Real-time virtualisation takes foothold in medical applications

30 July 2010

How do you reduce system costs through hardware consolidation? Gerd Lammers and Christian Eder explore a new option.

Figure 1

Medical systems such as ultrasound devices or patient monitors often need a sophisticated user interface in addition to hard real-time features. In general, this is realised with a real-time operating system on a microcontroller and a separate Microsoft Windows based embedded PC.

Thanks to virtualisation technology (VT) it is now possible to run both operating systems independently on a single embedded PC. The new Intel CPU features for virtualisation ensure that the parallel operation of the different operating systems is secure. This type of hardware consolidation is now also an option for embedded CPU modules with Intel Nehalem processors.

In medical devices it is critical that the real-time tasks of data collection and control run seamlessly and without any unexpected interruptions. These tasks should therefore be shielded from the visualisation software, network connectivity or other Windows applications. Traditional, so-called Windows Real-Time Extensions cannot meet these requirements, as the operation of the real-time part is handled by the Windows operating system. Faulty drivers, system components or viruses and other malware can compromise safe system operation or even lead to a complete system failure. The use of hardware-supported virtualisation technology is necessary to enable multiple operating systems to run safely in parallel.

Conventional virtualisation solutions often use Intel VT technologies to shield guest operating systems from each other, but this affects the real-time features of the system. These solutions generally also do not allow direct hardware access which is necessary for data control or acquisition.

RTS Hypervisor by German company Real-Time Systems GmbH meets all requirements by the medical device manufacturers in one product. In this example, the individual CPUs of a Nehalem processor (i3, i5 and i7) are exclusively assigned to the individual operating systems and separated via Intel VT technology. At the same time direct hardware access is granted to the real-time operating system to avoid any interference with its properties.

In this way, RTS Hypervisor does not affect the interrupt latencies, jitter or the I/O performance. The available memory is assigned to the individual operating systems on an exclusive basis.

Hardware devices, regardless of whether these are PCI or legacy devices, are configured so that interrupts are distributed exclusively and directly among the individual cores and operating systems. Cores, devices, I/O ports and memory, which are not assigned to a particular operating system, are hidden from that operating system. Because interrupts are assigned directly and each operating system is granted access to its own hardware devices no special or modified device drivers are required.

Figure 2

With RTS Hypervisor it is also possible to determine the boot order; individual processor cores or operating systems can be booted, re-booted or shut down independently from each other. None of the installed operating systems performs tasks assigned to other operating systems nor do they fulfill any of the functions of the Hypervisor itself. It is irrelevant whether the system runs multiple instances of the same real-time operating system or a mix of general purpose operating systems (GPOS) such as Windows with different real-time operating systems.

To allow communication between different virtual systems RTS Hypervisor offers the optional creation of shared memory and a virtual, TCP/IP-based internal network. The use of multiple operating systems on multicore processor architectures is a logical step in embedded systems design. It lowers total hardware costs while at the same time increasing system reliability and performance. With multicore technology, as used for instance in an Intel Core i7 processor embedded in computer modules, it is possible to reach a new dimension of modularity and flexibility in medical applications.

Hypervisor technology can be used on a variety of hardware platforms. The Z530 Intel Atom processor is ideal for very low-power or battery-powered devices. While this processor has only one core, thanks to hyperthreading there are two virtual processors available to share the computing power.

At the upper end of the performance spectrum there is the Intel Core i7 processor. The conga-BM57, a high-end COM Express module, uses an Intel Core i7-620M processor with two cores, and so four threads. Despite 4MB cache and 2.66 GHz clock rate, which can be increased up to 3.33 GHz in Turbo mode, this module rarely reaches the 30 Watt mark in practical use. The associated heat can also be realised in closed systems – which are a common requirement in the medical sector.

The integrated graphics controller supports the Intel Flexible Display Interface (FDI), thereby offering two independent video channels with one VGA, one LVDS, three HDMI or DisplayPort, or one SDVO interface. Five PCI Express lanes, eight USB 2.0 ports, three SATA, one EIDE plus one Gigabit Ethernet interface allow flexible system extensions with high data bandwidth. Additional features include fan control, an LPC bus for slow expansion, and the Intel High Definition Audio interface.

A compact design (95 x 125 mm²) coupled with low power consumption and high performance ensure that the conga-BM57 can easily be integrated in medical devices. These computer modules have been successfully tested with the new RTS Hypervisor from Real-Time Systems. With this extremely compact embedded PC module it is possible to run up to four random operating systems in parallel, in hard real-time and completely independently of each other.

Figure 3

Microsoft Windows, Windows CE, QNX Neutrino, Linux, Pre-Emptive Linux, VxWorks, Microware OS-9 and Ontime RTOS-32 are currently supported. Operating system vendors can port any additional operating systems onto the RTS Hypervisor platform themselves at any time.

With Hypervisor technology allowing the use of one embedded computer module for two independent applications, overall hardware costs of a system solution can be reduced by up to 50%. For medical applications, this means that the necessary real-time applications, as well as the visualisation and the user interface can run in parallel on the same system. Two previously separate systems are thus brought together on one hardware platform without affecting the functionality and without changes to the software.

The combination of embedded computer modules with the RTS Hypervisor creates a universal platform which is suitable for almost all medical applications. The seamless integration of multiple operating systems and applications on a single computer system reduces costs and increases flexibility. As an added benefit, virtualisation technology increases the reliability of the overall system, because hardware components become redundant, power consumption and heat decrease and the number of connectors and cables is reduced.

Gerd Lammers is Director of Real-Time Systems GmbH; and Christian Eder is Sales & Marketing Manager for congatec AG

Figure1. With Real-Time Hypervisor operating systems can directly access the allocated hardware resources.

Figure 2. A COM Express module with an Intel Core i7 processor is mounted on a customised carrier board.

Figure 3. COM Express modules measure 95 x 125 mm² and are easy to integrate. They have many interfaces and can simultaneously run up to four operating systems in real-time.


Contact Details and Archive...

Related Articles...

Most Viewed Articles...

Print this page | E-mail this page