World Academy of Science, Engineering and Technology

International Journal of Computer, Electrical, Automation, Control and Information Engineering Vol:3, No:7, 2009

Distributed e-Learning System with Client-Server

and P2P Hybrid Architecture
Kazunari Meguro, Shinichi Motomura, Takao Kawamura, and Kazunori Sugahara

AbstractWe have developed a distributed asynchronous Web no server computers. Secondly, each exercise in the system
based training system. In order to improve the scalability and ro- is not only data but also an agent so that it has functions,
bustness of this system, all contents and a function are realized on such as scoring users answers, telling the correct answers, and
mobile agents. These agents are distributed to computers, and they
can use a Peer to Peer network that modified Content-Addressable showing some related information without human instruction.
Network. In this system, all computers offer the function and exercise The proposed system equally treats all nodes, and distributes
by themselves. However, the system that all computers do the same the functions.
International Science Index, Computer and Information Engineering Vol:3, No:7, 2009 waset.org/Publication/649

behavior is not realistic. In this paper, as a solution of this issue, However, when the system is composed of nodes which
we present an e-Learning system that is composed of computers
have a low machine specs or use a narrow bandwidth of
of different participation types. Enabling the computer of different
participation types will improve the convenience of the system. network, the performance of the entire system is decreased.
A node which executes only exercises cannot participate in
KeywordsDistributed Multimedia Systems, e-Learning, P2P, Mo-
bile Agent
the system.
In this paper, as a solution for this issue, an e-Learning
system that is composed of computers with different partici-
I. I NTRODUCTION pation type is proposed. That is to say, one system is composed
Asynchronous Web-Based Training systems (hereafter we with nodes which have same functions like a pure P2P model,
abbreviate as WBT) are very popular in e-Learning systems. another system is composed with nodes which have function
A WBT allows a learner to complete the WBT on his own or not like a client/server model. A node only offers data
time and schedule, without live interaction with an instructor. and function like a server, and another node only learns. By
A large number of studies have been made on asynchronous constructing the system which uses the nodes with only server
WBT [1] [2] [3], and all of them are based on the client/server function, even if the trouble occurs in one of the nodes that
model. The features of the client/server model are that all are compose a server group, the system will continue service. By
to execute management and to offer the exercise by the server constructing the system which uses the nodes to perform only
machine. Although the client/server model has an advantage of learning, a performance of the whole system will be stable.
easy construction and maintenance, the client/server systems This paper is organized in 5 sections. The proposed e-
generally lack scalability and robustness. There is Peer to Peer Learning system is described in Section 2. The design
(hereafter we abbreviate as P2P) model to supplement the overviews and implementation of the proposed system are
disadvantage of client/server model. The feature in the system described in Section 3 and the experimental result in Section
is based on P2P model that each computer works as a client 4. Finally, some concluding remarks are drawn in Section 5.
or a server. The feature can distribute the load to a node. The
function of the entire system doesnt stop even if some nodes
II. P ROPOSED E -L EARNING S YSTEM
break down.
We have proposed and implemented a distributed e-Learning A. Overview
system based on P2P architecture [4] [5]. The proposed e-
Learning system has two distinguishing features. Firstly, it All exercises in the proposed system are classified into
is based on P2P architecture to improve the scalability and categories, such as Math/Statistic, English/Grammar, etc.
robustness of the system. In the proposed e-Learning system, A user can obtain exercises one after another by specifying
every users computer plays the role of a client and a server. categories of the required exercises. While a user uses the
While a user uses the system, his/her computer (hereafter we proposed e-Learning system, his/her node is to be a part of
refer to such a computer as a node) is a part of the system. the system. The node receives some number of categories and
The node receives some number of contents from another node exercises from another node when it joins the system and
when it joins the system and has responsibility to send appro- has responsibility to send appropriate exercises to requesting
priate contents to the requesting nodes. In addition to the above nodes.
advantages of using P2P architecture, the proposed e-Learning The important point to note is that the categories a node
system can be constructed at low cost because the system need has are independent of the categories in which the nodes user
is interested, as shown in Figure 1. Figure 1 illustrates that
K. Meguro is with Ph.D. student in Tottori University. user As request is forwarded first to the neighbor node, and
S. Motomura is Information Media Center Tottori University, T Kawamura
and K Sugahara are with the Department of Information and Knowledge the request is forwarded to the node which has the requested
Engineering Tottori University. category.

International Scholarly and Scientific Research & Innovation 3(7) 2009 1731 scholar.waset.org/1999.4/649
 World Academy of Science, Engineering and Technology
International Journal of Computer, Electrical, Automation, Control and Information Engineering Vol:3, No:7, 2009
International Science Index, Computer and Information Engineering Vol:3, No:7, 2009 waset.org/Publication/649

Fig. 1. Migration of exercise agent.

B. P2P network

Fig. 2. P2P network of the proposed e-Learning system.

