A Study on Email-based Mobile Agent Runtime Environment

全文

(1)國立交通大學電機資訊學院資訊科學系博士論文植基於電子郵件的行動代理人執行環境之研究 A Study on Email-based Mobile Agent Runtime Environment. 研究生. ：何敏煌. 指導教授. ：袁賢銘博士. 中華民國九十三年六月.

(2) 植基於電子郵件的行動代理人執行環境之研究 A Study on Email-based Mobile Agent Runtime Environment 研究生：何敏煌. Student:. 指導教授：袁賢銘博士. Advisor: Dr. Shyan-Ming Yuan. Min-Huang Ho. 國立交通大學電機資訊學院資訊科學系博士論文. A Dissertation Submitted to Department of Computer and Information Science College of Electrical Engineering and Computer Science National Chiao Tung University in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Computer and Information Science June 2004 Hsinchu, Taiwan, Republic of China. 中華民國九十三年六月.

(3) 植基於電子郵件的行動代理人執行環境之研究研究生：何敏煌. 指導教授：袁賢銘博士. 國立交通大學資訊科學研究所摘. 要. 現今，電腦硬體和軟體技術在網際網路的應用上其發展非常地快速。應用這些新的軟硬體技術在資訊擷取都已經獲得了相當大的成功。因為在網路上從事資訊擷取的應用非常地重要，在此方面的研究已經有相當多的成果。然而，目前這些資訊擷取方面的研究成果，主要大多是假設其使用的作業平台都具有穩定且高速的網路環境和強大的運算能力，對於那些網路連線速度慢且不穩定、運算能力薄弱且電源供應和相關資源都缺乏的行動運算設備和環境來說，要應用這些研究成果是非常困難且不容易的。本論文的目的即在於廣泛地研究這一類資源缺乏的行動計算環境在資訊擷取及相關技術方面的問題，並提出一個可行的架構以解決這些問題。在論文中，我們考慮所有移動裝置使用者在從事資訊擷取時可能遇到的狀況，然後提出一個建構於有限資源計算能力和低品質網路移動式設備上的工作架構，讓使用者可以在此類網路上擷取出他所需要的有用資訊。在我們的設計中，使用者可以輕易地利用他能上網的移動設備在區域網路及網際網路上擷取他有興趣的資訊，而這些移動式設備可以包含行動電話、PDA、甚至是掌上型的筆記型電腦。由於這些裝置在操作行為模式及介面上非常地多樣化，所以我們使用標準的電子郵件通訊協定讓使用者可以用一致性的操作介面在區域網路及網際網路中擷取資訊。電子郵件是所有可以上網的裝置中最受歡迎的網路服務之一，使得電子郵件的協定和其客戶端操作軟體在不同的可上網裝置中具有一致性。也因此，透過電子郵件來做為資訊擷取的媒介也將被一致化。也就是說，使用者可以利用一致性的介面和操作方式在我們設計的工作架構下設計、和操作一些在伺服器端的可移動式的行動代理程式，而這些行動代理程式即可在具有良好網路環境及高速計算能力的遠端伺服器中為使用者高速地擷取資訊及分析資料，不用去擔心他現在使用的是那一種行動裝置。.

(4) A Study on Email-based Mobile Agent Runtime Environment Student: Min-Huang Ho. Advisor: Dr. Shyan-Ming Yuan. Department of Computer and Information Science National Chiao Tung University. ABSTRACT Currently, the technologies of hardware and software about the Internet and information processing are developed in a very high pace. Applying these new technologies into the area of information retrieval is very successful.. Because of the importance of information. retrieval via the network, there are lots of prior studies dedicated in these related topics, and they have already got plenty of successful outcomes, but most of those researches always assume that the platforms by means of those technologies have good quality networks and high computing power CPU.. For those poor quality networks, low computing power CPU,. and resource-constrained mobile devices, obviously, most of the previous research outcomes usually can’t be easily adopted. The objective of this dissertation is to study such resource-constrained mobile computing circumstances for information retrieval and related applications, and to propose a feasible framework to solve such situation. In this dissertation, we take care all situations that a mobile user may run into, and then propose a framework for the mobile devices with low-computing-power and poor-quality-networks to retrieval the information he is interested in from the network. In our design, a mobile user can retrieve the information from the network or the Internet via his/her networked mobile device. include mobile phones, PDA, or mini-notebook PCs.. The mobile devices may. Because the interface and operating. model of those mobile devices are very diverse, to uniform the user interface among these devices, we utilize e-mail protocol and mobile agent technologies for the user to retrieve the information from the network or the Internet. Because the e-mail is one of the most popular applications for all networked devices, and the e-mail protocol and e-mail client operation are common and uniform in those networked devices, the information retrieval operating interface will be unified. That is, the user may utilize the unified interface of our proposed framework to design, operate, and execute some ii.

(5) mobile agents in the server-side, and those mobile agents will retrieve the information for the user in a very good computing environment. The user doesn’t need to worry about what kind of the mobile device he/she will use in the whole information retrieval process.. Keyword: Mobile Agent, Information Retrieval, Mobile Computing, E-mail. iii.

(6) 誌. 謝. 能夠順利取得博士學位，要感謝的人實在太多了，首先是我的指導教授袁賢銘博士，在漫長的研究過程中，如果沒有您全力的指導和充份的信任，要完成博士論文及相關研究工作，對我來說幾乎是不可能的任務。我的良師益友，台北大學教授張玉山博士，由於你適時的指引和關心，對於我的論文研究方向及內容不斷地加以斧正，讓我的博士研究內容得以更加地充實且切合要旨。當然，我也要感謝交大資科所各位教過我的教授們，在我研究的過程中給我很大的幫助，尤其是楊維邦館長和孫春在教授，一路從論文計畫審查、校內口試到學位考試，都能適時給我重要的指引，並給予鼓勵和建議，讓我的論文研究內容更加地實用，且不至於蕪蔓龐雜。施仁忠系主任在校內口試時給予的建議，適切地指正了論文部份的缺漏，讓我能夠及時地補正。中興大學資管所朱延平系主任和成功大學資工所鄭憲宗教授不辭辛勞地幫我審閱博士論文，並於學位考試時給我建議和鼓勵，在此一併向你們致上最高的敬意和謝意。從台南每個星期通車到新竹求學研究的漫長過程中，多虧了分散式系統實驗室的各位學長學弟們的協助，才得以安然渡過研究過程的各個難關。焦信達、葉秉哲、許瑞愷、邱繼弘、蕭存喻、鄭明俊、高子漢、吳瑞祥、孫文駿、及其他的學弟們，謝謝你們。我的妻子－美堯，從在大學的時候認識妳，我就一直是個只知道讀書的窮學生，如今，總算得以結束學生的身份，在家裡專心當個好老公了。回頭想想這段走過的路，如果沒有聰慧的妳從旁協助及打理，今天我可能還是一個留在台北獨自過生活的傻小子，謝謝妳。女兒小晴，兒子小暘，你們這兩個寶貝蛋，爸爸終於可以專心地陪你們了，從你們出生到現在，你們的爸爸都好忙好忙，現在，爸爸可是你們兩個人專用的喔！我的母親張玉英女士，希望我這樣一點點的努力成果能夠稍稍彌補您在我身上的用心；媽媽，您辛苦了。最後，謹將本論文獻給已故的父親，沒能讓您親自參加我的畢業典禮，是我這一生中最大的遺憾。. iv.

