You are on page 1of 8

Toward Agent Based E-Learning Grid Middlewares: A Layered

Architecture
Hajar Kashfi Mohammad Reza Razzazi
kashfi@qazviniau.ac.ir razzazi@ce.aut.ac.ir
CEIT Department CEIT Department
Islamic Azad University-Qazvin Branch Amirkabir University of technology
Qazvin,Iran Tehran,Iran

Abstract: E-learning users have many different interests and objectives and they will need to access a huge amount
of information. Therefore, a successful system will be one that addresses all learning issues for all types of users
across the world. Such a system should be scalable, available, interoperable, extensible, adaptable, and it should be
based on novel technologies. Elements of traditional e-learning environments such as software, hardware, and
Internet, suffer heavy workload that leads to slower computing speed. In order to solve this problem, it is necessary to
upgrade and increase the storage capacity and computing power of equipments which is barely affordable for
institutions that have limited budget.
By the aid of Grid technology, scattered computer resources can be connected via Internet links. This paper will
outline the efforts that have been done in using of grid technology as scalable, flexible, coordinated and secure
resource sharing among geographically distributed individuals and institutions in the context of e-learning. Meanwhile,
by implementing this middleware based on web service and agent technology, we are able to reuse functionalities in
a flexible and high performance manner. This way, many service and content providers can contribute to develop a
very large scale integrated e-learning system. Our middleware is implemented as a multi agent system. These agents
are located in different layers of the middleware. Each layer consists of some cooperating agents working together to
accomplish corresponding services. Using the concept of virtual organization in Grid, we can effectively group users
and organizations especially for cooperative learning.
Key Words: Grid, Resource Sharing, E-learning, Web service, Agent.

1. Introduction
E-learning systems can be broken down into four layers, namely presentation layer, common service
layer, e-learning service layer and resource layer. The presentation layer allows end-users to interact with
the service or application. The common service layer represents services that are common between all
users and are role and educational program independent; for instance, user management, user
administration, event management (All interaction between users and services is captured by the event
management service) and collaboration service. Collaboration Service supports multiple forms of
interaction and provides a backbone for learning environment. E-Learning Service layer consist of these
services: learning content management service, learning management Service (that is responsible for
delivery and administration of the course instance or offering), and learning administration service.
Resource layer is composed of learning content repository, learning metadata, learning assessment
repository, learning administration repository and user repository. More information about e-learning
layers and services can be found in (Sun, 2003).
E-Learning systems have been topics of increasing interest in recent years. The number of users
that are interested in such systems increases daily. These users have many different interests and
objectives and they will need to access a huge amount of information. Therefore, a successful system will
be one that addresses all learning issues for all types of users across the world. Such a system should be
scalable, available, interoperable, extensible, adaptable, and it should be based on novel technologies.
Since these kinds of systems are very huge, their creation can be done in such a way that many
organizations and institutes contribute to construct these systems. So, the development costs of these
systems will highly decrease. These organizations can cooperate under some policies to reach their
common objectives. Here we need a middleware for uniform access to all theses resources that belong to
different administration areas. Also, an infrastructure is needed that insure availability, scalability, security,
extensibility and interoperability for implementing such distributed systems. Each organization should be
able to reuse functionalities and services as well other resources that are provided by other organizations.

2. E-learning and Web services Convergence


As it has been mentioned above, e-learning systems consist of complex activities that interact with
each other to accomplish learning goals. Hence, recently, service oriented architectures have gained
more interests in this area. Service-based architectures take legacy application functionality and expose it
to the Internet in a reliable, highly available, scalable, .extensible, manageable, and secure manner. Web
services present another alternative distributed computing infrastructure that is strongly promoted as
preferable to the use of distributed object middlewares such as Java RMI or CORBA (Sun,2003). A web
service is a stand alone software component that has a unique URI 1. Web services are commonly based
on standards; currently, the most common ones are the XML-based specifications SOAP2, UDDI3, and
WSDL4 (Newcomer, 2002). Two advantages of web service frameworks, significantly in e-learning
context, are: First, supporting dynamic discovery and composition of services in heterogeneous
environments. Second, widespread adoption of web service mechanisms which means that a framework
based on web services can exploit numerous tools and extend services. The most critical requirement in
next generation e-learning is an infrastructure to simplify the delivery equation and lead to mainstream,
large scale, service based learning. Web services open the door to such infrastructure.
Instead of building large ,closed systems, recently, the focus has been on building flexible
architectures that provide interoperability of components and learning content and rely on open standards
for information exchange and component integration (Rodriguez,2003). Next generation e-learning
systems will use Web services that can match learning content to user context in a way that provide a
customized, personalized experience. Information retrieval (IR) functions that serve a critical role in many
e-leaning systems can be distributed or their functions can be made available through the web services
framework (Vossen, 2003). Also, by using web services, developers can provide the flexibility a learner
needs in tertiary education in which many institutions offer courses for education. Interoperability among
e-learning content and system components is a key factor in successful implementation of these
environments. Recently, there are several interoperability specifications and standards launched by a
number of organizations and consortiums (IMS, 2003).

