While life cycles of electronic components tend to become ever shorter, automotive suppliers need to keep up ECU supply for up to 30 years. This causes high storage costs or requires redesigning legacy ECUs in a way that obsolete components are replaced by newer ones. The legacy software problem is not specific to ECU software but is prevalent in virtually any real-time and non-real-time software field. A successful approach for non-real-time and soft real-time software is emulation, i.e. representing the original programming environment (e.g. instruction set architecture, device registers, application binary interface) on the target machine and instruction-wise interpreting the original machine code. Ideally, emulation is a black box approach in the sense that it does not require any knowledge about the legacy code. It is highly flexible but inherently inefficient which can be compensated by translating software such that it can be natively executed on the target machine. The redesigns of legacy ECU may cause tremendous software redevelopment efforts, e.g. to guarantee correct real-time behaviour. This paper presents a both academically and industrially challenging approach inspired by the legacy ECU software problem, namely static binary translation, which aims at automatic retargeting of real-time software. Eventually, the approach has the potential to significantly cut the redevelopment costs and facilitate the interchangeability of electronic components.
The legacy ECU software problem - Approach and research challenges
Das Altlastproblem 'Steuergerät-Programm' - Lösungs- und Forschungsherausforderung
2008
5 Seiten, 1 Bild, 5 Quellen
Aufsatz (Konferenz)
Englisch
CHALLENGES AND LESSONS LEARNED FROM RESURRECTING A LEGACY RESEARCH FLIGHT CONTROLLER
British Library Conference Proceedings | 2010
|Legacy Software Re-engineering Technology
British Library Conference Proceedings | 1996
|