Integrating the analysis of multiple non-functional properties in the model-driven software engineering process

点击翻译按钮获取中文摘要

This proposal is about seamless integration of multiple formal analysis models for Non-Functional Properties (NFPs), such as performance, reliability, availability, fault-tolerance, scalability, security, maintainability, cost, etc., into the Model-Driven Engineering (MDE) process, in order to guide the design choices from an early stage and to insure that the system under construction will meet its non-functional requirements. Such analysis models (also known as quality models) are based on existing formalisms and tools (e.g., queueing networks, stochastic Petri nets, stochastic process algebras, Markov chains, fault trees, probabilistic time automata, etc.) and are automatically derived by model transformations from the software models built for development. Developing software systems that exhibit a good trade-off between multiple NFPs is hard because the design of the software under construction and its underlying platforms have a large number of degrees of freedom spanning a very large discontinuous design space that cannot be exhaustively explored. Another challenge is due to the fact that some NFPs are conflicting (for example, security and performance).***The long-term objective of the proposed research program is to add more "engineering" to model-driven software engineering by supporting seamless integration of the analysis of multiple NFPs into the MDE process. Different NFP analysis models based on appropriate existing formalisms can be automatically derived by model transformations from the software models built for development, as explored in previous research. The software and NFP analysis models must co-evolve together. An important research effort will go into investigating how multiple NFP analysis models can be used to find a good (preferably optimal) design solution, in which all non-functional requirements are met. Another important aspect of the proposed research is concerned with (semi)automating the software process tasks/activities related to NFP analysis, asking for human intervention only when the designers need to provide new information and/or their judgement or decisions are required. The intended purpose of such automation is two pronged: a) to raise the efficiency and usability of the NFP analysis during MDE by eliminating error-prone manual model manipulations, and b) to enhance the quality of the software products by verifying the NFPs throughout the development process, from its early phases.***The short-term objectives of the proposed research are as follows: a) Development of an ecosystem of modeling artifacts which can support synchronized co-evolution of the software and analysis models; b) Inter-model traceability between two models related by a given transformation; c) Metaheuristic approaches for multi-NFP optimization; d) Semi(automated) software process tasks related to NFP analysis.*****