Research on Dynamic Role Assignment and Federal Supporting Platform for Open Multi Agent System - CORE CORE Search Search Services Access to raw data API Dataset FastSync Content discovery Recommender Discovery Managing content Repository dashboard Support FAQs About About CORE Blog Contact us Research on Dynamic Role Assignment and Federal Supporting Platform for Open Multi Agent System By 汪红兵 Abstract 多主体系统已成为建模大型复杂分布式信息系统的一种理想范型。随着多主体系统研究的进展，开放多主体系统由于其更大的应用范围和强调对系统开放性的支持而受到越来越多研究者的关注。但是，目前对开放多主体系统的研究仍然处于初级阶段。本文以动态角色分配和联邦式支持平台为重点展开研究。 首先，使用Object-Z建立开放多主体系统的形式规范。引入角色分配管理者主体动态维护主体和角色之间的分配关系，解耦了传统面向主体软件工程方法中主体直接继承角色的静态分配关系。Object-Z的继承和组合等面向对象特征使得形式规范可以作为实际应用开发的起点，其推理规则和相关定理提供了对系统性质进行形式证明的基础。 第二，提出一个具有并行约束目标的开放多主体系统动态角色分配模型。引入多个主体共同承担角色分配的计算任务，避免因单个主体可能形成的计算瓶颈。基于并行约束目标结构图，给出目标集合划分算法。设计实现了动态角色分配算法，分析了算法的时间复杂度，实验研究了算法的执行时间，表明理论分析与实验结果一致。基于目标集合划分对角色分配的计算任务进行分割，使得各个角色分配管理者的计算结果无需进行合并再进行并行约束检查。 第三，提出一个开放多主体系统的联邦式支持平台OMAS_HLA(Open Multi Agent System based on High Level Architecture)，其基本思想是将不同主体系统封装成不同联邦成员加入联邦。特别以中间件RTI_JADE(Run Time Interface and Java Agent Development Environment)为例，讨论了主体对象管理组件和主体交互管理组件，实验比较分析了隐式通信和显式通信的性能，表明该中间件是非常高效的。基于OMAS_HLA可方便集成已有的各种异构主体系统，并可支持多种形式的应用。 最后，基于OMAS_HLA实现一种联邦式网络协作。提出一种悲观算法HOAM(HLA Ownership and Agent Multicast)，解决当前高层体系结构的接口实现软件不能支持网络协作所要求的操作或消息因果约束问题。实验比较分析了HOAM和LICRA(Lock-free Interactive Concurrency Resolution Algorithm)算法，表明就平均响应时间而言，悲观算法HOAM可以达到与乐观算法LICRA大约相同的性能。OMAS_HLA为网络虚拟装配等应用所要求的主体智能扩展奠定了基础。MAS (Multi Agent System) is rapidly emerging as a powerful paradigm for modeling and developing large, complex and distributed information systems. With the development of research on MAS, OMAS (Open Multi Agent System) has attracted more and more researchers’ attention for its emphasis on supporting openness and more application range. However, research on OMAS is currently at the preliminary stage. Dynamic role assignment and federal supporting platform are two research focuses. The formal specification of OMAS is firstly developed by using the modeling language Object-Z. RamAgent (Role assignment manager Agent) is introduced to be responsible for maintaining the assignment relationships between agents and roles in order to decouple the static inheritance in traditional Agent-Oriented Software Engineering (AOSE) methods. The formal specification based on Object-Z, a language with many object-oriented properties, can be considered as the starting point to build real agent systems. The inference rules and relative theorems provide the formal basis for proving system properties. Secondly, the model of dynamic role assignment for OMAS with parallel constraints among goals is proposed. Multi agents are introduced to take the computational task of role assignment in order to avoid the computational bottleneck caused by a single agent. The algorithm of goal partition is presented based on the goal structure diagram with parallel constraints and its time complexity is analyzed. The algorithm of role assignment is given and the run time is investigated by the two ways of theoretic analysis and experimental verification. It shows that the theoretic result is consistent with the experimental result. This goal partition in the model assures that the computational results produced by RamAgents can be directly incorporated without checking about parallel constraints. Thirdly, the federal supporting platform, OMAS_HLA (Open Multi Agent System based on High Level Architecture), is presented. The basic idea is to develop different wrappers for different agent systems, which are wrapped as different federates to join a federation. With the middleware RTI_JADE (Run Time Interface and Java Agent Development Environment) as an example, agent object management component and agent interaction management component are discussed. The performance of explicit communication and implicit communication between agents are comparatively investigated. It shows that RTI_JADE works efficiently. Different federates may be heterogeneous so the integration of different agent systems can be supported by the platform. Furthermore, a variety of applications can be based on the platform. Finally, the federal network collaboration based on OMAS_HLA is developed. The pessimistic algorithm HOAM (HLA Ownership and Agent Multicast) is proposed. The problem of being not satisfied with message causal order by the current RTIs (Run Time Interface) of HLA (High Level Architecture) is solved. The comparative analysis is made in terms of the response time of HOAM and LICRA (Lock-free Interactive Concurrency Resolution Algorithm). It shows that HOAM, as a simple pessimistic algorithm, has the approximate same performance with LICRA, as a complex optimistic algorithm. OMAS_HLA provides the basis for intelligent expansion required by the application of network virtual assembly Topics: 开放多主体系统, 动态角色分配, 高层体系结构, 联邦, 网络协作 Year: 2007 OAI identifier: oai:ir.iscas.ac.cn:311060/7652 Provided by: Institute Of Software, Chinese Academy Of Sciences Download PDF: Sorry, we are unable to provide the full text but you may find it at the following location(s): http://124.16.136.157/handle/3... (external link) Suggested articles To submit an update or takedown request for this paper, please submit an Update/Correction/Removal Request. Useful links Blog Services About CORE Contact us Cookies Privacy notice Writing about CORE? Discover our research outputs and cite our work. CORE is a not-for-profit service delivered by the Open University and Jisc.
