Research Project: Trader
Within the TRADER project, we develop tools and techniques for improving the user-perceived reliability
of systems with embedded software, for the specific application domain of high-volume consumer electronics products. The tight constraints imposed by the market for these products, combined with the typical complexity of the systems create a unique situation, where it is not uncommon that products are released with several known software faults outstanding. The overall goal is to reduce the extent to which such faults manifest themselves to the user.
In this context, we currently apply source code analysis
to help developers focus their attention on those potential faults that are most likely to affect the users, and we apply fault diagnosis
to improve the overall efficiency of the debugging phase in the software development process.
Source code analysis
Static software checking tools are an important tool for creating reliable code, but in many cases, the large number of warnings generated by these tools prevents their practical use. Through automated source code analysis we try to identify those parts of a software system where an error would have the highest probability of propagating to a system failure, which would be noted by the user. One approach to this problem is to estimate the likelihood that a particular piece of code is executed. This information can then be used by developers to prioritize the advice generated by software checking tools. Because of the tight release deadlines for consumer electronics products, this is particularly relevant for the development of embedded software in this area.
Locating a software fault is an important step in actually solving it, and an automated diagnosis of errors detected during system testing can help the developers locate the responsible faults, thus allowing that more bugs can be solved in limited time. Our research focusses on a specific diagnosis technique, where we record a so-called program spectrum
for every scenario in a test suite. The recorded spectra tell us which parts of a system are active during the various scenarios, and we try to identify those parts whose activities appear to correlate most with the detected errors. Because the technique has a relatively small overhead with respect to CPU time and memory requirements, it is well suited for the resource-constrained environment that is typical for the development of consumer electronics products.
TRADER is a large project with industrial and academic partners, both from the hardware and software field. As a testcase, we focus on digital televisions produced by NXP (formerly known as Philips Semiconductors). The project is led by the
Embedded Systems Institute
in Eindhoven, which offers also the TRADER management summary
. The project is partly funded by the Dutch Government.
The following partners collaborate in TRADER:
NXP, Philips Research Laboratories, Philips TASS,
Design Technology Institute (a joint research institute of the
Eindhoven University of Technology and the National University of
Singapore), Delft University of Technology, Twente University,
University of Leiden, and the Embedded Systems Institute.
Presently, we have no
vacancies for PhDs or PostDocs in the Trader project but SERG
master students are welcome to discuss possibilities for doing their thesis assigment in the context of the project.
For more information on the Trader project contact Arjan van Gemund
Oops! Page Not Found
The server returned a 404 response.
- Did you type the URL?
- You may have typed the address (URL) incorrectly. Check it to make sure you've got the exact right spelling, capitalization, etc.
- Did you follow a link from somewhere else at this site?
- If you reached this page from another part of this site, please email us at [email] so we can correct our mistake.
- Did you follow a link from another site?
- Links from other sites can sometimes be outdated or misspelled. Email us at [email] where you came from and we can try to contact the other site in order to fix the problem.
- What's next