Graph-based Re-design Algorithms for Embedded Systems

Background

The importance of embedded real-time systems for our daily life is rapidly increasing. One important field of application for embedded real-time systems is in the telecommunications industry. The complexity of these systems is rapidly growing while at the same time the software part is becoming more and more important. The E-CARES research project is a coopertaion between Ericsson Eurolab Deutschland GmbH (EED) and the Department of Computer Science III, RWTH Aachen. E-CARES is an acronym for Ericsson Communication ARchitecture for Embedded Systems.

Eighty percent of programming resources are allocated to maintaining and reengineering existing code. Our project concerns understanding and restructuring of complex legacy systems from the telecommunication domain. Such systems are real-time embedded systems using the signaling paradigm, thus they pose additional requirements regarding fault tolerance, reliability, availability, and response time.

Problem

In the first phase of our reengineering project we regarded the architecture modeling and the reverse engineering of telecommunication systems. The aim was to understand complex legacy systems. Current work concerns the restructuring of such systems including their redesign and reimplementation. The challenge is to provide concepts, languages, methods, and tools to improve the architecture of software, make software easier to understand, and help to find bugs.


After the source code of the given system was parsed and analysed, we obtain a graph representation of the software architecture. By using a redesign tool the engineer is able to interactively modify the system. This diploma/master thesis should examine the different possibilities of architecture redesign. In the first part different methods should be developed showing which architecture transformations are useful and how graph transformations can to used to achieve the goal. In the second part of the thesis, the student should find a convenient way to store information about the performed graph changes which later should be used by the unparser to perform the corresponding source code transformations.

The approach should consider different modeling languages, e.g. Progres and Fujaba.

Requirements

The following lectures are helpful:

• Einführung in die Softwaretechnik
• Compilerbau

Dipl.-Inform. Christof Mosler (Raum 4219)
Wei Zhao