Traditionally, vehicles used to have standalone systems such as telematics, instrument cluster, and the infotainment system. To reduce the cost of development and maintenance of these independent systems, vehicle OEMs are adopting a single hardware approach. But having all the functionality on a single hardware platform introduces security and safety concerns. This is where hypervisors play an important role. A hypervisor is a software component that provides a virtual environment allowing multiple systems to run on the same hardware platform, while providing isolation between each of these systems.
There are two categories of hypervisors: enterprise and embedded. As the names suggest, enterprise hypervisors run on enterprise servers, while embedded hypervisors are used on devices with embedded software (like vehicles). Software complexity is continuously increasing as users expect more features for safety and comfort needs. To meet these market demands, vehicle OEMs need to include a hardware platform with high computing resources, similar to a server platform. To accommodate the growing software needs in vehicles, the features of both categories of hypervisors will need to converge as automotive hardware platforms become similar to enterprise servers.
Why are Hypervisors Important?
Hypervisors have been used on servers for a long time to minimize the number of individual servers needed for each operating system. But there are many other features of hypervisors that have made them so appealing:
- Allowing selective access to device peripherals and hardware resources
- Scheduling computing resources to each of the VMs
- Providing isolation to VMs for security
- Securing inter-VM communication
Enterprise hypervisors are used on enterprise servers with high compute resources and have more control of VMs and resources. These are feature rich, but high in footprint.
By comparison, embedded hypervisors are used on embedded systems where resources are constrained (e.g. industrial equipment, medical devices, mobile devices, and automobiles). Embedded hypervisors have real-time behavior, are small in memory footprint, and have less interrupt latency.
Enterprise and Embedded Hypervisors Need to Converge
In the automotive sector, customers expect a vehicle to stay relevant for many years and enjoy up-to-date features for quite some time. The growth of software defined vehicles – where software enables a large number of functions – has made these expectations possible. However, for this potential to be fully realized, OEMs need to use high-computing hardware platforms that can accommodate future software needs. And, as network speeds increase due to 5G/6G, the trend to dynamically offload software functions to edge computing and cloud platforms will increase, further expanding the use of hypervisors. The high computing hardware platforms could be dual SoC (System on Chip) or quad SoC. Based on the load on each of the SoCs, hypervisors could move functionality from one SoC to another or to edge or cloud platforms. And, in the case of failures, redundancy of VMs will be needed to ensure continuity of service.
To accommodate these future demands, hypervisors will need to support dynamic addition/removal of VMs, change the configuration with respect to compute resources, and migrate VMs from one SoC to another. Considering that enterprise hypervisors already have these features, the next logical step is to bring enterprise hypervisor functionality to embedded platforms based on Advanced RISC Machines (ARM) and RISC-V (apart from x86), optimizing the footprint, providing real time performance, and meeting the latency requirements of automotive systems. The figure below explains the factors/features of a converged version of the hypervisor.