Information Sciences 166 (2004) 1–17 www.elsevier.com/locate/ins
Architecture design of grid GIS and its applications on image processing based on LAN q Zhanfeng Shen a,*, Jiancheng Luo a, Chenghu Zhou a, Shaohua Cai a,b, Jiang Zheng a, Qiuxiao Chen a, Dongping Ming a, Qinghui Sun a,c a
c
Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China b The Surveying and Mapping of Information Center, Beijing 100088, China Institute of Surveying and Mapping of Information Engineering University, Zhengzhou, Henan 450052, China
Received 15 July 2003; received in revised form 9 October 2003; accepted 10 October 2003
Abstract Computer technology and its relative subjects developed at very high speed in recent years, so is geo-information science, including Geographic Information System (GIS), remote sensing (RS) and global position system (GPS). But with the increase of data, many data cannot be used efficiently because of the tremendous amount of data and information and the difficulty of process and transfer through network. So how to develop internet technology to solve these problems becomes a difficult problem for current computer experts and geo-science experts. Fortunately, grid computing provides us the method to solve this problem effectively. Grid computing is a resources sharing model presented by computer experts to solve current network resources imbalance problem. Basing on the application of grid computing on geographical information system (GIS), q
Sponsored by Chinese ‘‘863’’ project (2002AA135230) and National Natural Science Foundation of China (40101021). * Corresponding author. Postal address: State Key Laboratory of Resource and Environmental Information System (LREIS), Institute of Geographic Sciences and Natural Resources Research (IGSNRR), Chinese Academy of Sciences (CAS), 917 Building, Datun Road, Anwai, Beijing 100101, PR China. Tel.: +86-10-6488-9764; fax: +86-10-6488-9630. E-mail addresses:
[email protected] (Z. Shen),
[email protected] (J. Luo),
[email protected] (C. Zhou),
[email protected] (S. Cai),
[email protected] (J. Zheng),
[email protected] (Q. Chen). 0020-0255/$ - see front matter 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.ins.2003.10.004
2
Z. Shen et al. / Information Sciences 166 (2004) 1–17
this paper analyzes the weakness and problems of traditional GIS, and then gives the method to solve these problems with the technology provided by grid computing and web services. After analyzing the characteristic of grid computing this paper expatiates on current application status of grid computing on GIS and the problems it faces, with the technology of middleware, this paper presents the architecture of grid GIS and lists the techniques it needs. In conclusion, this paper concludes that the distributing middleware architecture based on grid geographic markup language (GridGML) and web services technique is a good solution to current problems, this architecture can also solve those problems such as effective resources sharing through internet and advancing international application’s efficiency, at last we discuss its implementation process based on LAN. 2003 Elsevier Inc. All rights reserved. Keywords: Grid computing; GIS; Grid-GIS; Middleware technology; Grid-middleware; GML; Web service; Metadata
1. Introduction Geographic information system (GIS) has developed rapidly since it came into being in the 1960s [3]. In recent years, with the progress of computer science and remote sensing (RS) and global position system (GPS), GIS has developed at very fast speed. At the same time, the data we obtained from RS, GPS and so on become more tremendous and the problems we need to process become more complicated, and these problems always involve a great deal of data transfer on the internet. So it has become an important task of GIS field [1] to discover information and knowledge from different sources and to find a method of knowledge integration. ‘‘Digital Earth’’ strategy tries to solve this problem by describing everything relative to geographic position in digital form and store those data into computers, and then provide services for the users through data sharing [4]. Digital Earth also needs to solve the problems that involve much data processing and computing [19]. So how to harmonize GIS project with colossal data and how to balance huge network flux become a main problem that current computer experts and GIS experts must solve. So according to the problems listed upon, it has become a main development direction of GIS field to avoid the waste of resources and information caused by lopsided tremendous data storage and universality of geographic data’s distributing and to find a good method to access and process these data and information. Many researchers have developed their method to access and analyze data on the internet, but their methods are always very complicated, and many data even need hundreds or thousands of billion scales’ computing, while the current data management system structure and method and technology cannot meet the need of high-performance disposal capability. So on the basis of present network development, many researchers present the conception of grid computing to solve these questions.