(7) Contents Abstract in Chinese..............................................................................................i Abstract in English .............................................................................................ii Acknowledge in Chinese ...................................................................................iv Contents ...............................................................................................................v List of Figures.....................................................................................................ix List of Tables ................................................................................................... xiii. Chapter 1 Introduction .................................................................................. 1-1 1.1 Problem basics......................................................................................................1-1 1.2 Objectives and our contributions..........................................................................1-7 1.3 The challenges in our research .............................................................................1-8 1.4 Organization of this dissertation...........................................................................1-9. Chapter 2 Problem Description..................................................................... 2-1 2.1 Information retrieval on the resource-constraint mobile devices.........................2-1 2.2 Styles of information gathering ............................................................................2-5 2.3 Classification of information retrieval models.....................................................2-6 2.4 Basic concept of e-mail based information retrieval..........................................2-10 2.5 Classification of e-mail based information retrieval ..........................................2-13 2.6 Applying mobile agent runtime mechanism to our design.................................2-16. Chapter 3 Background and Related Works ................................................. 3-1 3.1 Background and terminologies.............................................................................3-1 3.1.1 Mobile phones ...............................................................................................3-1 v.

(8) 3.1.2 WAP (Wireless Application Protocol) ...........................................................3-7 3.1.3 SMS/EMS/MMS of GSM .............................................................................3-8 3.1.4 i-Mode ...........................................................................................................3-9 3.1.5 PDA and PC.................................................................................................3-10 3.1.6 Mobile Execution Environment (MExE) ....................................................3-11 3.1.7 Information retrieval....................................................................................3-11 3.1.8 Mobile agents ..............................................................................................3-13 3.2 Related works .....................................................................................................3-14 3.2.1 AGORA .......................................................................................................3-14 3.2.2 Info-On-Demand .........................................................................................3-15 3.2.3 iSMS............................................................................................................3-20 3.2.4 AvantGo.......................................................................................................3-21 3.2.5 Unified interface in integrated information retrieval ..................................3-22 3.2.6 Mobile agent runtime environment in mobile computing...........................3-24 3.2.7 Current commercial products for remote information accessing ................3-24. Chapter 4 Mobile Agent and Integrated Information Retrieval................ 4-1 4.1 Mobile agent concepts and their advantages on mobile computing environment4-1 4.2 Migration for e-mail based mobile agent across mobile devices .........................4-8 4.3 The idea of hybrid mobile agent runtime environment ......................................4-12 4.4 A unified interface for integrating information retrieval ....................................4-14 4.5 Agents technologies for information retrieval by applications ..........................4-19 4.6 Automatic wrapper generator for integrated IR .................................................4-24 4.7 Overview of IBM Aglets ....................................................................................4-33. Chapter 5 System Design and Implementation ........................................... 5-1. vi.

(9) 5.1 System overview ..................................................................................................5-1 5.2 System workflow..................................................................................................5-4 5.3 System architecture ..............................................................................................5-6 5.4 Design Issues........................................................................................................5-9 5.5 System implementation ......................................................................................5-15 5.5.1 System implementation basics ....................................................................5-15 5.5.2 System server deployment...........................................................................5-16 5.5.3 System components and services supports in e-MARE..............................5-19 5.5.4 Commands supported in e-MARE ..............................................................5-22 5.5.5 Communication mechanism of e-MARE ....................................................5-23 5.5.6 Result-Slicing mechanism of e-MARE.......................................................5-26 5.6 Detail workflow of e-MARE..............................................................................5-28 5.7 Instruction set .....................................................................................................5-30. Chapter 6 Experiments and Examples ......................................................... 6-1 6.1 Software and hardware details .............................................................................6-1 6.2 The measurements for composing the operational e-mail....................................6-2 6.3 The steps of the mobile users for accessing e-MARE..........................................6-3 6.4 An introduction to web interface of e-MARE ......................................................6-3 6.5 An introduction to WAP interface of e-MARE ..................................................6-13 6.6 Examples of e-MARE to access the information on the Internet.......................6-16 6.6.1 Example I: Querying the available search engines of e-MARE .................6-16 6.6.2 Example II: Querying keywords via several search engines.......................6-18 6.6.3 Example III: Fetching any web page on the Internet ..................................6-25 6.6.4 Example IV: Executing a program in execution server of e-MARE...........6-29. vii.

(10) 6.6.5 Example V: Parallel and distributed execution programs in e-MARE........6-38 6.6.6 Example VI: Executing mobile agents in e-MARE ....................................6-39 6.7 Discussions about transferring bytes and time comparisons..............................6-40. Chapter 7 Applications................................................................................... 7-1 7.1 Information retrieval on the Internet ....................................................................7-1 7.2 Information retrieval from various information sources ......................................7-9 7.3 Remote displaying via e-MARE framework......................................................7-13. Chapter 8 Conclusions ................................................................................... 8-1. viii.

(11) List of Figures Chapter 1 Introduction Figure 1.1: The ratio of the Internet among cable, wireless, and wireline .............................1-3 Figure 1.2: The trend of Internet users ...................................................................................1-3 Figure 1.3: The model of traditional static agent for doing some user’s delegated works.....1-5 Figure 1.4: The model of mobile agent for doing the same works in Figure 1.3 ...................1-5. Chapter 2 Problem Description Figure 2.1: The basic concept of meta-search technologies...................................................2-2 Figure 2.2: The unified user interface for the mobile device users ........................................2-3 Figure 2.3: The model of Traditional Web-based Information Retrieval Service ..................2-7 Figure 2.4: The model of E-Paper Style Information Retrieval Service ................................2-8 Figure 2.5: E-mail based information retrieval model .........................................................2-10. Chapter 3 Background and Related Works Figure 3.1: The architecture of Info-On-Demand. ...............................................................3-16 Figure 3.2: The dialog for a simple query e-mail of Info-On-Demand................................3-17 Figure 3.3: The setting and register dialog of Info-On-Demand..........................................3-18 Figure 3.4: The fetched results from the Info-On-Demand system......................................3-19 Figure 3.5: The basic architecture of iSMS..........................................................................3-20 Figure 3.6: The typical workflow of meta-search services approach in ref [18] .................3-22 Figure 3.7: The role of a semi-automatic wrapper generator ...............................................3-23. Chapter 4 Mobile Agent and Integrated Information Retrieval Figure 4.1: The basic concept of the migration of mobile agent............................................4-3 Figure 4.2: The ideal mechanism for mobile agent migration ...............................................4-3 Figure 4.3: The popular event-based migration mechanism ..................................................4-4 ix.

