【作者】 韦敏；

【导师】 朱跃龙；

【作者基本信息】 河海大学 ， 计算机软件与理论， 2006， 硕士

【摘要】 无线技术和移动计算设备的迅速发展,改变了传统的分布式计算。与固定网络计算不同,移动计算在无线环境中运行,其资源有限且可用资源经常变动,应用需要感知并适应环境变化以保证服务质量。利用中间件处理环境感知及自适应问题,可以有效的简化和便利移动应用的开发。 利用反射原理可以动态重配置中间件,是实现环境感知的有效方法,因为反射系统可以通过自省和调整来对自身进行修改。London大学的CARISMA平台采用了元数据和反射技术来实现移动计算中间件的环境感知。由于不同用户的服务质量需求不同,应用为满足用户需求,会以不同方式定制中间件,这些方式互相之间可能存在冲突。CARISMA利用微观经济学中的暗标拍卖法来选择执行策略,解决了策略冗余的问题,却不曾处理策略缺失的情况。 本文基于CARISMA平台,改进中间件的元数据,使之能更好的描述基层系统,细化策略冲突及其产生原因,提出追求最大用户满意度的让步协商机制,解决了因用户服务质量需求差异而导致的策略缺失问题。并且对策略选择中的标值算法进行改进,将策略提供的服务质量与用户对服务质量的要求进行比较,使所选择的策略能够更好地满足用户的服务质量需求;引入标值扣减反映执行代价消耗,简化计算。 在提出的策略选择机制的基础上,论文分别给出通信代价和计算代价最小的两种分布式算法,对两者进行比较分析,并实现此算法,说明让步协商机制的可行性和正确性。最后,给出一个环境感知的反射式移动应用中间件CARMAM的体系结构,并用一个在其上开发的具体应用实例来显示CARMAM为环境感知的移动应用开发提供的支持。更多还原

【Abstract】 The rapid growths of wireless technology, coupled with advances in mobile computing devices, have fundamentally changed the landscape of traditional distributed computing. Unlike fixed network computing, mobile computing running in a wireless environment suffers from limited resources and often experiences constant changes in the availability of resources, therefore, applications have to be aware of, and adapt to, variations in the execution context, in order to deliver a good quality of service to their users. It can simplify and facilitate the development of mobile application that using middleware to deal with the awareness of context and adaptation to it.Reflective principle, used to reconfigure middleware dynamically, is effective to achieve context-awareness, as a reflective system can modify itself through introspection and intercession. The CARISMA platform of London University is a mobile computing middleware, which exploits the principle of reflection to enhance the construction of adaptive and context-aware mobile applications. As different users may have different quality-of-service needs, and applications, in an attempt to fulfill these needs, may customize middleware behavior in conflicting ways. The CARISMA platform uses a microeconomic approach that relies on a particular type of sealed-bid auction to solve the problem of policy redundancy, while helpless to policy short.Based on CARISMA, this paper improves the metadata, application profile, to describe the baselevel more appropriately and refines the policy conflicts and the cause of them. To fulfill most satisfaction of every application involved, this paper proposes a concession mechanism to solve the problem of policy short. In addition, this paper modifies the algorithms of bid computing in CARISMA, compares QoS afforded by policy with the required QoS of user to make the elected policy satisfy the quality of service of different users, and imports minus bid to express running cost in order to simplify bid computing.According to the policy choice mechanism proposed above, this paper provide two different distributed algorithms to implement it, the first one that attempts to minimize communication costs, and the second one that aims to minimize computation costs. Then, this paper discusses their similarities and, mainly, their differences. More over, this paper implement the algorithms to argue the feasibility and validity of our concession mechanism. Finally, this paper describes the architecture of a Context-Aware Reflective Mobile Application Middleware, and provides an instance of a concrete application development on top of it to show its support for context-aware application development.更多还原