Modern vehicles are often described as ‘computers on wheels’, and considering the growth in software-based functions, it is hard to argue otherwise.
Consumer demands for electronics within the vehicle have soared over the last decade or so, fuelled by a familiarity with connected services afforded by modern smartphones. The automotive industry, along with various municipalities and other bodies around the world, have also set lofty targets to improve road safety statistics through autonomous driving. As a potential answer to the challenge of powering all these functions, so-called ‘hypervisors’ can enable automakers to pack in more functionality with greater ease, and without a raft of additional hardware.
Hypervisors work by separating an operating system from its physical hardware to allow for several virtual ‘guest’ operating systems to run. Hypervisors have been around since the 1960s and have become widely used in consumer electronic devices. Moving forward, they are expected to gain similar favour with automotive manufacturers across the industry.
In the not-too-distant past, some premium passenger cars would have featured anywhere between 70 and 150 distributed computers. These would range from very small microcontrollers with dedicated functions, such as a window lifter or side indicator, to larger system on chips (SoCs) for cockpit controls.
But with each model cycle, a growing number of ECUs are entering the vehicle and with them, more functions: autonomous driving features, over-the-air (OTA) software updates, fleet telematics and even high definition video streaming. To get around this, developers have utilised more advanced ‘multicore’ ECUs that can process multiple functions simultaneously. A hypervisor is designed to enable even greater functionality, but without increasing any physical requirements. This trend is apparent across the automotive market, touching the computing systems used in anything from passenger cars up to a Class 8 heavy-duty truck.
“Automakers want to sell more features to customers,” explained Kai Lampka, Architect, EB corbos Hypervisor at Elektrobit, during a recent Automotive World webinar. “It is not only about enabling remote diagnostics, fleet analysis and data collection. Autonomous driving, for example, will also artificial intelligence (AI) algorithms, deployed on high performance platforms. For this you need high computing power available in the car.”
All of these requirements will have an impact on the kind of software that is used in future vehicle architectures, with automakers expected to move away from a hardware-centric approach and instead utilise hardware-independent software. As Lampka puts it, this will create a ‘virtual hardware environment’.
No way around the hypervisor
While interest in hypervisor technology grows, the question remains as to whether it is truly vital for vehicle architectures moving forward, or whether it is simply a useful addition. From Lampka’s position, a widespread preference for hypervisors in future is inevitable.
“I have devoted a large chunk of my professional career toward understanding the future of embedded computing, and I truly believe that we have only seen the beginning,” he said. “We will see massive multicore architectures that are extremely inexpensive compared to what we have today, and I don’t see how there is a different path to orchestrate different software stacks running side by side, whilst also giving health, safety and security monitoring.” In fact, he believes that the automotive industry will follow the path of high performance and Cloud computing, suggesting that “the approach of using a hypervisor is key… There is no other way around it.”
But as with any computer system, the subject of cyber security remains, and hypervisors do not alleviate this challenge. “If you use open source software this poses new security threats, and you must harden your software stacks so that a hack into the system is not possible, or that the risk is at least limited,” warned Lampka. “You want to minimise the attack surface.”
Elektrobit is not the only firm invested in automotive-specific hypervisor technology. BlackBerry has been developing its own offering for some time, and in June 2017 launched its second-generation hypervisor based on its QNX operating system. By creating so-called ‘software containers’, any “hiccup or breach” in a functional domain can be isolated, says BlackBerry, without creating vulnerabilities in other domains of the car. Chip giant NXP has been working with OpenSynergy on hypervisor tech, while Tier 1 suppliers Visteon, Qualcomm and Harman have also utilised hypervisors for digital cockpit platforms in recent years.
