Automating system-level data-interchange software through a system interface description language.

摘要

Today's platforms, such as full mission simulators (FMSs), exhibit an unprecedented level of hardware and software system integration. In this context, system integrators face heterogeneous system interfaces which need to be aligned and interconnected together in order to deliver a platform's intended capabilities. The sole aspect of the data systems exchange is problematic ranging from data misalignment up to multi-architecture environments over varying kinds of communication protocols. Similar challenges are also faced by integrators when interoperating multiple platforms together through distributed simulation environments where each platform can be seen as a system with its own distinct interface. On the other hand, enabling system reuse across multiple platforms for product line support is challenging for system suppliers, as they need to adapt system interfaces to heterogeneous platforms therefore facing similar challenges as integrators. Furthermore, the introduction of system interface changes in order to respond to late business needs, or unforeseen performance constraints for instance, is even more arduous as impacts are challenging to predict and their effect are often found late into the integration process.;Consequently, this thesis tackles the need to simplify system integration and interoperability in order to reduce their associated costs and increase their effectiveness along with their efficiency. It is meant to bring new advances in the fields of system integration and system interoperability. Notably, by establishing a common taxonomy, and by increasing the understanding of system interfaces, the various aspects impacting system data exchanges, multi-architecture environment considerations, and the factors enabling interface governance as well as system reuse. To this end, two research objectives have been formulated.;The first objective aims at defining a language used to describe system interfaces and the various aspects surrounding their data exchanges. Therefore, three key aspects are studied relating to system interfaces: the relevant language elements used to describe them, modeling system interfaces with the language, and capturing multi-architecture considerations.;The second objective aims at defining a method to automate the software responsible for system data exchanges as a way of simplifying the tasks involved in system integration and interoperability. Therefore, model compilers and code generation techniques are studied.;Consequently, this thesis tackles the need to simplify system integration and interoperability in order to reduce their associated costs and increase their effectiveness along with their efficiency. It is meant to bring new advances in the fields of system integration and system interoperability. Notably, by establishing a common taxonomy, and by increasing the understanding of system interfaces, the various aspects impacting system data exchanges, multi-architecture environment considerations, and the factors enabling interface governance as well as system reuse. To this end, two research objectives have been formulated.;The demonstration of these objectives brings new advances in the state of the art of system integration and system interoperability. Notably, this culminates in a novel system interface description language, SIDL, used to capture system interfaces and the various aspects surrounding their data exchanges, as well as a new method for automating the system-level datainterchange software from system interfaces captured in this language.;The advent of SIDL also contributes a new taxonomy providing a comprehensive perspective over system interoperability as well as a common language which can be shared amongst stakeholders, such as integrators, suppliers, and system experts. Being architecture-agnostic, SIDL provides a single architectural viewpoint overseeing all system interfaces and capturing multi-architecture considerations which was never achieved prior to this work. Furthermore, a SIDL code generator is introduced which has the novelty of generating the data-interchange software from a richer pool of information, notably from the high-level system relationships down to the low-level protocol and encoding details. Because multi-architecture considerations are captured natively in SIDL, this enables the code generator to be architecture-agnostic making it reusable in other contexts.;This thesis also paves the way for future research building upon its contributions. It even proposes a vision for software application development with the end goal being to push further the boundaries of simplifying and automating the tasks involved in system integration and interoperability.