TL;DR
This paper introduces SEPRS, a system designed to automate the complex process of configuring application systems from pre-existing primitive components. By adapting the UM-PRS reactive planning system, SEPRS aims to streamline the selection and integration of components, addressing the manual and tedious nature of current approaches.
What Problem Does This Paper Address?
Developing complex software systems often involves assembling them from a collection of pre-built components. The Scenario-based Engineering Process (SEP) is a methodology that aims to simplify this by grouping primitive components into application-specific ones. However, the current process of selecting these primitive components and defining how they interact is a highly manual and time-consuming task. This manual effort becomes a significant bottleneck, especially as systems grow in complexity, user requirements change dynamically, and the need to manage diverse domains increases. The paper addresses the challenge of automating this configuration process, making it more efficient and adaptable.
Methods
The core of SEPRS is built upon the University of Michigan's Procedural Reasoning System (UM-PRS), a reactive reasoning and planning system. SEPRS adapts UM-PRS's architecture by treating primitive components as the "plans" that UM-PRS would normally use. The system's interpreter uses application requirements as goals to guide the selection of these primitive components. Components that have already been selected and integrated are stored in an "in-process area." The interpreter activates relevant primitive components based on the intention structure, which then passes the chosen component to a "component integrator." This integrator utilizes Adaptive Semantic Language techniques to build the necessary interfaces and messages for integrating the component into the application architecture. The system also incorporates a "component monitor" that acts as a system engineer, feeding modifications back into the process.
Datasets
Not discussed in the paper.
Key Insights and Ideas
- Component Integration as Planning: The paper frames component selection and integration as a planning problem, leveraging existing reactive planning systems like UM-PRS. This is a novel application of planning techniques beyond traditional task execution.
- Dynamic Reconfiguration: SEPRS is designed to handle dynamic changes in user requirements. Modifications can be interjected at any point in the architecture creation cycle, potentially leading to immediate deselection of components and extensive reconfiguration.
- Extensibility: The system is designed to be extensible, allowing for the addition of new primitive components as technologies evolve, ensuring the system can adapt to future advancements.
- Manual Grouping for Marketing: The paper acknowledges that while component selection can be automated, the grouping of primitive components into higher-level application components may remain a manual task, driven by factors like marketing purposes.
Key Results
The paper describes the architecture and expected functionality of SEPRS rather than presenting quantitative results from a deployed system. The key outcome is the proposed system itself, SEPRS, which is expected to fit well within existing system engineering environments that include scenario modeling, object-oriented analysis, and simulation. The authors anticipate that SEPRS will automate the tedious manual undertaking of component selection and integration.
Limitations and Future Work
The paper acknowledges that the grouping of primitive components into application-specific components remains a manual task, as this can be influenced by various viewpoints, including marketing. Future work would likely involve further development and evaluation of SEPRS, potentially exploring more sophisticated methods for automated component grouping and demonstrating its effectiveness on real-world complex systems. The paper also implies that the system's ability to handle extensive reconfigurations due to dynamic requirements will be a key area for future investigation and refinement.
Summary updated