Agent Oriented Programming进展

1 引言近十年Agent和多Agent系统(MAS)的研究逐渐成为AI学科的热点之一。MAS中Agent是具有思维状态和交互能力的自治实体,彼此通过合作求解复杂问题,适合于动态开放环境。Agent的思维状态通常采用BDI模型,B(信念)、D(愿望)和I(意图)分别用模态算子给出,并在可能世界框架下给出其语义解释。这一模型语义明确直观,但实质上是一个计算资源无限的理想模型,因而在实际系统中,都采用限制和变通的方法来实现Agent,因此导致了所谓的理论脱离实践     

可移动Agent系统JMSAS中对象迁移的研究与实现

可移动软件Agent(Mobile Software Agent,MSA)是一种新的分布式计算模型,与传统的客户/服务器模型相比,在MSA计算环境中,计算实体不再是静止和被动的,而是能够自主迁移计算的Agent。MSA的计算方式就是将计算任务封装在Agent内部,通过Agent在多个位置迁移并与其他Agent交互和协作来完成这些任务。JMSAS系统是一个基于Java语言的供可移动Agent运行的平台。实现该系统的关键任务之一是实现Agent对象的迁移机制。     

一种新的基于复合弹簧网络的多Agent系统分布式问题求解方法

当多Agent系统（MAS）中Agent之间存在多种复杂的随机的社会交互行为时，当各Agent表现出不同程度的自治性和理性时，难以用现有的方法描述和求解MAS问题，即使对仅仅存在竞争和合作这两种社会交互行为，并且不考虑Agent之间自治程度的本质性差异时，现有的基于结盟的MAS问题求解算法也具有极高的计算复杂性，该文提出一种新的复合弹簧网络模型和方法，利用分布式弹性动力学方程，将MAS分布式问题求解过程转变对应的复合弹簧网络形变过程，这种模型和方法能够处理各种社会交互行为以及Agent不同程度的自治性，分析和仿真实验表明，在计算复杂性和适用性等许多方面，该文的分布并行算法优于文献[7,8]的Shehory-Kraus算法。     

基于组织的面向Agent 程序设计及其语言Oragent

当前,面向Agent程序设计在支持动态开放多Agent系统开发方面存在一系列的不足,如缺乏高层抽象、底层实现模型与高层设计模型相脱节、在支持系统动态性方面缺乏有效的运行机制和语言设施等.针对这些问题,提出一种基于组织的面向Agent程序设计方法.该方法将组织、Group、角色和Agent等高层抽象作为一阶实体,缩小了多Agent系统的设计模型与实现模型之间的概念鸿沟;借助于组织学中的机制——角色扮演机制、基于角色的交互——支持系统动态性的规约和实现,如Agent行为的动态组合、动态的交互等,基于该程序设计思想,设计了基于组织的面向Agent程序设计语言——Oragent,定义了其抽象语法和形式语义,并通过案例分析说明了如何基于该程序设计思想和Oragent语言来构造和实现动态而灵活的多Agent系统.     

基于网格的面向Agent可视化需求分析环境

面向Agent软件工程方法是开发复杂分布式系统的一种新范例,它在Agent抽象层次上建立软件系统的模型。多样的工具正在被设计、开发以推广面向Agent的需求建模方法。Aomg环境支持文献[1]中提出的一种面向Agent的分析和设计建模方法（Agent—Oriented Modeling based Grid,AOMG）,辅助基于网格环境的分布式应用的需求分析,建立可视化、直观的系统抽象模型（如系统组织模型、角色模型和Agent交互模型）,对之进行严格的XML描述,方便模型的一致性检查并为设计阶段提供支持。     

基于Agent技术的智能交通线控系统仿真环境研究

面向Agent技术综合和发展了分布技术和面向对象技术，适用于在功能或地理上分布的复杂自治系统的分析与设计。建立了基于Agent技术的智能交通线控系统仿真环境。结果表明，将Agent技术引入智能交通线控系统，可有效提高系统的动态环境适应性和自治性，降低系统设计与实现的复杂性。     

面向Agent的基本思想研究

提出面向Agem的基本思想是以社会学理论为原则构造和演化复杂软件系统。基于人的社会化、社会互动、社会组织理论论述了面向Agent关键抽象模型Agent、交互、组织的定义;基于社会变迁理论提出多Agent系统演化的目标是Agent和组织的自增长以满足需求的变化。从关键抽象模型和软件技术发展的三要素比较了面向Agent与面向对象的不同。     

