Design

New Architecture for Mobile News Agent System

Description
New Architecture for Mobile News Agent System
Categories
Published
of 21
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  New Architecture for Mobile News Agent System A. M. Riad Head of Information Systems DepartmentFaculty of Computers and Information SystemsMansoura University, EGYPTamriad2000@yahoo.com H. A. El-Ghareeb Department of Information SystemsFaculty of Computers and Information SystemsMansoura University, EGYPThelghareeb@mans.edu.eg Abstract: Mobile information agents are mobile agents that retrieve information to end users.This paper presents a new architecture for Mobile News Agent System (MNAS) that is basedon a mobile information agents to retrieve news to users from different news servers . Severalapproaches for implementing MNAS are illustrated and comparative analysis is performed toidentify the most suitable one. MNAS components are:  Mobile News Agent  that retrieves newsto user,  Directory Server  that is the central manager server, and  News Servers that holddifferent news categories and contents. MNAS provides the client an easy to use interface tospecify preferred news categories, a client application that initiates the mobile news agentwhich heads directly to the directory server  to retrieves list of  news servers that contain newsmatches user preferences,   and searches all news servers and collects news to finally return toclient. MNAS proves that mobile information agent technology can not just be used in allinformation retrieval systems, because under some conditions, it might loose some of itsadvantages. MNAS tries to find out the most appropriate information retrieval systemsconditions to implement mobile information agent technology. Keywords:  Mobile Information Agents,  News Server, Information Retrieval, Mobile Architecture, Distributed Processing.  1-2-3-  . 1  1. Introduction Agent is a software system that is situated in some environment and is capable of autonomousactions in this environment in order to meet its design objectives. Mobile agent is an executing program that can migrate during execution from machine to machine in a heterogeneousnetwork. Major characteristics of agents are: autonomy, interactivity, reactivity, proactivity,and mobility [13]. Autonomy means that agent can make its own decisions. Reactivity meansthat agent can react to changes in environment. Proactivity means that agent should not waitfor changes in the environment to take actions; instead it initiates its actions to start affectingthe environment. Interactivity takes place when more than one agent cooperates to achievecertain task that is the summation of all sub tasks performed by all agents involved in thesystem. Mobility happens when agent can transport itself across different systems architecturesand platforms. Information agent is an autonomous computational software entity that hasaccess to one or more heterogeneous and geographically distributed information sources, andwhich is able to proactively acquire, mediate, and maintain relevant information on behalf of its users or other agents preferably justintime [1]. Mobile information agent is one of the evolving technologies that proved lot of success though lot of limitations. The employment of mobile agents has been particularly attractive in many applications like query processingwithin widely distributed heterogeneous environments (for retrieving information amongdifferent sites), network management, electronic commerce, mobile learning, and mobilecomputing [48]. Retrieving information from distributed database systems is an essentialrequisite in nowadays distributed systems. Generally, it can be said that information retrievaland data collection are the most important requirements in the information technology.Mobile agent systems provide new perspectives for distributed ecommerce applications. SeaofData (SoD) applications are those that need to process huge quantities of distributed data.They present specific restrictions, which make mobile agent systems one of the most feasibletechnologies to implement them. A Mobile Agentbased Reservation System, MARS, was presented in [9]. MARS aimed to implement an online reservation system that uses mobileagents to determine what best option for the user is and reserve a ticket based on user  preferences. An architecture for implementing a multiagent system within the context of alearning environment is introduced in [5]. The system implements website, database, virtualclassroom, and agents. Intelligent agents roles in elearning systems were presented. Twoapproaches for information retrieval based on mobile agents are presented in [3]. The firstapproach utilizes the mobility of agent for moving the query to the desired site where the dataresided while the second one is based on reduction of the number of migrating agents. Mobileagents have been chosen for their intrinsic properties of autonomy, synchronicity, dynamicityof distribution, and adaptability to available system resources in VM (Virtual museum) byusing the internet as the network infrastructure to make available heterogeneous museuminformation, from pictures, figures and maps to texts, audio and animated images [6]. Withinany information network, users have many short term goals of information retrieval. Short termgoals are constituted as finding information in as little time as possible based on a specificquery. In recent years, peertopeer based information retrieval systems have raised significantinterests among computer scientists. A peertopeer based information retrieval system consistsof a set of nodes connected in a peertopeer fashion [78]. Mobile agents can be used as acomponent in clustering architecture for routing in mobile ad hoc networks [10]. This paper tends to build a mobile news agent system, presenting different approaches for mobile newsagent system, and