New Paradigms in Embedded Systems
Teknik Naturvetenskap
2011-09-02 11:53, reviderat 2011-09-02 12:11
A large part of the Swedish IT industry is concerned with the
development and production of embedded systems, including industrial
robots and systems for process automation, telephone exchanges, cars
and trucks, or systems tailored to the railway sector.
The multicore revolution we are currently experiencing gives rise to a
number of opportunities for embedded systems, ranging from the
availability of high-performance and power-efficient hardware to the
possibility to consolidate previously separate systems into a single
system with multiple processing cores. These potential benefits cannot
be exploited, however, without solving major technical challenges
introduced by multicore processors: predictability, reliability,
real-time guarantees, energy-aware design, and isolation between
applications of different criticality levels, which are all aspects
that are significantly more complicated on multicore than on
singlecore processors. Addressing such issues will require a variety
of new research results, both on a basic and the applied level. One
of the urgent challenges is the migration of legacy embedded software
to modern platforms with multi-processing units. Since embedded
software often consists of very hardware-dependent low-level code, a
change of processor architecture can incur substantial costs, which
become multiplied due to aspects of concurrency on multicore systems.
This challenge can be partly addressed by improved development,
simulation, and analysis tools for embedded software. In the long run,
it will additionally be necessary to devise better programming
paradigms for parallelised software, as well as programming languages
and frameworks that expose features of parallel hardware in a safer
and more comfortable way to software developers.
Besides the increasing integration of systems on a single platform,
there is also the trend of connecting previously separate systems in
networks, thus creating complex, distributed systems with dynamically
changing topology. The distributedness and pervasive character of such
"Systems of systems" give rise to new challenges for both industry and
academia. For this direction, we refer to the "Systems of systems"
description.
(Written by Philipp Ruemmer and Wang Yi)
Inga kommentarer inlagda.
Fasen är stängd, inte möjligt att kommentera.