一种基于Agent的自适应软件过程模型

传统的软件过程模型大多是静态的、机械的、被动的,它们要求软件工程人员在描述软件过程时预期所有可能发生的情况,并且显式地定义这些问题的解决方案.当软件过程所处的环境发生变化时,软件过程无法自适应地对这些变更作出相应的调整.提出了一种基于Agent的自适应软件过程模型.在这种软件过程模型中,软件过程被描述为一组相互独立而对等的实体--软件过程Agent.这些软件过程Agent能够对软件过程环境的变化主动地、自治地作出反应,动态地确定和变更其行为以实现软件开发的目标.     

基于生物网络的Web服务突现模型研究与实现

生物神经内分泌和免疫系统通过复杂的调节,构成了一个智能的生物网络.本文借鉴系统突现的自组织、自聚集和自协作特性及复杂的双向调节机制,提出了Web服务突现的研究方法.基于移动Agent设计具有免疫行为的生物实体并代理Web服务,服务请求是由突现形成的生物实体网络提供,生物实体通过自组织协作完成服务的动态合成和自主管理管理等一系列工作.Web服务突现模型由Java开发的功能模块和平台组成,并利用Linear Logic设计了实体的协商.     

一种基于组织学习的自适应Agent系统体系结构

在复杂适应系统和组织学习理论的基础上，探讨了一种基于组织学习的自适应Agent系统体系结构。在这种体系结构中，通过Agent交互中的适应性行为获取对系统复杂性的认知，而Agent的适应能力则依靠增强学习和动态自组织与重构来实现。文中给出了一个基于组织增长模型的分类器系统算法以及相应的软件实现技术途径。     

Intelligent distributed control systems

ContextThe paper² deals with distributed reconfigurable embedded control systems following the component-based International Industrial Standard IEC61499 in which a Function Block (abbreviated by FB) is an event-triggered software component owning data and a control application is a distributed network of Function Blocks. Nowadays, limited related works have been proposed to address particular cases of reconfigurations without considering distributed architectures. Our first problem is to be able to handle all possible forms of reconfigurations that can be applied at run-time to distributed Function Blocks. In this case, a coordination between devices of the execution environment should be applied to guarantee safe and coherent distributed reconfigurations. A second problem is to find the sufficient solutions for the correct implementation of this reconfigurable distributed architecture.ObjectiveThe paper defines an implementable multi-agent architecture for automatic and coherent reconfigurations of distributed Function Blocks.MethodTo address all possible industrial forms, we classify the reconfiguration scenarios into three levels. The first level deals with additions–removals of Function Blocks to-from the system’s implementation. The second deals with updates of compositions of blocks, and the third deals with updates of data. We define a Reconfiguration Agent for each device of the execution environment, and a unique Coordination Agent for coordinations between devices. Each Reconfiguration Agent to be modelled by nested state machines applies local reconfiguration scenarios in the corresponding device after coordinations with the Coordination Agent. We propose an Inter-Agents Communication Protocol to support correct and coherent reconfigurations of distributed devices. This protocol is based on Coordination Matrices to be handled by the Coordination Agent in order to define all reconfiguration scenarios that should be simultaneously applied in distributed devices. We propose XML-based implementations for both kinds of agents where XML code blocks are exchanged between devices to guarantee safety distributed reconfigurations. The contributions of the paper are applied to two Benchmark Production Systems available in our research laboratory.ResultsThe communication protocol is successfully applied to our platforms where simulations are executed to check distributed and coherent reconfiguration scenarios. The Reconfiguration and Coordination Agents are implemented in this platform by following the International Standard IEC61499. We show in addition XML-based successful interactions between devices when distributed reconfigurations are applied.ConclusionThe paper successfully defines a multi-agent architecture for IEC61499 distributed reconfigurable embedded systems where Coordination and Reconfiguration agents are proposed to allow feasible and coherent distributed reconfigurations by using a defined communication protocol. This architecture is implemented in XML and applied to real industrial platforms.     

Parallel reconfiguration in an image-processing context

Reconfiguration is considered in the context of image-processing applications. Dynamic switching is the basis of experiments into hardware reconfiguration described in the article. Implementation issues are detailed. Methods of analysing performance for the reconfigurable case are adumbrated. A discussion ensues on the practical problems of providing a usable system with reconfiguration provision. © 1998 John Wiley & Sons, Ltd.     