clarifies differences in performance among them. While doing so,advantages and disadvantages of mobile news agents will be discussed.2  1.1 Importance of mobile information agent There are many arguments with mobile information agents that advance them to other technologies. Mobile information agents are able to travel autonomously through the internet.Such agents enable dynamic load balancing in large scale networks, reduction of data transfer among information servers, and migration of small business logic within mediumrangecorporate intranets on demand [1]. Mobile information agent advantages include that [2]:1. It can invoke operations, retrieving and processing data, locally, eliminating the transfer of intermediate results, so, conserve bandwidth, reduce latencies, and responds to user actionsrapidly.2. It can continue execution even if the link between client and server goes down. This, because it executes on different machine(s) than the one initiated it, so, if the agent is alreadyinitiated, and is currently executing at another machine, and the link goes down, user,theoretically, should be able to retrieve results once the link is back. Besides, user gained timeof offline processing.3. It can choose different migration strategies depending on its task and current network conditions. This can be considered as an intelligent aspect of mobile information agents.Though distributed processing environments depend on certain predetermined route, mobileinformation agents can be intelligent and overcome this constraint by adapting its execution plan according to task and/or network condition.4. It allows traditional clients and servers to offload work to each other, and to change whooffloads to whom according to the capabilities and current load of the client, server, andnetwork.5. It can serve in distributed environments. More than one mobile information agent can beinitiated at the same time, and task can be subdivided on them. In this case, task becomes thesummation of all subtasks that constitutes it. In this case, certain aspects of management shouldtake place.There has been a renewed interest in mobile agent technology due to the continued exponentialgrowth of Internet applications, the establishment of open standards for these applications, aswell as the semantic web developments. Building mobile agent systems is a challenge becausefew guidelines exist on the topic and also because exhaustive testing is virtually impossible.The many (rather recent) Internet services, applications and technologies (e.g. Web Servicesand XML) are playing a major role in the further exploration of mobile agents as a viableInternet technology [11]. Mobile agents are programs that can migrate from host to host in anetwork of computers, at times and to places of their own choice. Unlike applets, both the codeand the execution state (heap and stack) move with the agent; unlike processes in processmigration systems, mobile agents move when and where they choose. A mobileagent programmer, thus, has an option not available to the programmer of a traditional distributedapplication: to move the code to the data, rather than moving the data to the code. In manysituations, moving the code may be faster, if the agent’s state is smaller than the data thatwould be moved. Or, it may be more reliable, since the application is only vulnerable tonetwork disconnection during the agent transfer, not during the interaction with the resource[12, 13].This paper presents several approaches for implementing Mobile News Agent System. It is amobile information agent based system that aims to retrieve news to users. Here comes thename " mobile news agent  ", because mobile information agent retrieves news to users as anexample of sort of information. The organization of this paper is as follows: In section 2, problem definition is presented. Section 3 includes development framework, components,analysis, design, and implementation of MNAS. Section 4 focuses on different approaches that3  can be used in MNAS. Section 5 illustrates measuring performance of each approach in acomparative analysis. Section 6, illustrates the results of comparative analysis. Conclusion andfuture work are presented in section 7. 1.2 Mobile Information Agents in Distributed Systems Distributed Environments must provide minimal level of objectives to be flexible and effective.Though mobile information agents are not distributed database systems, the fact that bothdistributed mobile information agents and distributed database systems share the sameconcepts can not be ignored. The fundamental principal of a distributed database system is tothe user, system should look exactly like a non-distributed system and this implied twelveobjectives [14]. Seven objectives of them will be applied to mobile information agents asdistributed system and one more will be adapted to fit the system. Objectives like  Fragmentation Independence, Distributed Query Processing, Distributed Transaction Management, DBMS independence will not be discussed because they are pure databasemanagement system topics. The eight objectives are: 1.   Local autonomy : autonomy should be considered for both mobile information agent, andhosting site. Autonomy is a main characteristic of mobile information agents, and it should be built in by default. Though mobile information agents still can not start themselves, they stilltake decisions in performing actions autonomously. Mobile information agent decides when,where to move, and how to perform tasks at hosting sites. Mobile information agent is aware of different destinations, and it determines the suitable next destination. Mobile information agentrecognizes different hosts’ types, and detects suitable actions to be performed at each host.Autonomy of hosting site lies within host capability to authenticate agents in order to present asystem security feature. 2.   