Z. Shen et al. / Information Sciences 166 (2004) 1–17
3
Grid computing is derived from distributing parallel computing and network high-performance computing [7], which both depend on specific computer hardware and are both expensive to be used universally, while the idea of grid computing solves the question primely. Computer network has come through internet and web period [4,6,13]. Internet makes it possible to correspond among different computer hardware, web makes it possible to communicate information in different places, while grid computing tries to make all the resources shared, including computing resources, storage resources, communicating resources, software resources, information resources, knowledge resources, etc. [5,15]. With grid computing as the basic running environment of our system, we set up a series of flexible and extensible information service system structure, which ensures the grid system’s running favorably. Basing on these techniques, we apply grid computing to GIS field to construct grid GIS (the application of grid computing on geographic information system), which has been proved to be successful by our grid GIS system based on LAN. For those projects which need much time such as image processing, the idea of grid computing matches the need well. So it is very necessary and essential to design and build grid GIS system based on grid
Fig. 1. Design and implementation of grid GIS based on LAN.
4
Z. Shen et al. / Information Sciences 166 (2004) 1–17
computing environment. It is no doubt that the implementation of grid GIS will support greatly current GIS theory especially web GIS theory, and will give GIS a new development direction. It will also give the users more convenient approach to complete their tasks, especially those difficult tasks which involve tremendous data and complex computing, and provide access to the required information for the analysis and interpretation of an initial dataset [2]. This paper will talk about grid GIS and its implementation from target analysis, technique support, method research and software implementation [18], according to Fig. 1. Firstly we analyze the application of grid computing in GIS project, then discuss the architecture and key technology to implement them based on LAN. We then put forward that the distributing middleware architecture which is based on grid geographic markup language (GridGML) and web services technique is a good solution to current problems, because it can solve the problems such as resource sharing. We consider grid GIS is a good development direction of current GIS, after that we discuss the implementation of the grid GIS system based on LAN.
2. Grid computing and its relative technology 2.1. Grid computing and grid middleware Grid computing was firstly put forward by Ian Foster in the 1990s; it aims to share all the resources on the internet to form a big, high-performance computing network. A main characteristic of grid computing environment is that a user may connect the grid computing system through internet, and the grid computing system can provide all kinds of services for the user. Ian Foster suggests that there are three definite characteristics of grid computing [7,17]: (1) it should coordinate computing resources and users that exist within a variety of control domains, which differs markedly from a local-area network. (2) It must coordinate such resources with ‘‘standard, open, general-purpose protocols and interfaces’’. (3) It must deliver nontrivial qualities of service that provide users with access to computing resources that are greater than the sum of the system’s nonintegrated parts, such stringent definitions classify the socalled grid middleware that is used to enable grid systems. Globus is one of the biggest corporations which do their research on grid computing. Cooperating with IBM, Microsoft, HP, Globus has constructed Globus Toolkit 2.2(GT2.2), GT3 as an architecture of grid, which formed the ‘‘five-tier sandglass structure’’ [12]. On the basis of five-tier sandglass structure and web services technology, they presented open grid services architecture (OGSA) which centered on grid services, in this paper we will present our architecture of grid GIS based on OGSA.