一种基于分布构件装配元信息动态重构的表示方法

动态重构技术使得软件系统自身进行演进的同时，仍然能够提供服务。本文提出了一种基于构件装配元信息的动态重构操作的表示方法。该表示方法以Jeff提出的动态重构系统设定为基础，采用集合来表达构件应用中的构件、连接子和节点，为动态重构规划的生成以及等价性验证等提供了量化方法，并在基于CORBA构件模型的StaCCM分布构件平台中加以实现。     

基于中间件的动态重配置容错对象管理框架

该文从容错QoS角度出发,为现有的基于中间件的容错对象管理框架增加自适应特性．首先把动态童配置定义成容错属性集（及其实现）、计算环境和资源变化描述、动态调整算法三者之间的互动关系,然后采用反射模型和发布／订阅模型建立了一个动态重配置管理框架,可以使系统在发现计算环境的变化、定制动态调整算法等多方面得到有力的支持,为系统有效实施动态重配置奠定了基础．     

RETRACTED ARTICLE: A multi-agent architectural solution for coherent distributed reconfigurations of function blocks

The paper deals with distributed Multi-Agent Reconfigurable Embedded Control Systems following the International Industrial Standard IEC61499 in which a Function Block (abbreviated by FB) is an event-triggered software component owning data and a control application is a network of distributed blocks that should satisfy functional and temporal properties according to user requirements. We define an architecture of reconfigurable multi-agent systems in which a Reconfiguration Agent is affected to each device of the execution environment to apply local reconfigurations, and a Coordination Agent is proposed for coordinations between devices in order to guarantee safe and adequate distributed reconfigurations. A Communication Protocol is proposed to handle coordinations between agents by using well-defined Coordination Matrices. We specify both reconfiguration agents to be modelled by nested state machines, and the Coordination Agent according to the formalism Net Condition-Event Systems (Abbreviated by NCES) which is an extension of Petri nets. To validate the whole architecture, we check by applying the model checker SESA in each device functional and temporal properties to be described according to the temporal logic “Computation Tree Logic”. We have also to check all possible coordinations between devices by verifying that whenever a reconfiguration is applied in a device, the Coordination Agent and other concerned devices react as described in user requirements. We present a tool applying simulations of this distributed architecture in order to check interactions and reactivities of agents. The paper’s contributions are applied to two Benchmark Production Systems available in our research laboratory.     

可重构密码处理结构中重构机制及对性能的影响

重构机制对可重构密码处理系统的性能有着重要的影响,该文从全局、局部、静态、动态几方面提出了流水化可重构密码处理结构中重构机制的分类,给出了各种机制的吞吐率和延迟公式,并分析了几种机制的性能和实现代价,最后给出了在采用局部动态重构机制的可重构密码处理结构中密码处理的性能。     

基于数据动态冗余的分布式并行系统重构机制

设计高可用分布式并行数据库系统,系统重构不能只局限于单节点状态变化,为此提出了基于数据动态冗余的分布式并行系统重构机制。该机制通过相关事件触发系统检测,及时分布采集节点状态信息,借助数据的再生和转移,在各节点并行完成负载失衡和节点状态变化下的系统自动重构。实际应用表明,该机制解决了多节点状态同时变化的系统资源和数据资源的重构问题,保证了数据库系统的高可用性。     

Bounded latency spanning tree reconfiguration

One of the main obstacles to the adoption of Ethernet technology in carrier-grade metropolitan and wide-area networks is the large recovery latency, in case of failure, due to spanning tree reconfiguration. In this paper we present a technique called Bounded Latency Spanning Tree Reconfiguration (BLSTR), which guarantees worst case recovery latency in the case of single faults by adopting a time-bounded bridge port reconfiguration mechanism and by eliminating the bandwidth-consuming station discovery phase that follows reconfiguration. BLSTR does not replace the Rapid and Multiple Spanning Tree reconfiguration protocols, which remain in control of network reconfiguration, whereas it operates in parallel with them.     

可重构技术及其在网络控制系统中的应用综述

可重构技术在对系统的维护、优化、扩展以及自适应、自组织、自恢复上有其独特的优势．对此,介绍了可重构技术的基本原理、实现方法和应用领域;阐述了其在网络控制系统中分别对系统、节点和协议的应用．并对其发展趋势进行了展望．