分散型市场信息系统的开发【外文翻译】

　　外文翻译

　　原文

　　Elaborating a Decentralized Market Information System

　　Material Source: Polytechnic University of Catalonia

　　Author: Felix Freitag

　　Research Question:

　　The planned DMIS framework has to deal with several constraints both from dis- tributed environments, such as scalability and robustness constraints, and from information systems, like privacy, time-sensitivity and historical data availabil- ity. That leads to the first open question about the feasibility and integration of the required functionalities and models into one framework. This means not merging directly components since the influence between efficient routing combi- nation of models,

　　The research in this thesis should answer the following research question: Is the proposed framework flexibly and reusable in different Grid applications? The result to this question should be answered by validating the DMIS framework with an implementation for market-based Grid projects like Sorma and Grid4All.

　　How can the information retrieval be optimized, needs to be answered. The information type does not stay static during the application, hence each request requires a different structure or algorithm. Also the storing and retrieval of a large-scale information can be changing meanwhile the application is running. As the prediction of all possible requests would be cost intensive, a dynamic solution during the runtime of the application is required. To ensure an efficient routing with les

　　Another questions we aim to resolve concerns economic aspects. Is it possible to obtain better price margins once information about the market is available? In central auctions, trading agents with a simple negotiation intelligence need at least some basic information about the market. But in decentralized auctions or bargaining the availability of these values can often not be ensured. The main question is how the economic benefit in form of competitive advantage can be optimized by using the i

　　Different, New and Better Solutions

　　Only a few researches target a distributed aggregation and a publish-subscribe model in one framework: Willow and DSIMS. Willow combines both approaches in one protocol, which has the trade-off of an effective routing and the design complexity by merging the two approaches. A more important difference is that it focuses on the protocol itself and its effective routing algorithms. Our work will go further in several aspects of privacy, complex queries and adaptive structures.

　　The most related work is the DSIMS that also aggregates information about large-scale network systems. This work focus on the aggregation of information considering scalability behavior, privacy, robustness and flexibility. A difference to our work is that there is no specific focus on time-sensitive data that can explicitly be addressed. Moreover, in our work the privacy model will be extended and adapted to economic information aspects.

　　Until now no known work investigates on economic aspects of an applied decentralized information or monitoring system. Therefore, required parameters of general on-line markets as well as Grid markets will be characterized within our thesis. They will be analyzed with a developed DMIS.

　　The economic literature has discussed about the value of information in mar- kets such as auctions. However, these aspects are still open and are rather the- oretical. Our work aims to execute simulations and prototype runs to extract the effort of provided information. These values obtained in simulations can be compared with existing theoretical approaches. Afterwards, its realization in a real environment will be shown and compared with an implemented prototype.

　　Another distinction is the possibility to integrate the DMIS into several Grid market applications. This integration will show the flexibility and the feasibility of integration into an existing market middleware, into a user interface as front end application and into different DHTs substrates.

　　译文

　　分散型市场信息系统的开发

　　资料来源：Polytechnic University of Catalonia 作者：Felix Freitag

　　研究问题：

　　在分散型环境中计划的DMIS软件框架，无论是处理数据的可扩展性和鲁棒性，都存在一些制约因素和限制，并会影响信息系统的比如隐私权，数据的敏感性和历史数据的可用性。这首先导致系统的可行性及所需功能和模型集成的框架成了一个悬而未决的问题。但这意味着不合理，因为它们之间有间接的联系，这个联系可以影响整个的组件模型，扩展性和灵活性，方便性，需进行评估。如果对这个问题进行研究，确定主要部件和它们的依赖，就可以得到一个精心设计的架构。本论文的研究的问题包括：如何扩展系统构架的灵活性，使之可以使用在不同的应用系统中，对这个问题的应该基于像Sorma和Grid4All DMIS系统一样通过市场的验证。

　　怎样才能优化信息检索，这个问题需要进行深入讨论。信息类型不应该集成在静态的应用程序中，因此每个请求都需要一个不同的结构或算法。还要求储存的和大规模的信息检索可以改变，并且可以同时在应用程序运行。由于预测所有可能的请求将使成本增加，所以在应用程序需要运行需要提供动态的解决方案。为了保证在部署期间的DMIS软件运行时能够稳定运行，需要深入市场进行调查。

　　另一个问题，我们的目标是通过信息系统来提高利润。是否提供更多有关市场信息就可能获得更好的价格和利润？在中部地区的拍卖行，使用一个简单的交易代理的谈判情报也会要求提供一些有关市场的基本信息。但在拍卖或讨价还价下，这些价值可往往得不到保证。主要的问题如何通过使用信息系统的优化在竞争优势中形成经济利益。

　　自我架构对于内部元件和外部框架的灵活性是一个主要的要求。这种集成能力将验证使用到至少两个不同的中间件DMIS软件，像电网市场CATNETS和Grid4All项目的DMIS软件。一个成功的整合将被批准时，请求和响应信息的选择性设置将会被所有的外部应用程序的组合提供。以后定稿的内容就是原型分析和内部流程优化了。

　　不同的解决方案的对比：

　　只有少数的研究目标将分散、聚集和发布订阅在一个框架模型：Willow and DSIMS。Willow将两种方法结合在一个协议中，而权衡一个路由有效性是通过合并这两种设计方法的复杂性。更重要的区别是，它的协议本身和它的重点是有效的路由算法。我们的工作将进一步在隐私性，复杂的查询和自适应结构这几个方面。

　　最重要的工作是，如何将几个大型的DMIS信息系统集成在一个商场中。集成工作的重点是考验资料的可扩展性，隐私性，鲁棒性和灵活性。相对我们的工作，不同的是，到底有没有对时间敏感的数据，是否可以明确具体重点加以解决。此外，我们的工作将隐私模式推广和适应到经济信息方面。

　　到现在为止，我们还没有发现任何人和单位去探讨如何应用分散的信息和监测系统经济方面的工作。因此，我们的论文最需要的参数将是在一线市场以及电网市场。我们将结合已发展DMIS软件进行分析。

　　三月的经济文献讨论内容比如拍卖kets信息的价值。但是，这些方面仍然是属于开放的模型和理论体系。我们的工作目标是努力执行模拟和原型并将信息提取出来。在模拟环境中提取这些值可以比现有的理论方法更有效。然后，我们把它实现在一个真实环境中去，与一个实现原型进行对比分析。

　　另一个区别是融入一些电网市场应用的DMIS软件。这种集成将显示软件的灵活性，能否与前端应用程序的用户界面和DHT方法彻底融合，就是实施到到现有的市场可行性。