A Software Component Technology for Vehicle Control Systems
Mikael Åkerholm · 2005
Software is fantastic! Numereous modern high-tech products incorporate software. This thesis focus on software for vehicles. As high-tech products, modern vehicles contain much software that provides advanced functionality, e.g., efficient engine control, anti-spin systems, and adaptive-cruise control. However, software engineering is not without problems! Software can contain errors, is often delivered later than promised and the cost of its development constitutes a major part of the total development cost of the vehicle. These problems are consequences of the complexity of the systems we build with software, in combination with the immaturity of an evolving discipline. We therefore need improved approaches to software engineering. Component-based software engineering is a promising approach. It is analogous with approaches used in other engineering domains. For examples, mechanical engineers build systems using well-specified components such as nuts and bolts, and the building industry uses components as large as walls and roofs (in turn assembled from smaller components). It has proven to be effective in application domains such as desktop and web-applications. However, it has not yet been widely adopted for use in the development of vehicular software; one of the reasons being that the commercial component technologies are developed specifically for other domains and to support other types of applications. This research aims at developing a component technology for embedded control systems in vehicles. Such a technology would enable software engineers in the vehicular domain to make use of component-based software engineering. We have addressed questions concerning quality attributes, and how component-based applications should be built and modelled, in the context of vehicular systems. Furthermore, based on our answers we have implemented, and evaluated a prototype component technology in cooperation with industry. The results confirm the suitability of our prototype, but also show that it must be further developed if it is to meet wider industrial requirements.