3. Reusable learning objects


In addition to web services, extensions in the area of learning objects are based on the fact that
reusable units of learning material can be created, shared and reused between different communities. A
novel architecture based on these technologies and extensions will actually reduce the development
costs of a high-quality, large-scale learning environment. This way, the e-learning environment consists of
learning objects and services to be used during accomplishment of activities. Modern development
strategies try to separate learning objects and services from the pedagogical methods used in the unit of
learning. By sharing and reusing available teaching resources, learners can benefit from highly
personalized learning paths to fit individual requirements.
Using the theory of learning objects, creation of standard contents, searching and classifying of these
contents have been enhanced. One of the most important specifications in this area is IMS Learning
design (IMS, 2003) easing the formal description of the learning process.

4. Grid technology
Grid is coordinate resource sharing and problem solving in dynamic multi-institutional virtual
organizations (Foster, 2001). Grid computing enables aggregation and sharing of resources by bringing
together communities with common objectives and creating virtual organizations (Foster,1998a) . Data

1
Uniform Resource Identifier
2
Simple Object Access Protocol
3
Universal Description, Discovery and Integration
4
Web Service Description Language
Grids have evolved to tackle the twin challenges of large datasets and multiple data repositories at
distributed locations in data-intensive computing environments (Foster, 1998b). The Data Grid
infrastructure is designed to integrate data storage devices and data management services in a grid
environment. Data grids consist of scattered computing and storage resources, which, though located in
different countries, remain accessible to users as a cohesive environment (Chervenak, 2000). Grid
infrastructures support sharing and coordinated use of resources in dynamic global heterogeneous
distributed environments. These resources include computers, data, telecommunication and network
facilities, and software applications provided by different organizations. A Grid is a collection of distributed
computing resources available over a local- or wide-area network that appears to an end user or
application as one large virtual computing system (Yang, 2005).

Figure.1.GLOB structure (Pankratius, 2003)

5. E-learning Grid
Elements of traditional e-learning environments such as software, hardware, and Internet, suffer
heavy workload that leads to slower computing speed. In order to solve this problem, it is necessary to
upgrade and increase the storage capacity and computing power of equipments, which is barely
affordable for institutions that have limited budget. By the aid of Grid technology, scattered computer
resources can be connected via Internet links. These resources include computer facilities, storage
media, and various input and output devices. Current e-learning systems are very limit; for example, it is
vastly impossible to compute photo-realistic visualizations in real-time and display the computation result
on a remote screen (Pankratius, 2003). With the advanced functionality of an e-learning
grid(Navarr,2005), students can be provided for the possibility to search, visualize and accomplish other
high computational e-learning services in an effective manner. Grid computing provides a convenient
solution for resource and service sharing. We should notice that the software and hardware facilities
many organizations use vary greatly. This causes major difficulties in sharing learning resources. Web
service techniques enable integration of different information systems within grids and solve this problem.
Therefore, the idea behind present researches is to integrate Grid Computing and web services to build
an e-learning platform (Yang, 2005).
Till now, E-learning and grid technology were two distinct areas. However, E-learning standards and
techniques increasingly addresses learning resource sharing, interoperability and various modes of
interactions, and it is the convergence point of grid technology and distance learning (Liao,2005). In
contrast with current e-learning applications which may use high performance computing as a secondary
task, modern e-learning systems can use novel technologies such as grid and parallel computing as well
as web services to reach interactive, collaborative and reality based learning environments (Reklaitis,
2001). By proposing the use of grid technology, many service providers and content providers can
participate in various e-learning VOs. Using the concept of virtual organization in Grid, we can effectively
group users and organizations especially for cooperative learning. Organizations and institutes that
contribute to develop such a system can share their resources and services forming different VOs. These
organizations can cooperate under some policies to reach their common objectives. Here, we need a
middleware for uniform access to all theses resources that belong to different administrative areas.
Additionally, an infrastructure is needed that insure availability, scalability, security, extensibility and
interoperability needed for implementing such distributed systems. Each organization should be able to
reuse services as well learning resources provided by other organizations.

6. Grid based reusable learning objects


Recently, a model permitting integration of grid technology and e-learning systems has been
proposed. This model introduces a new paradigm for reusable learning objects in grid named GLOB.
GLOB structure has been shown in figure1.” the structure of GLOB is designed in such a way that it can
contain both the conventional e-learning content and content that uses the grid functionality” (Pankratius,
2003:9). As it has depicted in that figure, the GLOB consist of several parts: overview of the lesson,
metadata, several reusable information objects (ROIs) and a summery.
This definition combines both the functionalities of a traditional learning object and the grid computing
power by adding grid functionality consisting of a specific grid application layer and a user interface to
traditional learning object. A GLOB is wrapped by a web service which let it easily integrate into learning
management systems. More information about this concept can be found in (Pankratius, 2003).

7. Agents as web services


Agents are intelligent interactive entities that are part of an environment shared with other agents.
These agents are capable to communicate and cooperate with each other (Luck, 2002). A multi agent
system is a loosely coupled network of intelligent agents working together to reach a common goal.
A number of activities have been initiated recently to apply agents to computational grids. Almost all
off these efforts tried to use agent technology from computational perspective (Avila-Rosas, 2003). They
use agent powers to manage distributed and parallel computations, scheduling the usage of grid
resources, and so on. Some of these efforts are described in (Avila-Rosas, 2003) and (Jennings,
2001).Grid and agent communities are both pursuing the development of open distributed systems, albeit
from different perspectives. Grid has a large number of requirements such as interoperability, platform
independent, dynamic discovery, etc. These requirements have been addressed by Multi Agent Systems
for open distributed systems before.
In order to build an open, large scale and interoperable multi agent system in the context of grid
computing, some researches are looking at integration of agent technologies and web services (Avila-
Rosas, 2003). These researches focus on technical aspects of creating, deploying, and publishing agents
as web services. Agent behaviors can be described advertised and discovered using web service
technology. WSDL is used to describe agent behavior, which can then be advertised in UDDI registries as
regular services.
On the other hand, in e-learning systems, agents play a great role in implementing services (Suzuki
,2002a); especially because of the nature of agents, in modeling real world people manners and
situations. Up to now, many systems have been developed that use agent mediated mechanisms to
create and deliver educational materials effectively as well as to reach hardware and software
heterogeneity in e-learning areas. Using agents, delivery of courseware and learning materials to users,
also automatic managing of metadata can be handled very effectively (Suzuki, 2002b).
8. Proposed E-Learning Grid Middleware
As mentioned above, e-learning platforms generally need a high capacity storage device to store
learning content and it is not affordable for most institutes. One solution is to connect computers used
by organizations for teaching or administrating together and combine their idle resources by means of
grid technology. Our middleware is a multi agent system which represents grid core services for
resource sharing and other basic services for e-learning. Each agent in this middleware is created,
deployed and published as a web service. Proposed middleware is responsible for management of
distributed learning resources and also discovery and selection of learning services. Experiments
show that in an agent based grid computing environment, transparency of the distributed environment
and system performance will improve (Avila-Rosas,2003), (Jennings,2001). Agents can communicate
and negotiate to establish an appropriate connection between a service provider and a learning
application. Learning object repositories and other learning services can be fitted in Grid framework to
act as grid services. Using the concept of virtual organization in Grid, users and organizations can be
effectively grouped especially for cooperative learning. Based on different roles specified in e-learning
environments, organizations and institutes that contribute to develop such a system can share their
resources and services with each other forming different VOs such as publishing organization,
Tutoring organization, Brokering organization, Training organizations, Research organizations. The
topology of this middleware can be seen in figure 2.