(12) Figure 4.4: E-mail based migration mechanism.....................................................................4-5 Figure 4.5: The distributed model of e-mail based mobile agent migrating mechanism .......4-9 Figure 4.6: The centralized model of e-mail based mobile agent migrating mechanism.....4-11 Figure 4.7: The hybrid architecture between on-line and off-line MARE ...........................4-13 Figure 4.8: Architecture of IIR .............................................................................................4-17 Figure 4.9: The preliminary architecture of supporting personalized search.......................4-21 Figure 4.10: Personal information filtering agent ................................................................4-23 Figure 4.11: Basic architecture of XML-based wrapper ......................................................4-26 Figure 4.12: An overview of the wrapper generator ............................................................4-29 Figure 4.13: The workflow of the proposed wrapper generation.........................................4-31 Figure 4.14: Architecture of automatic wrapper generator ..................................................4-32 Figure 4.15: The architecture of Aglet Viewer .....................................................................4-34 Figure 4.16: The infrastructure of Aglets .............................................................................4-35. Chapter 5 System Design and Implementation Figure 5.1: The basic infrastructure of our proposed e-MARE framework ...........................5-2 Figure 5.2: Basic system workflow of e-MARE....................................................................5-5 Figure 5.3: The basic system architecture of e-MARE ..........................................................5-6 Figure 5.4: The server deployment of e-MARE...................................................................5-18 Figure 5.5: The system components of e-MARE.................................................................5-20 Figure 5.6: The first web page of BlackBoard System of e-MARE ....................................5-24 Figure 5.7: User’s personal information page ......................................................................5-26. Chapter 6 Experiments and Examples Figure 6.1: The first page of e-MARE web site .....................................................................6-4 Figure 6.2: The user login page..............................................................................................6-5. x.

(13) Figure 6.3: The page after the user logged into the system....................................................6-6 Figure 6.4: The available links of the user named ‘minhuang’..............................................6-7 Figure 6.5: The list of available search engines on the web page ..........................................6-8 Figure 6.6: The querying results of entering the keyword ‘T65’ in Yahoo ............................6-9 Figure 6.7: The available remote programs displaying page. ..............................................6-10 Figure 6.8: Uploading the remote programs into runner’s home directory..........................6-12 Figure 6.9: The screen snapshot of listed available programs for user ‘minhuang’.............6-13 Figure 6.10: The main page of the WAP interface of e-MARE ...........................................6-14 Figure 6.11: The WWW interface for ‘e-MARE Go!’ ......................................................... 6.15 Figure 6.12: The shortcut of ‘e-MARE Go!’........................................................................6-15 Figure 6.13: The messages after user’s selection for executing the 5th shortcut ..................6-16 Figure 6.14: The results after user’s selection for executing the 5th shortcut.......................6-16 Figure 6.15: The menu item to enable the SMS funcation...................................................6-16 Figure 6.16: The webpage of http://www.whitehouse.gov...................................................6-28 Figure 6.17: The screen snapshot of http://news.yam.com/general .....................................6-31 Figure 6.18: The results after executing the rtnews.php ......................................................6-33 Figure 6.19: Installing some programs at the home directory on the operating system.......6-35 Figure 6.20: The screen of index.php...................................................................................6-36 Figure 6.21: The concise version of the fiancé news on the Internet ...................................6-37 Figure 6.22: The cron table of our example .........................................................................6-38. Chapter 7 Applications Figure 7.1: http://stock.kimo.com.tw .....................................................................................7-2 Figure 7.2: http://stock.yam.com............................................................................................7-2 Figure 7.3: All stock list of the company about automobile ..................................................7-3. xi.

(14) Figure 7.4: The screen of querying the stock id 2330 ............................................................7-6 Figure 7.5: The screen of http://tw.dictionary.yahoo.com/.....................................................7-8 Figure 7.6: An example of teacher’s timetable of high school...............................................7-9 Figure 7.7: The executing results of command ‘fetchTable 11’...........................................7-12 Figure 7.8: The basic architecture of remote displaying technology ...................................7-14 Figure 7.9: The main webpage of http://www.twvs.tnc.edu.tw/remote ...............................7-15 Figure 7.10: The screen snapshot after the system has received the command, &&RDON(1) ..............................................................................................................................................7-15 Figure 7.11: The screen snapshot after the &&RDC commands has issued........................7-19 Figure 7.12: Utilizing the &&RDL commands to show the URL list in PDP .....................7-20. xii.

(15) List of Tables Table 2.1: The comparison between mobile devices and PC .................................................2-4 Table 4.1: Some querying examples of integrating information retrieval approach ............4-18 Table 6.1: The measurements for composing a simple operational e-mail ............................6-2 Table 6.2: The five distinct typical querying operations ......................................................6-41 Table 6.3: File size and time consuming comparison...........................................................6-42 Table 7.1: The mean of &&RDC(id, 0, n) commands .........................................................7-16 Table 7.2: The table structure of actSite ...............................................................................7-17 Table 8.1: The comparison between e-MARE and other related works.................................8-1. xiii.

(16) Chapter 1 Introduction This chapter gives an introduction about the trend and the vision of the mobile computing environment, the idea and efforts in our researches will also be presented here. At the first, we point out the developing trend of the information retrieval technologies and mobile computing technologies nowadays, and then focus the encountered problems for the mobile users who are interested in retrieving the information on the network via their mobile devices. To solve the mentioned problems, we propose a novel and feasible solution.. A brief. introduction of our approach will be given in section 1.2. In section 1.3, we also list the current challenges in the related topics, and these issues will be solved in the following chapters of this dissertation. The final section of this chapter is a description about the organization of this dissertation.. 1.1 Problem basics The number of documents on the Internet is very tremendous nowadays, and the document numbers are also growing in a very fast speed. Thus, the Internet has become a very important information source for every information retriever or even every computer users. However, not only the volume of documents on the Internet is very huge, but also the formats of those documents are very diverse. Furthermore, because of lacking the proving process by editors, most of the documents on the Internet do not have a well structure, and those documents always lack a specified compact and precise description.. Consequently, the. technologies dealing with efficiently retrieving the information from the great amount diverse unstructured documents on the Internet are very urgent for the people who want to get more comprehensive and precise results from it. For approximately 4000 years ago, people had already known to organize the information for later retrieval and usage, but today, facing to those almost countless documents, we need some useful and efficient technologies to help us.. Information retrieval (IR) deals with the. representation, storage, organization of, and access to information items. 1-1. The representation.