No reliance on a central site : There should be no central controlling site. Disadvantages of central sites include single point of failure, and bottle neck. Unfortunately this objective has to be broken because mobile information agent needs central sites to serve as distributed processing systems. mobile information agent should be able to clone, that is replicate itself toall destined sites, or spawn, that is generate children to perform subtasks at destined sites, andthe sum of all mobile information agents subtasks satisfies the task. Generating replicas or children requires tracking, and summing subtasks. Tracking and summing can be performedeither at client machine, or at a central dedicated server, or at any host. 3.   Continuous operation : Mobile information agent's dead end destinations should becarefully managed. Dead ends might result due to communication disorder or site's failure. If amobile information agent is destined to a dead end, this mobile information agent is lost. Deadends can be managed either centrally or by all participating servers. Local autonomy propertyof hosts enable them to decide whether mobile news agent was serialized successfully to thedesired host, and keeps trying if not. Parallel processing requires a central manager server toenable management to enforce continuous operation. 4.   Location independence : user should not be aware of site's locations and open channels, sothe vulnerability to outside attacks might be decreased as much as possible. This objective can be achieved using Server Aliases. Server aliases can be translated centrally or at all participating sites. Server aliases reflects news servers URIs stored in directory server  database. 5.   Hardware independence : mobile information agent should be able to operate on differenthardware platforms. This objective gives the mobile information agent ability to retrieve and process data with no limitations or restrictions. This can be achieved through hardwareisolating layer, that is, operating system.4  6.   Operating system independence : mobile information agent should be able to run ondifferent operating systems. The more operating systems mobile information agent can executeon, the more efficient and flexible the system is. Host independence can provide this objective. 7.   Network independence : mobile information agent has to merge smoothly among sites atdifferent networks. Operating Systems, at different levels, provide this functionality. The effectof different networks on mobile information agents is the responsibility of transport, network,and link layer components of network service model. It is not the responsibility of mobileinformation agent at all, though it is mentioned for the sake of completeness. 8.   Host independence : this is the adapted objective to fit mobile information systems. Host isthe always running application at each site that enables mobile information agent to perform itstasks. Host layer should act as an isolating layer from operating system, moreover, is able torun any received mobile information agent. 2. Problem Definition Assume that we have a homogenous distributed database system that includes horizontallyfragmented database relations. These relations contain data for news system. User can issue arequest for news based on one or more preferred categories. These relations are fragmentedamong different (n) sites. Relation schema is the following: TB = Project(ID, Category,DateTime, Content). There is a query submitted by a user at client machine requesting "Findthe News for Sports and Business categories". A mobile news agent will be initialized at client machine , heads toward directory server  to obtain information about news servers and collectsnews from them   and return news to user. MNAS   can perform this task using more than oneapproach. To decide which approach to implement, a comparative analysis between differentapproaches is needed. 3. Mobile News Agent System3.1 Development Framework  Any programming language that supports connection and serialization among machines issuitable for building mobile information agents. Mobile agent systems were built using C/C++,Java, and many other programming languages. .Net framework is an emerging Microsofttechnology that advanced programming in this century. Microsoft thinks of integration as thesolution to lots of problems. Microsoft .net framework was presented in year 2000 to form aframework that enables executing any .net compiled code on any machine that has .netframework installed on it. Examples of those machines are: PDA; personal digital assistants,tablet PCs, smart phones, and computers. Besides, .net framework is not a software developingtool, neither a programming language. .Net framework supports more than 30 programminglanguages, including Delphi, Visual Java, and Visual C++ [15]. By using .net framework,mobile information agents make a step toward host independence objective. .Net framework isthe isolating layer between different hardware, operating systems, and mobile informationagents. .Net framework provides the most efficient programming capabilities. Communicationand Serialization of objects among machines can never be made simpler in programming andimplementation than they are in .net. Choosing .net framework as the developmentenvironment was affected by two other factors: efficiency of Microsoft SQL Server 2005, andseamlessly efficiency of integration between it and .net framework. Microsoft SQL Server 2005 presents new vision for database management systems that worth respect. Finally, .netframework deals with standards efficiently. News should be retrieved to users in a standardformat allowing them to display results in a customizable environment. As news will be in theXML (eXtensible Markup Language: standard language for information interchange) format,5
Search
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks