ModelingNon-functionalApplicationDomain

2016-05-31 20:20:15   0  举报





仅支持查看

AI智能生成

非功能应用域（Non-functional Application Domain）是指软件系统在性能、可靠性、安全性等方面的属性和特征。这些属性通常不直接与系统的特定功能相关联，而是关注于系统的整体运行效果和用户体验。在建模非功能应用域时，我们需要考虑诸如响应时间、吞吐量、可用性、容错性、可扩展性等关键指标，以确保软件系统能够满足预期的性能要求和用户需求。此外，还需要关注数据完整性、访问控制、审计跟踪等方面的安全性需求，以确保系统在各种环境下的稳定运行。通过有效地建模非功能应用域，我们可以更好地评估和优化软件系统的性能和可靠性，从而提高整体的产品质量和用户满意度。

作者其他创作

大纲/内容

Background & Motivation

1. Service robotics as a science of integration relies on the combination of individual expertise from various stakeholders involved in the overall development of a robotics software system

2. Recent robotics research trends mainly address isolated capabilities on functional level.

3.Model-based robotics approaches such as RTC already facilitate the description of functional components however system integration is currently not systematically addressed in robotics system development.

4.Some issues existing in model-based approaches beyond robotics such as OMG MARTE and SysML,offering concepts for describing the execution and interaction behavior of components on system level

central concepts are often hidden

concepts of model level are too fine granular

robotics expert is left alone with many system-level design choices, while application domain experts need to understand all the low-level technical details of the functional components

1.The authors of this paper pursue the opposed freedomfrom-choice approach by consciously restricting the modeling choices to the crucial concepts and abstractions that are necessary to systematically design and integrate functional components.

Contibution

Pay special attention to non-functional system-level aspects

Provide model-based mechanisms for robotics experts to clearly define activation semantics for the concurrent execution of functional blocks within components

Motivating Example

Navigation scenario with two basic robot capabilities: the fast, local obstacle avoidance (inner loop) and the slower grid-map-based path planning (outer loop).

Conclution of example:mentioned system properties need to be adjustable not the result of unmodifiable decisions inside of component implementations

Ecore Meta-Model Definition

For component definition: simplified from fledged robotics (meta)-models to focus on essential concepts for efficiency and clarity reasons,also extended them with additional concepts to address concerns related to inteconnected components with the aforementioned cause-effect chains

For System configuration:main concern is the initialization of ComponentInstances and the definition of initial Connections between In- and OutPorts

Design and development of a model editor

Graphical DSL for component definition

A textual DSL for system configuration

Model-to-text transformation

Conclusion & future work

This paper provides a meta-model which clearly separates different concerns from component developers and system integrators.

The presented work provides a contribution for making the step from function-driven coding towards model-driven system integration