managerAglet

ManagerAglet host replicaManagerAglet

brokerAglet

metadataManagerAglet
mediatorAglet

mediatorAglet host mediatorAglet host


mediatorAglet host
resourceManagerAglet
searchAglet

fabricAglet host
fabricAglet

EL Client agent
Figure 2. Middleware topology

As mentioned before, our middleware is implemented as a multi agent system. There are ten agents
in above figure as you can see (To prevent complicating the picture, not all agents in the middleware have
been shown). These agents are located in different layers of the middleware. proposed middleware has a
multi layer architecture. Each layer consists of some cooperating agents working together to accomplish
corresponding services. Layers of the middleware are shown in figure 3. More information about these
layers and their services are discussed in the following sub sections.
The most important agents in the middleware are Manager, Mediator and Fabric agents which are
responsible for establishing the grid topology. Since we have implemented these agents in aglet
environment (Aglet, 2006), we called them ManagerAglet, MediatorAglet and so on. Aglet is an
environment for implementing mobile agents using java programming language. Aglet gets its name from
java Applets. fabricAglet migrates to the client machine as the client login to the grid and connects its host
to other machines existed in the grid. This mobile agent is able to interact mediator agents and other
agents serving grid and e-learning services. Each administrative domain must have a server for managing
resources existing in that domain. to do that MediatorAglet is installed on the server and act as a local
manager for fabric agents in that domain. mediatorAglets are able to communicate to other mediators so
they can cover the hole grid consisting of multiple administrative areas. ManagerAglet is responsible for
managing mediatorAglets,creating them and sending them to local mediator servers. These three agents
are in grid core services layer. Broker agent and resource manager agents are in this layer too.
Additionally, other agents that are responsible for converting conventional agents to web service agents
are located in this layer.

8.1. Grid core services layer


Agents that are responsible for dynamic adding, removing and managing resources in the grid are
located in this layer. We introduced some of these agents like fabric, mediator and manager agents
before. The first step in developing the architecture is to provide a grid infrastructure. Users incrementally,
add their hosts to the grid. On the other hand system administrators can add new mediator servers to the
system anytime needed, and therefore extend the system to desired scale. In this architecture the entire
grid are divided to multiple administrative areas. In each area, there is one mediator server, that response
to all requests of clients in that area. There are multiple agents hosting on each mediator server. The
most important agent is mediator agent that is responsible for coordination of other agents. This agent
connects fabric agents which are under its control to the entire grid. Also this agent is responsible for
creating other agents that should exist on each mediator server to provide e-learning and Grid core
services.

Figure 3. EL middleware Layers

8.2. E-Learning basic Services layer


This layer is the heart of our grid middleware. In this layer, agents represent basic services for e-
learning systems using other agents that exist in first layer (grid core services layer). Applying these
agents, users are able to access and share all types of information including papers, magazines, e-books,
etc. with other users. Except sharing other resources, any user can contribute in publishing learning
objects in this system. On the other hand, any organization can publish its learning content for its
members as well as other organizations. This layer consists of services including searching learning
objects, user tracking, content management (including metadata management for learning contents),
content packaging, rating services for learning objects, and alerting services. There are multiple agents
cooperating to perform desired services. Some of these agents are ELClient Agent, SearchManager
Agent, MetadataManager Agent, ContentManager Agent, Usertracking Agent, LogManager Agent,
NotificationManager Agent and VOmanager Agent.
8.3.E-learning additional Services
Each organization can add its e-learning special purpose services in the form of new agents to the
most top layer in the architecture named e-learning additional services layer. New agent should be
presented to an agent responsible for converting typical agents to web service agents. We called this
agent gateway agent. Gateway agent is placed in Grid core services layer.

9. Conclusion
This paper proposes an integrated agent based middleware architecture for e-learning on the Grid .We
described how new technologies including agent, web service, Grid and emerging standards can be
applied to E-learning. Recently, most of E-learning systems are being designed based on client/server,
peer to peer; and recently Web Services architectures. These systems have major drawbacks because of
their limitations in scalability, availability, distribution of computing power and storage system, as well as
information sharing between users that contribute in these systems. By proposing the use of grid
technology as scalable, flexible, coordinated and secure resource sharing among geographically
distributed individuals and institutions, in the context of e-learning ,we are able to address these
concerns. Meanwhile by implementing our middleware based on web service technology we are able to
reuse functionalities. This way, many service providers and content providers can contribute in developing
a very large scale integrated e-learning system that leads us to a cooperative learning environment. The
content provider in the simplest case can be a user who shares information with other users.
In e-learning systems, agents play a great role in implementing services such that they can response to
users in an effective manner. Hence, we have implemented our middleware using agent technology, while
we tried to represent each agent as a web service in our architecture. Our middleware is a multi agent
system which represents grid core services for resource sharing and other basic services for e-learning,
all at the same time. Each agent in this middleware is created, deployed and published as a web service.