(17) and organization of the information items should provide the user with easy access to the information in which he is interested. In the computer centric view, the IR Problem consists mainly of building up efficient indexes, processing user queries with high performance, and developing ranking algorithms which improve the quality of the answer set.[1] Traditionally, libraries are the major information sources which the IR researches focused on.. The digitized. documents of digital libraries’ management system have well defined structure and semantics. Thus, it is easy and efficient for an IR client to get the user’s interested information from such well structured information sources. However, the documents of the libraries are not the only information source for the people to gather. As mentioned earlier, the information sources from the Internet or private local area network are more and more important nowadays.. Thus,. the related issues accompanied with retrieving the documents from the Internet have to be envisaged and be solved. Currently, given the ever-increasing scale and diversity of information and applications on the Internet, improving the technology of information retrieval on the Internet is an urgent research objective.. The documents retrieved from the Internet are either semi-structured or. unstructured in format and its sources are extremely heterogeneous.. In consequence, the task. of efficiently gathering and extracting information from documents can be both difficult and tedious.. Given this variety of sources and formats, many choose to use mediator/wrapper. architecture [11], but applying these IR technologies also means that the information retrieval and gathering works need a high speed CPU and a good quality network environment. Some reports predicate that there will be more than 2 billion mobile users all over the world between 2007 and 2010. That is, there is one mobile phone for every three living persons on the earth.. Now, many people are familiar to use various kinds of portable. terminal devices, such as PDAs or smart phones, to access the information on the Internet at any place in any time to assist their daily businesses.. Some people also utilize some sort of. agent technologies of desktop devices to help themselves taking cares of their daily affairs. Moreover, personal mobile agents can help them to search, to filter, to analyze, and to translate the large volume information of the Internet. In most cases, utilize portable devices to connect to the information source has to make use of wireless network.. As the previous. search [44] estimates, the trend of Internet connecting technologies is shown in Figure 1.1.. 1-2.

(18) Figure 1.1: The ratio of the Internet among cable, wireless, and wireline.. As Figure 1.1 shows, in the near future, the wireless technologies for connecting to the Internet and private local area networks will be developed in a very great pace, and its percentage is large among all three main network technologies.. Some reports even estimate. that wireless connection will replace all the wired connection in near future.. As well as the. development of wireless networking technologies, people will also familiar with connecting to the Internet by multiple devices.. Figure 1.2 shows the trend of Internet users. [44]. Figure 1.2: The trend of Internet users.. Today, most people mainly connect to the Internet by PC, and retrieve the information on the Internet to his mobile device via the application of the PC (such as the services from AvantGo).. It is a PC centric Internet connection model.. However, according to the fast. development of wireless technologies, it is foreseeable that more and more mobile devices will be equipped the high power CPU and stable network connection. 1-3. There will be more and.

(19) more users are willing to utilize their mobile devices directly connecting to the Internet.. And. the services and applications on the mobile devices will be more and more fantastic and powerful.. Needless to say, the software agent technologies and frameworks also can be. easily applied into the mobile computing environment, but before the coming of the 3rd Generation Internet which is shown in Figure 1.2, there are some considerations have to be considered for providing more powerful applications on the mobile devices.. The software. agent technology is one of the savers. In the viewpoint of our research, there are two major developing trends of software agent technologies.. The one is the researches about the enhancement of the autonomy and. intelligence, and the other is the enhancement of the mobility and accordance of the agents. The highly autonomous and highly intelligent agents can accept the delegation ordered by the user to execute his specified works intelligently and autonomously.. The former kind of agent. technology is focused on the intelligence and autonomous for process and analysis of collected information from the network. In such case, the agents always stay on the same machine or host, and retrieve the information via the network. The processed information is always returned to the user, and let him to make the decisions, but in many agent frameworks, to those more intelligent agent systems, they even can autonomously make some decisions for their delegates if they have the user’s appropriate authorization. The model of traditional static agent for doing the user’s delegated works is depicted in Figure 1.3. As shown in Figure 1.3, the user or client may execute a buyer’s program on the buyer’s computer.. Through the slower Internet communication (compared to the inner I/O. speed of computer), the program always directly retrieve the great volume of information and documents from supplier’s programs or services on the remote supplier’s computer for further analysis and making decisions. If the retrieved documents are large enough, the process will consume lots of times.. 1-4.

(20) Figure 1.3: The model of traditional static agent for doing some user’s delegated works.. The technologies of mobile agents are mainly devoted to solve the previous mentioned situation. Consider the picture shown in Figure 1.4.. It is the same situation with Figure 1.3,. but this time, the user delegates a mobile software agent, and sends it directly to the remote supplier’s computer.. In such case, all the works, includes retrieving, analyzing, and. processing of documents, will be completed in mobile software agent on the remote supplier’s computer.. The data through the Internet communication are only the results and the mobile. software agent itself. Lots of times for transferring the unneeded and temporary documents will be saved. Figure 1.4: The model of mobile agent for doing the same works in Figure 1.3.. Applying the mobile agent technologies to the IR problem has many advantages. Consider the situation, if a user has to fetch all of the employee data from the remote database server, and then analyze and mining these data to find the some interested statistic trend. Using the traditional model has to fetch all the data items across the network into local machine.. If the employee data is huge and is distributed across several database servers, the. data fetching works will take very long time, and will waste lots of memory spaces of the local machine to store them.. Of course, in the whole fetch process, the network connection has to. be kept in a very high speed and good quality condition.. If the database servers are. distributed across many buildings or even several different cities, it will cost lots of money for temporary documents transmission. 1-5.

(21) In such case, the quantity of data items is much larger than the size of processing program agent itself. Thus, we can transfer the agent into each database server respectively, and then execute the agent's program code to analyze the data in the remote data server.. Consequently,. it can save lots of network transmitting time and refine the utilization of local machine's memory space.. This is the key benefit of mobile agent technologies. Utilizing the mobile. agent technologies, all the remote-accessed data items will be acted as local-accessed data. Thus, not only the processing time will be reduced dramatically, but transmitting will also be reduced too.. the cost for data. Furthermore, the network connection is no more. needed at most of the time during the data processing.. Currently, the mobile agent. technology is feasible for desktop computing environment, and it is more and more popular for many information retrieval projects. Most of the mobile devices are the resource-constrained devices.. It is not easy for such. limited devices to execute too complex works. The battery duration also limits the execution time for any software on the mobile devices. Accordingly, the features of mobile agent technology are very suitable for the mobile devices to do some useful and long-term information retrieval projects.. However, applying the mobile agent technologies to mobile. devices also has some considerations. The traditional mobile agent system is always an on-line system.. That is, no matter. what the processing node is at local or at remote, they have to keep the network connection available, especially for the host in the server-side.. In the on-line mobile agent system, the. host always has to be available on the network and wait the client's request.. For the desktop. PC and server, this is an acceptable solution, but for a mobile computing environment, considering the networking cost and networking stability, it is not easy to fulfill such requirements.. As mentioned earlier, it is important to propose a feasible design to conquer. the issues produced in mobile computing environment for retrieval information on the Internet. For the technologies of mobile agent system, first of all, the key point we have to emphasize is that the off-line mobile agent model is very important than the on-line one in the mobile computing environment.. The on-line mobile agent always stays at a device with well. networking and high computing power.. It is easy and fast for such on-line mobile agent to. migrate itself from one host to another. The off-line mobile agent is different. stays on the mobile device with unstable networking and low computing power.. It always The cost of. wireless networking for mobile devices is also more expensive than the desktop's service. 1-6.

(22) The target of our approach is to design a complete framework.. The solution has to minimize. the utilization of expensive wireless networking for mobile devices, but keeps the flexibility of mobile agents.. The framework not only provides on-line mobile agents on the host with well. networking environment, but also provides off-line mobile agents on the mobile devices with poor networking environment and low-computing power.. The user can utilize his mobile. device (mobile phones, PDAs, or even pocket PCs and notebook PCs) to launch some mobile agents, and let those agents assist him to retrieve his interested information from the private database or the Internet by any e-mail client applications. For the purpose of information retrieval, those mobile agents will migrate from one host to another. It also can migrate itself to another mobile device as well.. 1.2 Objectives and our contributions According to the explanation in section 1.1, the objective of this dissertation is to propose a feasible solution for the user to comprehensively access the information on the Internet or the local area network by the resource-constrained mobile devices.. Since the e-mail is almost. the most popular service for any network connected devices, it will provide a unified operating interface to the user for accessing the Internet. In this dissertation, we propose a framework named Email-based Mobile Agent Runtime Environment (e-MARE).. Utilizing the. instructions provided by e-MARE framework, the user can easily operate the mobile information retrieval agent embedded in the main server to retrieve the interested information from the Internet or private databases via his mobile devices.. Not just the embedded mobile. agent of the server, the user can design his own mobile agent at their mobile device, and then launch that agent to the remote host for his own purpose. The main contribution of this dissertation is the proposed email-based mobile agent and information retrieval application operating framework. named e-MARE.. We implement a prototype system. Through the framework, no matter what the device is, and no matter where. the user is, the user can use the unified user interface to design, launch, and operate the mobile agent to achieve his own purpose. Because of the off-line and asynchronous characteristics of e-mail protocol, the mobile agent has more time to retrieve and analyze the documents fetched from the network and the Internet.. The system may prepare some results with good. quality as an e-mail, and then send it to user's mailbox. Under this framework, the user can do anything just as if he is using the desktop PC with good networking environment, even though he is just using a mobile device with poor networking environment and poor 1-7.

(23) computing power. Finally, to demonstrate the feasible of e-MARE, we provide many examples and useful applications in later chapters of this dissertation.. 1.3 The challenges in our research Although we use the simple e-mail protocol as the interface between mobile users and servers, due to the original difficulties of information retrieval for complex and versatile documents on the Internet, there are still some difficulties have to be considered in our proposed framework. The challenges for design and implementation of our framework are briefly shown as follows: . Most of the documents of the information source on the Internet are semi-structured or even unstructured, they are hard to parse and analyze completely by application program.. . Different information source has different protocol for getting the documents.. It needs some mature wrapper technologies to solve the. diversity among several information sources. . Each mobile device has its distinct user interface and applications.. For. client side, the proposed framework has to adapt all the client devices, let all devices can easily operate the mobile agent with our proposed framework.. For server side, the result e-mail has to adapt itself to. fulfill the different requirements and environment of that specified mobile device. . For completely operating the local and remote mobile agent, the system also has to provide a complete instruction set or the programming language solutions for user. How to install, or compile and execute these commands and programming languages are also a great challenge of our design framework.. . Furthermore, for some mobile devices (such as mobile phones) with limited input interface, it is hard to input too many characters in a short time.. . The security issues are always the most important consideration for the 1-8.

(24) users and system developers.. In our research, we applying the. id/password mechanism to avoid some malicious users. Due to the advanced security issues are the future works of our approach, we just exclude these topics in the dissertation.. 1.4 Organization of this dissertation This dissertation is organized as follows. Chapter 1 gives a fundamental introduction including introductions, objectives, contributions, and challenges.. Chapter 2 gives some. discussions about the proposed problems and motivations more detail. Chapter 3 surveys the background and previous works related with the topic. In Chapter 3, we also survey the present and future technologies related to our research more detail.. Chapter 4 mainly. introduces some theories of mobile agent technologies, and their applications on the information retrieval. In Chapter 5, we propose the system design and discuss the issues derived from the design.. We will evaluate some real-world examples in Chapter 6.. Some. more complex and useful applications which are applying the e-MARE framework are extensively introduced in Chapter 7.. Finally, the concluding remarks are given in the. Chapter 8.. 1-9.

(25) Chapter 2 Problem Description This chapter describes the related issues and technologies of our research. At the first, we review many obstacles about utilizing the mobile devices to retrieve the information on the Internet.. Then, some related technologies, including agent technologies, mobile agent. technologies and information retrieval technologies, for improving the whole process of information retrieval are briefly introduced in the following sections.. Finally, we classify the. type of information retrieval models, and state the advantages for applying the e-mail protocol and related technologies to those mobile devices information retrieval approaches.. 2.1 Information retrieval on the resource-constraint mobile devices Information Retrieval (IR) is one of the most important applications of computer science nowadays.. The traditional IR is mainly focused on the well structured digital library. database and management system.. However, today, the Internet provides the user a cheaper. and easier way to access various data sources, and the advances in all kinds of digital communication provided greater access to networks.. Moreover, the Internet also provides. anyone an easier way to publish their own documents to the web as one of the data item of the information source.. Therefore, the information on the Internet becomes a very important but. unstructured and not easy manageable information source for information retriever. That is, the user is now facing the astronomical unstructured documents on the Internet. There are many search engines on the Internet which indexing lots of documents, and providing the users to query their interested information, but there is no search engine can cover all documents on the Internet.. Even the most popular search engine, Google. (http://www.google.com), has been pointed out that its estimated coverage is about 45%-55% until year 2002 [53].. Hence, many researches develop meta-search technologies, trying to. cover more and more Internet documents.. The basic concept of meta-search technologies is. shown in Figure 2.1. 2-1.

(26) Figure 2.1: The basic concept of meta-search technologies.. As Figure 2.1 shows, a good organized meta-search system integrates several information sources to provide a unified user interface, letting the users or applications to access those information sources in a unified and integrated manner.. The users or the application. developers don’t need to worry about the format of querying string at all.. However, the. diversities and differences of the terminal which is used by the client user or application are not considered in these approaches. For more and more greatly increasing mobile device users, to provide a unified user interface for them to access those versatile information sources is urgent. the basic idea of such idea.. Figure 2.2 shows. As shown in Figures 2.2, one of the main objectives of our. research is to contribute such a unified user interface mobile information retrieval framework. The detail of such framework will be explained in the following chapters of this dissertation.. 2-2.

(27) Figure 2.2: The unified user interface for the mobile device users.. Beside the consideration of the great number of documents on the Internet, how to efficiently fetch the useful information from the astronomical documents is the most important issue for many Internet users and information investigators.. There are lots of former. researches were dedicated on related topics, and they also have got lots of successes. are also lots of products with these technologies and serves millions of users.. There. However, most. of the traditional approaches mentioned above assume that the client user is sitting at the front of a desktop PC, and the PC is also equipped some sort of well networking connection. All the information retrieval works need a powerful computing environment to achieve the information retrieving requirements.. For more and more mobile users using mobile device. with limited computing power, those traditional methods will fail. The major difference between the desktop PC and the mobile device are the limitation of battery power, computing ability, network speed and quality, and other computing resources. These differences will dominate the technologies it will use for retrieving information from the Internet. Different to the desktop platform, most of mobile devices are equipped with low-speed CPU, fewer memory, smaller display size, and low-speed and unstable network connection. Most of the traditional information retrieval technologies are not suitable for such limited mobile devices.. Table 2.1 compares many popular mobile devices with notebook and. desktop PC.. 2-3.