When the proposed system bootstraps, one initial node has
all categories in the system. When another node joins the
system, it receives certain number of categories from the initial
node. The categories are distributed among all nodes in the C. Components of system
system according as nodes join the system or leave the system. In order for the proposed system to be considered as a
In existing P2P-based file sharing systems, such as Nap- distributed WBT system, it is not enough that only exercises
ster [6], Gnutella [7], and Freenet [8], each shared file is are distributed among all nodes. Functions to provide the above
owned by a particular node. In the systems, files are originally services must be also distributed among all nodes. Mobile
distributed among all nodes. On the other hand, the categories agent technology is adopted to achieve this goal.
in the proposed system are originally concentrated. When a There are following agents and user interface programs on
new node joins the system, not only location information each node. These agents have implemented in the mobile agent
of a category but the category itself must be handed to the framework that we have developed [10].
new node. Considering that, the P2P network of the proposed Node Agent: Each node has one node agent. It manages
system can be constructed as a CAN [9]. the zone information of a CAN and forwards messages
Our P2P network is constructed with 2-dimensional coordi- to the Category Agents in the node.
nate space [0,1] [0,1] to store exercise categories, as shown Exercise Agent: Each Exercise Agent has questions and
in Figure 2. The figure shows the situation that node C has functions to score users answers, to tell the correct
just joined the system as the third node. Before node C joins, answers, and to show some related information about the
node A and node B shared the whole coordinate space half exercise.
and half. At that moment, node A managed Math/Geometry, Category Agent: Each Category Agent stands for a unit
Math/Statistics, and History/Rome categories and node of a particular subject. It manages Exercise Agents in
B managed English/Grammar, English/Reader and His- itself and sends them to the requesting node.
tory/Japan categories, respectively. When node C joins the Interface Agent:There is one interface agent for each
system, we assume node C already knows IP addresses of user interface, such as a student interface and an exercise
some nodes in the system and node C sends the join request manager interface on each node. It plays a role of in-
to some node in the list. Then node C is mapped on a certain terfaces between the interface program and other agents,
coordinate space according to a random number and takes on and between agents and applications.
corresponding categories from another node. For example, in User Agent:Each user has its own User Agent. A User
the case of Figure 2, node C takes on the History/Japan Agent manages its users information that includes login
category from node B, then exercises of the category move to name, password, IP address of the users computer, on-
node C. After joining, node C gets a list of IP addresses of line/offline status, and log of studying or a list of created
neighbor nodes in the coordinate space, such as node A and exercises.
node B. Therefore, neighbor nodes can communicate with each Student Interface: One student interface is on each node
other. of which a user logs in as a student. It is a user interface

International Scholarly and Scientific Research & Innovation 3(7) 2009 1732 scholar.waset.org/1999.4/649
 World Academy of Science, Engineering and Technology
International Journal of Computer, Electrical, Automation, Control and Information Engineering Vol:3, No:7, 2009

program for studying. connected node. Therefore, the Interface Agent for the User
Interface Node to exercise is prepared.
III. T HE TYPE OF PARTICIPATION NODE In the proposed system, the agent ID is essential to com-
municate with the agent. The agent ID is necessary to avoid
The following nodes is used in the proposed system. the confusion of the message, it is the same with commu-
Server Node:Only data and function are offered. nication between the user interface and the Interface Agent.
Common Node:Data and function are offered, it has In communicating, since the Node Agent is only one in each
agents system requirements, and it is operated directly server, it doesnt need the agent ID. The User Interface Node
as a learner and a teacher. requests the Node Agent in the Known Node to generate the
Client Node:Neither data nor function is offered, it has Interface Agent. Then, the Node Agent generates the Interface
the environment that agent operates, and it is operated Agent, and returns the ID to the User Interface Node. The User
directly as a learner and a teacher. Interface Node can exercise by this mechanism.
User Interface Node:Neither data nor function is offered, However, there is one issue that should be thought when
it doesnt have the environment that agent operates, and User Interface Node is designed. It is the case that the node
it is operated directly as a learner and a teacher. which generates the Interface Agent leaves. The method of
the migration to the node that integrates the DHT areas of the
International Science Index, Computer and Information Engineering Vol:3, No:7, 2009 waset.org/Publication/649

