eSOL has joined the SOAFEE (Scalable Open Architecture for Embedded Edge) Special Interest Group (SIG). SOAFEE is an industry-led collaboration defined by automotive manufacturers, semiconductor suppliers, cloud companies, and open-source and independent software vendors.
As a SOAFEE partner, eSOL plans to help the company develop a cloud-native architecture for mixed-critical automotive applications. The architecture will be based on open-source software, and will incorporate commercial and non-commercial offerings.
Normal cloud services
Cloud services are familiar from smartphones, PCs or embedded consumer electronics (smart TV, Internet radio): Netflix, Spotify, WhatsApp. These cloud services have also been available in cars for a long time. However, they are only executed in specially protected software areas (sandbox).
These areas can tolerate application errors, crashes or security vulnerabilities without causing a physical accident. If an app slows down a smartphone or exposes one's account data, it's not nice, but neither is it directly relevant to survival.
Development in the cloud
On the other hand, millions of cloud service developers have now learned to take advantage of the cloud for their software development processes as well. Around the clock, they extend their cloud services, build, and test them in different stages from subcomponent level and finally up to the whole application. As a result, the functionality of the cloud services can be changed whilst retaining a working version available at all times.
For a long time, something like this was unthinkable in automotive for safety reasons - bug fixing gladly, but never introducing function changes or even function extensions under no circumstances. Every changed line of code always carries the risk of new errors or unintended side effects.
Cloud service developers have solved this problem through continuous testing. While the functions are continuously extended, extensive tests are also carried out in parallel in an automated manner and continuously extended.
SOAFEE focus: Functionally safe development in the cloud
SOAFEE is now focusing on developing just such an environment for automotive development that satisfies applications with functional safety requirements. It is not about cloud services like Netflix and not primarily about OTA. On the contrary, it is about placing the car as a digital twin in the cloud in such a way that automotive developers of functionally safe applications can use the advantages of a virtual environment, while still taking functional safety into account.
The hype scalable computational performance of cloud hardware allows complex functions to be tested in the cloud through extensive testing, saving a lot of development time. Keyword "SDV" (Software Defined Vehicle): The future functionality of a car is to be realized more and more by software. However, large software systems also require more extensive test scenarios. A local development server will sooner or later reach its capacity, whereas in the cloud, computing power can be added and removed as needed.
It should not only be possible to fix bugs after delivery of the vehicle, but also to add new functions. As already mentioned, this does not apply to smartphone apps like Netflix, but to safety-relevant functions such as enhanced facial recognition or an additional display effect for the LED turn signal.
OS for functional safe applications
eSOL's role in SOAFEE will be to define a common platform for future OEM software architectures for software-defined vehicles. In doing so, eSOL will leverage its expertise regarding functionally safe development processes and the corresponding RTOSes and hypervisors.
Furthermore, only eSOL can currently provide an RTOS that is manycore capable and that runs on manycore hardware such as the Arm Neoverse CPUs defined by SOAFEE as the standard for HPC. Traditional RTOSes were developed in the days of single/dual cores, and on such systems exhibit a communication bottleneck that slows down the entire application. With one or two CPU cores there wasn’t any problem, but semiconductor suppliers already supply 128-Arm-core CPUs for vehicles. For many developers in the embedded automotive sector, this is currently unthinkable: soon we will also see chips with even more cores!
For functionally safe applications on this hardware, eSOL with eMCOS will be a good real-time platform for complex systems.
Senior Product Marketing Manager
Michael has been working in the field of embedded applications for 25 years. He started as a Product Development Engineer for gas sensors in explosives hazardous areas across multiple industries, and then moved to the area of chip manufacturing of 32-bit MCUs as a system engineer. Finally, he was responsible for 64-bit multicore SoCs for automotive as a Product Manager. Since April 2020 Michael has served as Product Marketing Manager for embedded software products at eSOL, especially in the area of real-time operating systems for 64-bit multicore SoCs for automotive and industrial applications.