Z. Shen et al. / Information Sciences 166 (2004) 1–17
5
Grid GIS is the application of grid computing on GIS field, and we can understand its conception from three points listed below: 1. Grid is idea. The target of grid computing is to share resources and data and every node to work together, which suits to the needs of GIS projects. A main characteristic of grid is sharing of all kinds of resources under some specific regulars, including data, applications, computing capacity, and these regulars can ensure that all kinds of resources in the grid system work well together. 2. Grid is technology. Grid computing must have many techniques to ensure resources sharing and cooperation so as to share all kinds of resources and make them cooperate well with one another. There must be many criteria to ensure the security and communication of every part of the whole system. And grid GIS needs these techniques to share all kinds of resources more efficiently. 3. Grid is basic establishment. Grid system is made up of all kinds of computers, data, equipments and services. With the development and maturation of grid, it is important to set up some resource nodes all over the countries and even all over the world; it should integrate many resources on the internet, and should provide very excellent services for the users. When grid environment is established, users can use its resources from internet as the manner as we use electric power currently, we need not know where the power comes from. With the software and data grid computing provides, our investments and maintenances will lessen sharply. Digital Earth can be looked as an application of grid computing on geoinformation science. Its implementation needs many subjects, especially some information science and technology, including information autobahn, computer broad band, remote sensing and secondary planet images, spatial information technology, data processing and storage, virtual reality technology, etc. According to XueYong, grid computing is the best way to implement Digital Earth strategy and solve the problems it involved. At present, grid computing is only in its primary period, and many famous grid projects have been experimented such as IPG Grid of NASA, USA, and ASCI Grid of energy sources, USA, Euro Grid and Data Grid, EEC. China has done some research in grid computing since 1990. Chinese government and Chinese technology ministry has invested much on grid research, the main target is the breakthrough of grid key technology and the setup of grid standard and the application of grid to other fields. By this way, we strengthen the construction and research on grid computing and industrialization [8]. And the application of grid computing on other research fields has been in their experimental periods. Here with the grid computing and the middleware technology we present our grid GIS architecture based on LAN.
6
Z. Shen et al. / Information Sciences 166 (2004) 1–17
Middleware is a sort of system software or service program which runs in client computer or server computer, and it can serve as many roles, such as process management, spatial information and resources sharing, information storage and access, system security login and authorization, services monitor and so on [4,6,9]. In grid computing, middleware serves as not only the connection of every part of the application, but also the implementation of many complicated operations. Middleware is always located between the application layer and the basic layer. By transparent encapsulation of the relative function and application, it makes the application software independent of computer hardware and operation system, and makes it possible to communicate among different operating systems. Middleware is widely used in many huge enterprises’ standard platform and increases the haleness and the modularity of the whole system. Basing on these specific characteristic of middleware, our grid GIS system was designed and implemented with the technology of middleware. Grid middleware is a sort of middleware in grid computing system. Other than the characteristics listed upon, grid middleware has some extra characteristics in grid GIS system. According to grid middleware’s functions in the system, we can divide them into message middleware, object middleware, security middleware, registry middleware, resource management middleware, duplication middleware, task management middleware and error management middleware. In our grid GIS system, the main function of grid middleware is as follows: 1. It accords with the principle of software modularization. In GIS project, especially in many big GIS projects, using middleware can help us to lighten our ‘‘software’s weight’’, which accords with the principle of component software [9,16]. At the same time, grid middleware helps our system to develop toward software modularization, and we can easily advance the efficiency and the service quality of our software just by improving some grid middleware. 2. It accords with the principle of reuse of software. From the history of GIS we can learn that with the development of GIS software, many sorts of software become more and more complicated with the increase of their function. While if we change their architecture and reconstruct their architecture based on different middleware, and every middleware is an essential part of the software system, every part has its specific function, we will benefit from it. As long as we have designed and reserved the interface it needs, we will be able to complete the task easily. Every middleware will become ‘‘light’’, and by this means it is easy for programmers to complete their work. 3. It is easy to communicate among different platforms. Different middleware has different mode or architecture, and it is independent of system platforms, we can use this characteristic to communicate between different platforms.