10.References
Sun Microsystems (2003), E-learning Framework, Technical White Paper,Available : www.sun.com/products-n-
solutions/ edu/whitepapers/pdf/framework.pdf.

Booth.D,Haas.H,McCabe.F,Newcomer.E,Champion.M,Ferris.C.and Orchard.D.(2003) “Web Services Architecture”


World-Wide-Web Consortium (W3C), Available http://www.w3.org/TR/ws-arch/

Resmer.M (1998)”Internet Architectures for Learning” IEEE Computer, vol. 31, No. 9, September 1998.

Newcomer, E. (2002) ”Understanding Web Services: XML, WSDL, SOAP, and UDDI”, Addison-Wesley Professional;
1st edition (May 13, 2002).

Rodríguez.J, Anido.L ,Fernández.M.J(2003) “How can the Web Services Paradigm improve the E-learning?”,
Proceedings of the The 3rd IEEE International Conference on Advanced Learning Technologies (ICALT’03).

Vossen, G., P.Westerkamp (2003)” E-learning as a Web service” (extended abstract). In Proc. 7th International
Conference on Database Engineering and Applications (IDEAS), Hong Kong, China, IEEE Computer Society Press,
pp. 242-249,2003.

IMS consortium (2005),“IMS Content Packaging Specifications “ , Available


http://www.imsproject.org/content/packaging/index.html

Foster,I.and Kesselman,C.and Tuecke ,S.(2001)”The Anatomy of the Grid:Enabaling Scalable Virtual


Organizations”,International journal of high performance Computing applications,15(3).200-222.2001.

Foster,I.and Kesselman,C.(1998) “The Grid:Blueprint for a New Computing Infrastructure”,Morgan Kaufmann


publishers,San fransisco,1st edition (November 1, 1998).

Foster,I.,Kesselman,C.and Tuecke ,S.(1998)The Physiology of the Grid: Open Grid Services Architecture for
distributed systems Integration,”, Proc. 4th Global Grid.
Chervenak.A,Foster.I,Kesselman.C, Salisbury.C, and Tuecke.S (2000) ”The data grid: Towards an architecture for the
distributed management and analysis of large scientific datasets”, Journal of Network and Computer Applications, vol.
23, no. 3, pp. 187–200.

YANG.C,HO.C (2005)An e-Learning Platform Based on Grid Architecture, JOURNAL OF INFORMATION SCIENCE
AND ENGINEERING 21, 911-928.

Liao.C,Yang.F,Hsu.K.C (2005)“A service-Oriented Approach for the Pervasive Learning Grid”, Journal of Information
Science and Engineering 21,959-971.
Reklaitis.V,Baniulis.K,Okamoto.T,(2001)“Shaping E-learning applications for a service oriented grid”, 2nd International
LeGE-WG Workshop on e-learning and Grid Tchnologies: a fundamental challenge for Europe.

Pankratius.V,Vossen.G, (2003)”Towards E-Learning Grids: Using Grid Computing in Electronic Learning”, Proc.IEEE
Workshop on Knowledge Grid and Grid Intelligence.

Navarr.L.I,Diaz.A.S,Such.M.M(2005) ”Learning units Design based in Grid computing”,Journal of Interactive Media in


Education 2005(10)

Luck, M., McBurney, P. and Preist, C.(2003)”Agent technology: Enabling Next Generation Computing”,Available
www.agentlink.org/roadmap/al2/roadmap.pdf.

Avila-Rosas.A,Moreau.L,Dialani.V,Miles.S,Liu.X;(2003) “Agents for the Grid: A comparison with Web Services”, In


IEEE International Symposium. on Cluster Computing and the Grid, Available www.agentcities.org/Challenge02/
Proc/Papers/ch02_51_avila-rosas.pdf.

Jennings, N.R. (2001) ”An agent-based approach for building complex software systems” Communications of the
ACM, 44(4). 35-41.

Suzuki.J,Yamamoto.Y,(2002) “Building Next-Generation Infrastructure for Agent-based Distance Learning”,


International Journal of Continuous Engineering Education and Life-Long Learning, Vol. 12.Nos. 1-4.

IBM corporation (2006), Aglet Software Development Kit, Available: http://sourceforge.net/projects/aglets/

You might also like