SGER: Analysis of Fundamental Control Mechanisms for Mobile Agents in Large Network Infrastructures

本提案的目标是对基于代理的系统的基本控制机制进行可行性研究，以促进网络资源管理和流量控制。主要的重点是协调多个智能移动代理的工作，这样所产生的协作就可以很好地随规模、资源可用性和基础设施的动态而扩展。提出的研究探讨了智能体群体的大小作为直接或间接描述基础设施属性的参数的函数的可行性。其基本思想是将网络中心功能从节点分离到移动子层。研究人员期望这种方法将产生一个更可扩展的架构，用于解决各种不同的以网络为中心的问题。为了促进这种方法，他们定义了一些基本机制，如代理克隆、代理合并、知识获取和代理迁移。正是通过使用这些机制，研究人员试图展示将智能移动代理范式集成到系统级任务中的可行性。提议的工作将涉及作为基础设施属性和单个代理的计算进展的函数的代理群体行为的正式分析研究。进一步的分析将基于精心构建的仿真实验和大型分布式网络环境的仿真。该建议有意忽略了基于智能体系统的几个问题，如知识表示、容错和安全性，而将重点放在基本机制的设计和分析上。我们认为，拟议项目的有限范围及其内在的探索性使其成为探索性研究小额赠款（SGER）计划的理想候选人。

The objective of this proposal is the feasibility study of fundamental control mechanisms foragent-based systems that facilitate network resource management and traffic control. The primaryemphasis is on orchestrating the work of multiple intelligent mobile agents such that the resultingcollaboration scales well with the size, the resource availability, and the dynamics of theinfrastructure. The proposed research investigates the feasibility of autonomously adapting thesize of the agent population as a function of parameters that describe properties of theinfrastructure directly or indirectly. The underlying idea is the separation of network centricfunctions from the nodes into a mobile sub-layer. The researchers expect that this approach will yield a morescalable architecture for solving a variety of different network centric problems. To facilitate thisapproach, thet have defined fundamental mechanisms such as agent cloning, agent merging,knowledge acquisition, and agent migration. It is through the use of these mechanisms that the researchersattempt to show the feasibility of integrating the intelligent mobile agent paradigm into system-leveltasks. The proposed work will involve a formal analytical study of the behavior of the agentpopulation as a function of properties of the infrastructure and computational progress ofindividual agents. The analysis will further be based on carefully constructed simulationexperiments and the emulation of a large distributed network environment. This proposal intentionally disregards several issues of agent-based systems, such as knowledgerepresentation, fault tolerance, and security, and focuses on the design and analysis of thefundamental mechanisms. We believe that the limited scope of the proposed project and itsintrinsic exploratory nature makes it an ideal candidate for the Small Grants for ExploratoryResearch (SGER) program.