Z. Shen et al. / Information Sciences 166 (2004) 1–17
7
4. It accords with the principle of software system. Although traditional client/ server architecture ever worked well in designing different GIS software system and it provides the system many flexible characteristic, it leads to some bed phenomena. For example, some software aggravate the burden of server or client computer, but in the client/middleware/server architecture, we can assign different grid middleware to complete different work and then assembly them to form a whole system, by this means we lighten the weight of client and server and boost up the haleness of the system. 2.2. XML and GML Extensible markup language (XML) was a subset of standard for general markup language (SGML) released by W3C organization in 1998; it is a sort of internet description language for communication. Since XML is very easy to master, and can be extended easily, it has become a necessary means of communication across internet. In our system XML serves: (1) as an internet description language. XML defines many standards for communication, in our system it can serve as task description language and resources description language; (2) as a main standard exchange language. XML provides many mechanisms to ensure the security and integrity of data exchange. XML is also a good language for internet communication considered from data expression, and it is convenient to publish, find and bind services. Geography markup language (GML) is the extension of XML, and it was launched for geographic information communication on the internet by Open GIS Consortium Inc. GML is the encoding for the transport and storage of geographic information, including spatial and non-spatial properties of geographic features. By defining the XML Schema syntax, mechanisms, and conventions, GML provides users methods to store and transmit all kinds of property information and geometry information. XML also provides exact definition of metadata and spatial entity model and every reference frame [10,14]. And of course, GML has the characteristic of XML, that is, extensible, and we can easily define some other criteria based on GML. GridGML is an information description language in our system based on GML, in our system its main functions are task description, resources description and internet communication. GridGML has the characteristic of both GML and grid computing, and in our system it also serves as a main communication language among different middleware. 2.3. Web services and WSDL Web services are platforms where you can set up your new distributing application, i.e. web services could provide many Application Programming Interfaces (APIs) to complete many internet operations [11]. From the
8
Z. Shen et al. / Information Sciences 166 (2004) 1–17
definition of web services we could know that web services provide us the method to communicate information on the internet. Web service description language (WSDL) is a language based on XML which is used to describe web services and their functions, parameters and values. Since being based on XML, WSDL has the characteristic of XML. Simple Object Access Protocol (SOAP) helps us to communicate through fire wall and is prone to understand, we can call web services’ method by remote procedure calling (RPC) provided by SOAP, and later in this paper we will discuss its implementation specifically. Web services are independent of platform, independent of location, and independent of programming language. They provide many security guarantees; basing on these characteristics we build our grid GIS system with web services technology.
3. Architecture of grid GIS and its key technology 3.1. Grid GIS and web GIS Grid GIS is the application of grid computing on GIS, in this paper we apply the idea of grid computing to GIS field and use grid GIS to serve for geographical information computing. Grid GIS is different from web GIS, we can understand them in this way: 1. They have different software structures: grid GIS is the application of grid computing in GIS field, which is constructed on the basis of grid computing environment; while web GIS is constructed on the basis of today’s internet. 2. They have different functions: As what is stated above, the main target of grid GIS is a universal resource sharing, including computing resources, data resources, storage resources, information resources, knowledge resources and so on; while web GIS emphasizes mainly data sharing. 3. They have different implementation methods: the architecture of grid GIS is grid computing, while the basis of web GIS is today’s internet. The implementation of grid computing is on the basis of today’s internet but is more complicated than today’s internet, which has more advanced functions and can be called ‘‘next generation’s network’’. Its implementation depends on many other subjects, and of course it will have more mighty functions, later we will talk about how to implement grid GIS. 4. They have different service methods: web GIS serves a user by whole software, that is, a user must buy the software even if he only needs part of the software’s function, which can lead much waste of money; while grid GIS serves a user by the means of ‘‘service’’, a user can select the services he need and build ‘‘his software’’, which solves the problem above.
