open standards usage in this context is the JPEG data format for the compressed video images.
According to Freescale, the MPC5604E is a direct result of a cooperation with BMW. The auto maker in the past had heralded the use of Ethernet in cars at several opportunities.
Systems based on the new microcontroller and Ethernet in cars will go into volume production in 2013, Lehmann said. He however declined to name the customer.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.