(28) Table 2.1: The comparison between mobile devices and PC.. Devices SonyEricsson T68. CPU/OS. Memory/Networking. Display Resolution. N/A. 800KB RAM GPRS IrDA BlueTooth. 101x80 256 colors. 4MB MMC supported GPRS IrDA BlueTooth. 176x208 4096 colors. 16 MB SDRAM 8MB Flash ROM Memory Stick supported IrDA Serial MODEM. 160x160 16 gray colors. 64MB SDRAM 32MB Flash ROM SD/MMC supported CF Type I/II supported IrDA GPRS IEEE 802.11b. 240x320 3.5“ TFT LCD 65536 colors. 256MB (2GB) SDRAM 40GB Hard Disk CD-ROM/DVD-ROM IrDA 10/100/1000 Ethernet IEEE 802.11b 56K MODEM. 1024x768 14.1” TFT LCD 16.7 million colors. 256MB SDRAM 80GB Hard Disk DVD-ROM 10/100 Ethernet IEEE 802.11b 56K MODEM. 1024x768 15” TFT LCD 16.7 million colors. Nokia 3650 ARM9 CPU 104 MHz Symbian OS 6.1 series 60 Acer s10 Motorola DragonBall VZ 33MHz Palm OS 4.1 Acer n10 Intel PXA255 300MHz Microsoft Pocket PC 2003 Acer TraveMate C300 Tablet PC. Intel Pentium M 1.7GHz Microsoft Windows XP Tablet Edition. IBM ThinkCentre M. Intel Pentium 4 2.8 GHz Microsoft Windows XP Professional. 2-4.

(29) Therefore, to reduce the gap between these two kind of devices, it may have two possible solutions for a mobile device to retrieve the information on the Internet, 1) downsizing the scale of these technologies and executing them on the resource-constraint devices, or 2) executing those on the server-end, and leaving the mobile devices as the simple front end for the user.. In our survey and research, later choice is a better solution to adapt the most mobile. devices for information retrieval.. 2.2 Styles of information gathering For the daily information gathering works of Internet users, we divide those works into two major styles: On-line Information Gathering and Off-line Information Gathering.. We. call an information gathering style On-line if the user is utilizing the WWW browser or any other WWW surfing applications to browse and fetch the documents on the Internet continuously.. In general, the users of this type are always contiguously initiate some query. commands or operations to some sort of information sources, such as search engines, on the web, sitting on the chair in front of computer, and waiting the results be displayed on the screen.. Because the user is always waiting the results after he has issued his requests, the. results have to be returned as soon as possible. To return the query results to the users faster, it means, there is no enough time for the information retrieval application (in this case, it is some sort of web browser or web surfing application) to provide the complete results which have been completely analyzed from the fetched documents on the Internet.. In the meantime,. the technologies for the time-restricted information retrieval processes are always utilize the simple keyword-based technology to reduce the response time for information retrieval, or the information provider has to completely analyze the documents in advance. On the other hand, we call an information gathering styles Off-line if the user doesn’t care the response time from the Internet or network after he has issued his queries.. That is, in. most cases, the most important consideration of those users is the completeness and correctness of the result sets.. They would rather spend more time to wait the complete. results from the information retrieval application.. Thus, the information processing. application has more time to parse and analyze the fetched documents, and to rank the display order for the user according to the specified preferences and other important arguments of the user.. Compare the style to On-line Information Gathering, the user always can get more. comprehensive and more precise results from the Internet, but the cost which the user has to pay is more time to wait. 2-5.

(30) In the traditional information retrieval working environments, spending more time to wait the results from the Internet is always not acceptable for most Internet users.. However, in the. unstable network with low-speed CPU mobile computing environment, such time waiting situation is common and acceptable for most mobile users. Furthermore, most of the mobile devices can not efficiently execute full function Internet browser to directly surf the web documents on the mobile devices for the user, it is hard to utilize On-line Information Gathering Style on the mobile computing environment. Consequently, utilizing the e-mail based solution, Off-line Information Gathering Style, on the information retrieval from the Internet for mobile computing environment is acceptable and feasible.. 2.3 Classification of information retrieval models As mentioned in the section 2.2, to retrieve the information from the Internet in mobile computing environment needs a doable and convenient method.. It has to let the mobile user. getting the satisfied information by his mobile device under acceptable operating environment. To further explain the feasible e-mail based solution for mobile computing environment, we first illustrate the current major models for getting information from the Internet. We divide the current information retrieval model from the Internet by retrieving style into three categories as follows: . Web-based Information Retrieval Model, WIRM Web-based Information Retrieval Model, which is shown in Figure 2.3,. is a traditional model for the user to retrieve the documents on the Internet. It always also means the On-line Information Gathering Style which has been explained in section 2.2. Generally, the user in this model always executes a full functioned Internet WWW browser or information retrieval application to fetch the information and to operate the databases in real time. It always requires high computing power platform and always on-line networking connection.. In many cases, the client machine should be a. powerful desktop PC or notebook PC and to be able to run such complex and powerful applications.. Also, the client application always provides. the good-looking and fantastic user interface for the user to issue his query to the specified information source, and the user always stays in front of the PC during the whole retrieving process. Because the user always keeps 2-6.

(31) concerning the whole progress of information retrieval, the response time from the information sources should be very critical.. Figure 2.3: The model of Traditional Web-based Information Retrieval Service.. Although most information sources of the Internet provide some kind of good-looking user interface for the user to fetch their documents, there still are few Internet search engines (such as Google, http://www.google.com) provide the concise but straightforward user interface for mobile devices’ (such as cellular phones or PDA) WAP browser or simplified mini-browser. However, although the user interface is simple, those information sources still need to equip some sort of WAP-HTML gateway technology.. Thus,. when the user browses Google's web pages via mobile device, Google will first transform the original HTML format into WAP format.. The mobile. user then may use built-in or plug-in WAP browser to browse the Google's web pages. The WAP–HTML gateway technology is partly solving the information retrieval issues of mobile devices, but due to the translation technology between WAP and HTML is not mature (especially for Chinese web pages), and it can't optimize the results in a very short time, moreover, the mobile phone's wireless networking is very expensive, this is not a good solution for mobile devices to retrieve the information from the Internet. . E-Paper Style Information Retrieval Model, EPSIRM The E-Paper Style Information Retrieval Service, which is shown in. Figure 2.4, is a more flexible method for a user to fetch information from the Internet.. A user can subscribe some kinds of E-Paper on the E-Paper. provider web site in advance by their subscription interface which is provided by those famous information sources. After the user has been 2-7.

