Organizational Model and Dynamic Goal Model for Autonomous, Adaptive Systems

摘要

Systems are becoming more complex, in part due to increased customer requirements and the expectation that applications should be seamlessly integrated with other existing, often distributed applications and systems. In addition, there is an increasing demand for these complex systems to exhibit some type of intelligence as well. No longer is it 'good enough' to be able to access systems across the internet, but customers require that their systems know how to access data and systems, even in the face of unexpected events or failures. The goal of this research is to provide a framework for constructing complex, distributed systems that can autonomously adapt to their environment. Multiagent systems have become popular over the last few years for providing the basic notions that are applicable to this problem. A multiagent system uses teams of self-directed agents working together to achieve a common goal. Such multiagent teams are widely proposed as replacements for sophisticated, complex, and expensive stand-alone systems for similar applications. Multiagent teams tend to be more robust and, in many cases, more efficient (due to their ability to perform parallel actions) than single monolithic applications. In addition, the individual agents tend to be simpler to build, as they are built from a single agent's perspective. However, unpredictable application environments make multiagent teams susceptible to individual failures that can significantly reduce the ability of the team to accomplish its goal. To overcome this and other problems, the authors have developed a framework that allows the team to design its own organization at runtime. They propose to provide the team with organizational knowledge and let the team design its own organization based on the current goals and team capabilities. While the designer can provide guidance, supplying the team with key organizational information will allow the team to redesign, or reorganize, itself to match its scenario.