Z. Shen et al. / Information Sciences 166 (2004) 1–17
9
5. Grid GIS does not conflict with web GIS: they may develop together, and grid GIS may be implemented in the environment by the technology of web GIS. 3.2. The architecture of grid GIS The architecture of grid GIS is the running structure of grid GIS, which decides the stability and extensibility of the whole system. The architecture of grid computing defines many protocols and APIs between grid nodes to ensure all kinds of operations and applications of grid system running successfully [12]. Fig. 2 lists the main running architecture of grid GIS. It is made up of five layers according their main functions, and each layer has its specific functions. From below to upon are basement layer, resource layer, control layer, implementation layer and application layer. The basement layer includes basic structures and protocols of network and some other protocols for grid system, which ensure the information communication between grid middleware and the grid system. The resource layer contains many resources, which are mainly made up of local resources and remote resources which have been registered to resource registry center for sharing. The control layer is the core of the whole system, which ensures the whole system to run correctly. The implementation layer is made up of many middleware, which connects with the system by some specific interfaces. And the application layer is the applications of users based on GridGML. Fig. 3 displays the running frame of the grid GIS system. From the figure we can learn that a user can apply for resources from system after login and submit his task by GridGML description language. After receiving the task presented by user, system searches for usable resources that have been registered to resource registry center, then parses the user’s requirement (task), task management middleware then hands out and distributes the task to grid nodes. Grid nodes do their jobs given by task management middleware and return the results and reports (including task source, start time and end time, error info, etc.) to task management middleware, the task management middleware
Fig. 2. Architecture of grid GIS system.
10
Z. Shen et al. / Information Sciences 166 (2004) 1–17
Fig. 3. Running-frame of grid-GIS system.
merges the results and info and returns them to the user proxy. Meanwhile, security middleware and error disposal middleware are in their monitor status to ensure the system’s running stably all the time. 3.3. Implementation of grid GIS From Fig. 3 we can know that task management middleware and resource management middleware are the most important parts of the system. In fact, these two parts are the key parts which may decide the success or failure of the whole system, now we will discuss these two parts’ technical implementation in details. 3.3.1. Task management middleware Task management middleware takes charge of all the tasks including task management and dispatch, disintegration, distribution, result merge, report generation, etc. Fig. 4 shows the functions of task management middleware. The mission of task management middleware in this system is mainly completed by task object. After receiving users’ apply, task management middleware may dispatch many task objects to complete different functions, every task object has the object-oriented characteristic, and every task object may have its child object. Users cannot operate with resources directly except by task object. Different task objects have different missions, including searching, displaying, suspending, resuming, stopping the operation, which were decided by task management middleware when they were generated.
Z. Shen et al. / Information Sciences 166 (2004) 1–17
11
Fig. 4. Task management middleware and task object.
Fig. 5 lists the main bind principle of the task management middleware object and their child object. In our system, when a task of image segmentation was launched, the task object then generates some child objects and binds them to grid nodes. Child objects then gain the information of grid nodes by the function GetGridInformation(. . .), including the status of computer, CPU, storage and so on, and then by function Segmentation(. . .) to bind with remote computer grid nodes, and transfer image data, segmentation information and requirement. After gaining the task, the child objects return its status info by function SendGridRoport(. . .) when it is time of 23 · initInterval, then task management object will decide the next task by function SendGridMessage(. . .) according to these messages: if the running status and task are done well, the task will continue, or else stop.
Fig. 5. Bind principle of task management middleware object and grid nodes.
12
Z. Shen et al. / Information Sciences 166 (2004) 1–17
3.3.2. Resource management middleware Resource management middleware deals with the problems how to describe resources in resource registry center and how to publish, find and bind service resources. Fig. 6 displays the implementation flow of resource management. Different users may have different authority to use the resources. After login, users can declare their resources as shared resources (local shared resources or public shared resources) by registering them in the resource registry center. The resource description language (RDL) is based on GridGML, and it is also an important mechanism for resource search. Resources and information in resource registry center are organized by directory structure, and they are organized with light directory access protocol (LDAP) based on database, which may provide technique support for the mission transfer, mission dispatch and mission balance of computers in different places. In this way it is convenient to search and control all the missions, and it is convenient to transfer information among different parts of the system. Table 1 lists the main content of resources description. The registry mechanism of information resources is authorized by administrators. All the resources registered in resource registry center are described by GridGML description language, in this way it is convenient to find and bind services for different parts of the program. Using GridGML between different parts of the system as an information exchange language may also solve the problems such as information expression, information process and information
Fig. 6. Implementation flow of resource management middleware.