(32) subscribed and verified, the E-Paper provider web site will actively send some e-mail with the content of E-Paper specified by the user in advance to the user's mailbox periodically.. All the contents of E-Paper from the web. site are actively sent into user's mailbox, the user is not required to take care of any other chores except subscription.. Of course, in most cases, there is. no any customization mechanism for the user to adjust the contents of those e-papers.. Figure 2.4: The model of E-Paper Style Information Retrieval Service.. As mentioned earlier, this method is not required an always on-line networking connection. In addition, lots of networked mobile devices can read the e-mail directly. Thus, it is a feasible solution for most information retriever of traditional client platform, but one of the defects of the method is that the types of contents of E-Paper are limited, and in most cases, it has no way to customize the content to fulfill the user's personal requirements. Moreover, the web site always assumes the client's machine is desktop PC or notebook PC, so it lacks some acceptable technologies to adapt the contents into different applications and platforms.. Again, the most important defeat. of EPSIRS is lacking of interaction mechanism between the user and the information providers. In most current E-Paper providers, there is no way for a user to operate any specified E-paper, and to further customize the contents of those E-Papers. Consequently, EPSIRM is not suitable for a mobile user who wants to actively and comprehensively retrieve the information on the Internet. . E-Mail based Information Retrieval Model, EMIRM Using e-mail as the tool of information retrieval is not a new idea. 2-8.

(33) There were lots of researches and products using the e-mail protocol as the front end for other services (includes WWW, FTP, NNTP, Gopher, etc.) on the Internet at the era of poor infrastructure of network. Today, the quality of Internet networking for desktop computing environment is fast and stable. The role of e-mail for information retrieval and other Internet services is replaced by multimedia-enabled web browser. However, for a mobile user with limited resources mobile device, most of the mobile devices have not enough computing power to execute the complex multimedia-enabled web browser and information analyzing works.. The primary basic problem. about limitation of resources for information retrieval is rising again. In such a case, the best solution is letting the mobile device be treated as the front end client for the whole information retrieval process, and letting a powerful server as the back end server to fetch and analyze the documents from the Internet.. The role of mobile device can be treated as the front end. of information retrieval and only to provide a unified operating interface for the user to program, operate, and launch his remote applications, programs, or even mobile agents, and then wait the results returned from the programs or agents in the back end server. Due to most of the mobile devices has not the ability to execute the powerful web browsing, information retrieving applications, and the information analyzing applications, the interactions between the front end and back end are all through the e-mail protocol. That is, if the user wants to retrieve some information from the Internet, he may have to compose an operational e-mail (which is a mail in a proper format for operating the remote applications or information retrieval process) to specified the interested information, and send it to the specified e-mail server.. Then, the e-mail server will launch the information retrieval process. for the user, and prepare the results into an e-mail. Few minutes later, the results will be sent back to the user’s mailbox.. The user may browse the. retrieved results from the Internet by opening the e-mail client application again.. The mechanism of EMIRM is depicted in the Figure 2.5.. 2-9.

(34) Figure 2.5: E-mail based information retrieval model.. As Figure 2.5 shows, the user will not directly access the information from the information sources on the Internet. On the contrary, the user directly access to the e-mail server.. The works for actually retrieving and analyzing the documents are all done by. e-mail server.. The major advantage of this model is the efforts of the mobile devices will be. dramatically reduced.. All the chores can be finished on the powerful and high speed. desktop PC or workstation with broadband networking connection. In summary, the listed three models are suitable for each different hardware and software environment respectively.. For the high speed computing environment with broadband. networking, the first one is the most popular information retrieval model, but if the user has no time to sit at the front of desktop computer, or the user wants to get regular news and stuffs from some information providers, the second model is the best candidate for him. However, if the user is on the road and grasps some kind of resource-constraint mobile device in his hand, to fetch documents or information from the Internet in such mobile device, the third model is the only suitable solution.. 2.4 Basic concept of e-mail based information retrieval At the client side point of view of e-mail based information retrieval model, when the user wants to do something on the Internet or private database in remote location, he may use any front end device (Desktop PC, notebook PC, PDA, or mobile phone) to launch the e-mail client application, and then writing the commands provided by proposed framework or mechanism as the e-mail subject and body, sending this operational e-mail to a specified 2-10.

(35) e-mail account on the back end server which is registered in advance, and finally, waiting the reply from the back end server.. Few minutes later, he may execute the e-mail client. application again to check and see his mailbox. In most cases, he will get a response from the back end server, and the contents of the e-mail always are the executed results of his delegated remote application or mobile agent in the remote server. At the server side point of view of the e-mail based model, the back end server always checks the registered e-mail account to see any incoming messages. When a message has received, it first parses the subject of the e-mail to determine the type of the e-mail and to verify the id and related information according to the information of this e-mail.. If the. e-mail is with some simple operating commands (such as ‘&&SS;’ to show the status of the server), the server will operate these commands right now, and will pack the results into an e-mail, and then replies the e-mail to the user's e-mail account as soon as possible. In other case, if the user's operation sheets are complex, the server will reply a status report e-mail to the user, and then packs these operations as a program for executing. Meanwhile, the server will send the current progressive statistics to the user every 10 minutes (the default value, it can be changed by the user). The key mechanism of this model is the role of the e-mail server. It is responsible for accepting the user’s request, and will act as a coordinator to manage the information retrieval application or mobile agents on the server.. All of the interactions between the user and the. programs on the server should be through the e-mail protocol.. Every result e-mail will also. be rearranged in a specific manageable format, which can be further operated by user to fetch more information from the Internet by specifying the identification number in the e-mail. Using e-mail protocol as the gateway between the front end device of information retrieval and other Internet services has many advantages which are shown as follows: . Real Cross-Platform: No mater what the devices the user will use, includes desktop PC, notebook PC, PDA, or even mobile phone, most of these devices with networking ability are all equipped the e-mail client application.. So, the e-mail based IR mechanism provides the. real cross-platform service. . User Simplicity: Due to the simplicity of e-mail protocol, there is no any extra works has to be done for the mobile device of the users.. All. he needs is to look up some information retrieval commands, and 2-11.

(36) composes those commands into a proper operational e-mail. The other operations are all the same as the normal e-mail application. . Compatibility and Flexibility: The e-mail protocol is a very simple but complete protocol.. The contents of e-mail can be represented in. standard ASCII encoding.. Every program can easily read and process. the subject and contents of any standard e-mail messages.. Thus, the. mobile device with limited resources can also easily process the contents of e-mail too. Nevertheless, for the more powerful platform, such as desktop PC or notebook PC, of course they can process more complex data by attaching those data files into the operational e-mail. . No client application installation: Because of the e-mail is the common application for many networked devices, most of these mobile devices don’t need to install any related program for adapting this model.. This. is very convenient for system developer to propose their new services. All the new services are only to be installed into server. There is no need to modify any application at client side.. The cost of deployment. will be greatly reduced. . No user training: Almost every mobile device users have enough knowledge and abilities to make use of e-mail application to process their own messages.. So, this e-mail based model can save lots of extra. training works. . Low cost, and low and short bandwidth required: Because the e-mail based model doesn't need to educate the user, and it also doesn't need to install any program into client side, the cost and time will be greatly reduced. Moreover, because the e-mail protocol basically is some kind of off-line network protocol, it needs the network connections only while the client application is being transferred the e-mail. Thus, the networking cost will be dramatically reduced too.. For a mobile. network (such as GSM/WAP/GPRS connection services) with high charged networking service, this is a great deal. . User-Oriented and autonomous information retrieval process: Another important characteristic of this e-mail based model is that the model is a 2-12.

