Using Architecture Analysis to Evolve Complex Industrial Systems

  • Authors:
  • Tommy Kettu;Eckhard Kruse;Magnus Larsson;Goran Mustapic

  • Affiliations:
  • Industrial Software Systems, ABB, Corporate Research, Västerås, Sweden 72178;Industrial Software Systems, ABB, Corporate Research, Ladenburg, Germany 68526;Industrial Software Systems, ABB, Corporate Research, Västerås, Sweden 72178 and Industrial Software Systems, Mälardalen University, Västerås, Sweden 72123;Industrial Software Systems, ABB, Corporate Research, Västerås, Sweden 72178

  • Venue:
  • Architecting Dependable Systems V
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

ABB is a large industrial company with a broad product portfolio that contains products that can be categorized as highly complex industrial systems. Software embedded in complex industrial systems must support rigid system dependability requirements. It is not only a challenge to design and implement these systems as dependable, but it is also difficult to maintain this important property over time. There are several factors that make software evolution a challenging task, such as: size of the software base is measured in order of MLOC, products are long-lived and extended to support new requirements over time longer than 10 years. Because of personnel turnover important knowledge is lost from time to time, and the only artifact that is really up-to-date is the implementation itself. Therefore, to obtain an up-to-date view of the system and prevent expensive mistakes during system evolution, it is beneficial to find practical ways to obtain an up-to-date view on an architectural level without having to read thousands of lines of source code. These activities should be seen as an important contribution for preventing the introduction of faults into software systems since they contribute to improve and maintain the overall system dependability. This experience paper provides practical advices on how to reconstruct the architecture of existing systems by combining the use of tools and the existing knowledge within the organization. The paper is based on experiences from two cases in different sub domains within industrial automation.