Z. Shen et al. / Information Sciences 166 (2004) 1–17
13
Table 1 Contents of resources description Data resources 1. Data resources address 2. Data access method 3. Metadata 4. Data type 5. User protocol 6. Valid date 7. Authority 8. Pay
Service description contents 1. User protocol 2. Description method 3. Authority 4. Service type 5. Service explain 6. Service provider 7. Certificate 8. Pay
Computing resources 1. Resources address 2. User protocol 3. Valid date 4. Authority 5. Pay
Storage resources 1. Resources address 2. User protocol 3. Valid date 4. Security 5. Pay
control, and the exchange of information resources may be transferred by SOAP message. 3.3.3. Other middleware Besides middleware listed above, we need some other middleware to construct the grid GIS system, including user management middleware, login management middleware, message management middleware, security middleware, duplicate management middleware, error disposal middleware, log management middleware and so on. They are independent of one another and can cooperate with one another so as to complete all kinds of tasks that user submitted. 3.4. Key technology of grid GIS From what we have discussed above we can know that it involves many technologies to implement the grid GIS, besides some protocols of grid computing, grid GIS must stand by some other protocols, which include: 3.4.1. Metadata management and information service Grid computing must solve the problems of publishing, storing and accessing resources and services, so is in grid GIS system. In a grid system, resources and their providers are distributed at different places. Metadata are used to manage, describe, collect and organize all the resources in the grid, and are used to describe information about resources, methods, datasets, and users and so on. Information services are the basic services that metadata provide for outside applications, which can implement the registry and publish of the
14
Z. Shen et al. / Information Sciences 166 (2004) 1–17
resource entity services. Information services can communicate with registered entities by metadata information services. So in a grid GIS system, we must set up a flexible and extensible information service architecture, which could ensure the information sharing and security. According to this requirement, we design GridGML based on GML to describe metadata information, which is simple to use and is extensible and prone to be parsed by many explorers or applications. 3.4.2. Data access service All the metadata in the grid system constitute metadata directory, which could describe all the metadata in an efficient way. So metadata must be as follows: (1) they form a system which should be arranged by tier structure like directory, (2) they should not destroy the present method of describing metadata, and should work well with them. To fulfill this requirement, we take a uniform form to access data using GridGML to describe messages. Although data and their storage forms may be different from each other, we can get what we want rapidly by using GridGML, and gain uniform data access interface and method to search, access, read or write. 3.4.3. Task management service Task management service is completed by task management middleware. The main targets of task management service are assignment and management of the user’s tasks and merge of the returned results. Different task need different method. In our GIS field, there exist many different requirements, so we take different methods to construct different task management middleware and construct different task object and combine them to the task management service according to some specific interface. 3.4.4. Security service We must ensure the whole system’s security since there involve many data exchanges in our grid GIS system. Besides those protocols that are relative to security provided by grid computing, we must also solve other security problems such as data transfer bankrupt and data encrypt.
4. A sample of grid GIS based on LAN Fig. 7 shows a sample of grid GIS based on LAN. Using some computers as the resource registry center, task management center, error disposal center and message management center, we can set up our grid environment according to Fig. 7. Firstly every node registers its shared resources in resource registry center (and updates in definite intervals), which is
Z. Shen et al. / Information Sciences 166 (2004) 1–17
15
Fig. 7. Implementation of grid GIS based on LAN.
described by GridGML. Suppose that a user logins from node A, and presents a mission of RS image division, the RS image was stored in node B, and the mission was completed by node A, C, D and E using the same division method. After receiving the user’s mission, the task management middleware parses the mission and searches the available resources, then decomposes the task and distributes it to C, D, and E, and defines the initial time as X s. C, D, and E do their jobs by the method according to Fig. 5 and send a report to task management middleware when it is time 23 X s; task management middleware then analyzes the report: if one node has done its job well, then task management middleware sends a message to it and make its counter become zero, then continues the next counter; or else task management middleware must do some error disposal. After A, C, D and E have completed their jobs and return their results and reports to task management middleware, task management middleware merges the results and sends them back to B (according to the user’s directions, here we suppose the user has asked the system to returns the result to B), then return the report to A and interrupts the connection with C, D and E.