(37) user-oriented and autonomous information retrieval process.. Under. the architecture, all the complex retrieval operations have to be programmed in advance.. After the system have received and parsed. this e-mail, it will launch a static agent to monitor and schedule the user's mobile agent to do its works, and the static agent also will report the executing status to the user periodically. All the works will be automatically run at the server or across each needed hosts. The user also can write the event handler in their commands sheets of the e-mail to guide the static agent treating every possible occurred event during his works. Finally, after all the works have been done, the static agent will send a complete and well quality result to user's mailbox.. That is,. if the user's commands sheet is a good design for information retrieval, the boring progress for information retrieval can be discarded. All his works have to be done are just to wait the returned results form his own mobile agent. . Server scheduling optimization: Beside, because all of the user's information is handled at the server side, the system can optimize the system's performance according to the system's states.. Many. technologies such as cache mechanisms, scheduling mechanisms can be applied in the system to boost the system’s performance.. 2.5 Classification of e-mail based information retrieval According to the description of the previous sections, we conclude that the e-mail based information retrieval model is the best candidate for the resource-constraint mobile devices. To utilize the e-mail by mobile devices to fetch the documents and to retrieve the information from the Internet or private local area network, according to the complexity of the model, we also divide them into three categories which are shown as follows: . Data-Driven System: All information retrieval functions are embedded in the main server.. The user can only specify the interested data field,. and then expects the corresponding results of those data fields returned from the remote server.. In this type of e-mail based information. retrieval system, the user has no flexibility to determine the control flow 2-13.

(38) of the information retrieval process.. That is, the user only can specify. some simple query commands with simple arguments to get the information from the server.. There is no any control statements and. other further advanced statements can be specified by the user. Although such the system can provide the user having the ability to install their own program into the server to achieve their own information gathering purpose, in the client side, he only can just call the program with some simple arguments.. He can’t specify the. repeated execution times for some specified statements or programs, and he also can’t determine the control flow according to the return value of any program or application. . Control-Driven System: This type of e-mail-based information retrieval system is more flexible than data-driven system.. It is supporting some. kind of information retrieval script language. The user can utilize such script language to write some simple batch programs for retrieving information from the Internet.. Moreover, the user can also operate the. remote mobile agents by writing programs with this kind of language. This type of e-mail based information retrieval system at least provides basic flow control statements, such as decision statements and loop statements, letting the user has more ability to control the execution flow of process of information retrieval.. Furthermore, the well-formed. results can also be treated as the reference for the next operational e-mail.. The user can manage all of the functions and applications built. in the system by writing a program in their operational e-mail.. In most. cases, this type of information retrieval is enough for mobile user to comprehensively fetch their interested information from the Internet. . Full Interactive System: In most cases, the Control-Driven System is enough for most mobile users. However, sometimes, if the system has not provide the user’s desired application, and there is no appropriate desktop PC for the user to install or deploy his own application, in such cases, the system also needs to provide the user the ability to install their own application into the remote server by his simple mobile device.. 2-14.

(39) This is the main purpose of the invention of the Full Interactive System. This type of e-mail based information retrieval system supports all the ability of programming language.. Actually, it just extracts the. program from the operational e-mail, and dispatches the extracted program into the operation system.. After the execution of the program. has been finished, the results will be wrapped into an e-mail, and then be sent back to the user’s mailbox.. In this style, the user can write any. customized mobile agent with this kind of programming language.. For. convenience of users, the programming language can be the traditional language such as C, JAVA, Perl, and any acceptable shell language etc. In practice, the interpreter style language, such as Perl and some shell language, is easier to implement. Different type of e-mail based information retrieval system has different characteristic and its applicable domain.. The Data-Driven System is simple, and it is just for the user who. only wants to query his simple interest information from the private database or some specified information sources on the Internet.. Getting the item price of product of our. company, getting the timetable of the school, or quoting the stock value from stock market are all the application examples which can be implemented by Data-Driven information retrieval system. For more complex applications, such as surfing the Internet, or retrieving the information across many search engines, the Control-Driven information retrieval system is more viable than Data-Driven System since it provides more flexibility for the user to control the results from the server.. In practice, the system will provide the control statements such as for/next,. if/then, while/wend letting the user to control the execution flow for his information retrieval process.. Furthermore, the system will also rearrange and reformat the results in a. well-formed format, letting the user can directly utilize the results to retrieve the information from the Internet more intensive. Sometimes, the simple control flow can’t fulfill all the requirements of the user.. To. support full function programming language is also important.. The major advantage for an. e-mail based information retrieval system is the great flexibility.. In such system, the user can. not only operate the remote application to obtain his interested information, but also write his own application by mobile device. Although the application is written on the mobile device, 2-15.

(40) the execution place is the remote server.. Since the programming language can do everything. that the computer can do, the user can do everything by his mobile device as well.. 2.6 Applying mobile agent runtime mechanism to our design To further maximize the ability of the e-mail based information retrieval system, applying the mobile agent technologies into the system is one of the most important approaches.. A. mobile agent is an autonomous software program which can be migrated from one host to another host during the execution time.. Since the mobile agent technologies can provide the. programmer to write a program which can easily execute itself on any available host, this mechanism can achieve maximum flexibility in any stable and unstable network environment. However, applying the mobile agent technologies into mobile computing environment is very different in typical desktop environment. about.. It still has some considerations which have to care. Some of these main considerations are shown as follows:. 1. The host always has to provide a running mobile agent runtime environment to accept the migrating request from the other mobile agent in remote host. 2. The network of hosts always has to be available. 3. If the migrating request from mobile agent has received, the host has to response as soon as possible. 4. The mechanism recommends more network bandwidth for processing the migration of mobile agents and other related works.. Sure thing, the. networking cost is more expensive. 5. Both the client and server side have to install specified runtime environment. Then they will have the ability to launch, migrate, and operate the mobile agents. Because the client side applications always need more computing power, lots of mobile devices can't fulfill these requirements. The mentioned considerations let the mobile agent technologies can’t be easily adapted to mobile computing environment.. That is, applying mobile technologies into mobile. computing environment needs many modifications.. To defeat the shortcoming mentioned. above, we propose an Email-based Mobile Agent Runtime Environment called e-MARE. The e-MARE provides the users the ability to utilize the unified interface for many different 2-16.