An embedded system is a compact computing system designed to perform one or a few dedicated functions which usually operates in real-time conditions. Embedded systems are integrated into industrial machinery, cars, vessels, trains, planes, satellites, medical and scientific equipment. They represent a global market of € 200 billion per year and constitute a strategic sector for the EU.
A more robust embedded system which requires less space and energy consumption and has shorter response times than a conventional computer. Nevertheless, these advantages are diluted in high performance equipment, such as those used in the medical and aerospace industry, where it is necessary to control various hardware devices simultaneously with applications and even different operating systems. In these cases, the most common course of action would be to integrate several microprocessors in the system; one for each task or set of tasks, which increases the cost, necessary space, energy consumption and heat generated.
An alternative solution is to use virtualisation techniques. These techniques, when applied to a microprocessor, allow the creation of virtual partitions that, in practice, operate as independent microprocessors. Virtualisation is rather widespread in conventional computers, but not so much in embedded systems, a research area that is one of the priority lines of the Institute of Automation and Industrial Computing (ai2). Specifically, the ai2 has a long history in the development of hypervisors, a virtualisation technique that provides particularly solid and efficient results.
As a result of the work developed over the years in the optimisation of high-performance embedded systems, in 2008 the ai2 team developed the open source hypervisor “Xtratum” for the French National Centre for Space Studies (CNES) which is a reference in the development of embedded systems for satellites.
Currently, together with the company EADS Astrium, the ai2 develops an evolved version of the Xtratum hypervisor in order to optimise it to operate on the LEON 3 processor used in the new generation of satellites of the European Space Agency (ESA). In addition, ai2 works alongside ESA to adapt Xtratum to the LEON 4 multiprocessor, designed for higher performance satellites.