5. Conclusions Grid GIS is a new issue that faces many researchers in computer and GIS fields and it is only in experimental period at present. While it provides us a
16
Z. Shen et al. / Information Sciences 166 (2004) 1–17
good foreground of application in GIS field, it provides us more challenge than ever. Grid GIS is a sort of project that involves in different regions, different platforms, different subjects and different fields’ researchers. According to the characteristic of GIS and the development of current grid computing, we present in our system GridGML as a main information description language for internet communication; with web services technology, we organize the whole architecture based on middleware structure. According to the specific questions and data, we set down some protocols in this system to implement the information and resources sharing, and then we construct our application on this distributed system. Our architecture of grid GIS is based on OGSA with the technology of dotNET and web services. Since different tasks can be described in different ways, so how to describe a task is very important for the system. How to make the system generate the intelligentized task description will be a very important problem. At the same time, this is also an important norm to evaluate the capacity of the system, which is our next work.
References [1] M.-L. Shyu, C. Haruechaiyasak, S.-C. Chen, Category cluster discovery from distributed WWW directories, Information Sciences 155 (2003) 181–197. [2] N. Giannadakis, A. Rowe, M. Ghanem, Y.-K. Guo, InfoGrid: providing information integration for knowledge discovery, Information Sciences 155 (2003) 199–226. [3] S. Chen, X. Lu, C. Zhou, Guide of Earth Information Sciences, Science Publishing House, 1999. [4] J. Luo, C. Zhou, S. Cai, The design of middleware-based grid-GIS, Geo-information Science 1 (3) (2002) 17–25. [5] I. Foster, C. Kesselman, The Grid: Blueprint for a New Computing Infrastructure, Morgan Kaufmann Publishers, Inc., San Francisco, CA, 1999. [6] J. Fang, J. He, Architecture of grid GIS and its implementation techniques, Geo-information Science (4) (2002) 36–42. [7] I. Foster, What is the grid? A three point checklist. Available from: . [8] N. Xiao, X. Lu, H.M. Wang, Four forms of grid computing, Computer World Journal 40 (2002), B1, B2, B3, B4. [9] Z. Zhang, Middleware, Techniques & Products & Applications, Chinese Petrifaction Publishing House, 2002. [10] Open GIS Consortium, Inc. OpenGIS Geography Markup Language (GML) Implementation Specification, version 2.1.1.2002. Available from: . [11] S. Yasser, Interdiction to web service, Programmer (2002) 69–70. [12] The Globus Projecte Argonne National Laboratory USC Information Sciences Institute, Grid Architecture. Available from: . [13] A. Ashton, Getting the grid, IEEE distributed system online, 2002. Available from: .
Z. Shen et al. / Information Sciences 166 (2004) 1–17
17
[14] Open GIS Consortium, Inc., OpenGIS Geography Markup Language (GML) Implementation Specification, version 2.1.2.2002. Available from: . [15] N. Xiao, W. Fu, SDPG: spatial data processing grid, Journal of Computer Science & Technology 18 (4) (2003) 523–531. [16] L. Chen, C.L. Wang, C.M.L. Francis, A grid middleware for distributed Java computing with MPI binding and process migration supports, Journal of Computer Science & Technology 18 (4) (2003) 505–514. [17] I. Foster, C. Kesselman, J.M. Nick, S. Tuecke, Grid services for distributed system integration, Computer 35 (6) (2002) 37–46. [18] M.T.Y. Stanley, V.L. Hong, S. Antonio, Distributed agent environment: application and performance, Information Sciences 154 (2003) 5–21. [19] Y. Xue, A.P. Cracknell, H.D. Guo, Telegeoprocessing: The integration of remote sensing, geographic information system (GIS), global positioning system (GPS) and telecommunication, International Journal of Remote Sensing 23 (2002) 1851–1893.