leaving node cannot be realized for two reasons. One of the

A. Design and Implementation of the Client Node
reason is that the Interface Agent exists only by one or less in
In the proposed system,the Distributed Hash Table(hereafter one node. The other is that the user interface cannot find the
we abbreviate as DHT) that does the mapping on the coordi- integrated node even if two or more the Interface Agent can
nate space two dimensions is used. The area of DHT is given exist in one node. In the proposed system, communication
to the node which participates in the system. As a result, the to the agent can be established from the user interface, the
node composes a P2P network, and the contents are managed. opposite communication cannot be established. Being not able
However, the Client Node doesnt manage the contents, to establish the communication of the user interface from the
and the function of P2P network is used. To realize the agent, it means that the Interface Agent cannot tell the user
participation type which use only the function of P2P network, interface migrating to the integrated node, as a result, the user
when the Client Node participates in the P2P network, it interface cannot find the Interface Agent.
should not receive the area of DHT. Instead of not receiving As a solution of the above issue, the generated Interface
the area of DHT, the Client Node registers the address of Agent is mapped in DHT. When the Interface Agent is mapped
the node that participates in the P2P network as a connection in DHT, the user-name is treated as a key. Then the Interface
destination. The registered node is named the Proxy Node. Agent automatically migrates the node according to the change
The Proxy Node is chosen by the following procedures. in the area. As the user interface can use the function of
When the Client Node participates in a system for the the Node Agent as long as the Known Node is identified,
first time, the Client Node connects with the node that it it becomes possible to find the Interface Agent that makes the
has already known. The node is named the Known Node. user-name a key and uses the search function of the Node
The Known Node generates a coordinate. Agent. When the User Interface Node is exercising, the User
A node including the coordinate is chosen as Proxy in Agent of a user and the Exercise Agent under exercise are sent
the domain to manage of its own. to the node where the Interface Agent exists. If the Interface
When the Proxy Node leaves, the inquiry of the Known Agent manages the User Agent and the Exercise Agent, these
Node is executed again, and the node that includes coordinate agents can be migrated together. This method makes it possible
in the area is retrieved, then it registers as a new Proxy Node. to continue exercising, since the Interface Agent can be find
When the area of the Proxy Node changes and the area doesnt even if the user interface loses sight of the Interface Agent.
include the coordinate which is generated by the Known Node,
it retrieves by the inquiry of the Known Node, and it registers IV. E XPERIMENT
as a new Proxy Node. The Client Node can use all functions This section presents the experimental results for the e-
except the content delivery. There is an advantage that there Learning system where the node with different participation
is no influence on the P2P network even if the trouble occurs type exists together. The following experiments are examined.
in the Proxy Node, for the Client Node one-sidedly executes Table I shows the machine specification.
the registration of the Proxy Node. Confirming the operation of the system
Comparison of response speeds by number of participa-

B. Design and Implementation of the User Interface Node tion nodes

The Interface Agent intermediates between an agent and a
user interface which a learner uses. If the procedure can be A. Confirming the operation of system
remotely called, which is one of the functions of the Interface The system is constructed N1:Server Node, N2:Common
Agent, users can exercise by using the Student Interface. In Node and N3:User Interface Node. TableII shows the
other words, to make the User Interface Node participate as experimental conditions. In the experiment, it aims to confirm
a learner, it only has to prepare the Interface Agent in the the following.

International Scholarly and Scientific Research & Innovation 3(7) 2009 1733 scholar.waset.org/1999.4/649
 World Academy of Science, Engineering and Technology
International Journal of Computer, Electrical, Automation, Control and Information Engineering Vol:3, No:7, 2009

TABLE I
T HE MACHINE SPECIFICATION . the interface of the learner. When the corresponding Interface
Agent left, the learners can continue studying.
CPU Intel Pentium4 3.0GHz
Memory 1GB
Network 1000Base-T
OS TurboLinux 10 Desktop

When the node with a different function in itself exists,

the category and the exercise can be made and can be
acquired.
When the User Interface Node participates in the system,
the Interface Agent corresponding to it can be made.
Even if the proxy node leaves, the User Interface Node
can be connected with a new proxy node.

TABLE II
International Science Index, Computer and Information Engineering Vol:3, No:7, 2009 waset.org/Publication/649

T HE EXPERIMENTAL CONDITIONS .

Number of node N1:49, N2:1, N3:5

Number of group 50
Number of user account student:50, teacher:3
Number of category 20
Number of exercise 60

Figure3 shows the appearance of area of DHT divided with

the node that participates, where the Category Agent and the
Interface Agent are made corresponding to a learner. The P2P Fig. 3. Division of management area and the distributed agents.
network is constructed according to the following procedures.
1) First of all, nines node of the N1 and one node of the
N2 participate in the system, then forty node of the N1 B. Comparison of the response time corresponding to number
participate. Each node is named node 1, node 2, ..., and of the Server Node
node 50 in the order of participation respectively.
2) On node 10 that is the node of N2, the system is logged The system is constructed N1:Server Node and N3:User
in as a teacher by using the user interface for the teacher. Interface Node. It aims the examination of the number of
3) The teacher makes four subjects of English, Math, Server Node of the relation to response time. The response
Biology and Physics, and makes five categories for each time in this experiment means the time which takes to display
subject, and makes three exercises for each category. after the learner request the exercise on Student Interface
4) The teacher logs out once by the system after executing The number of nodes of the N1 is 1, 2, 3, 6, 12, and 20. The
these operations, and logs in the system again as a number of nodes of the N2 is 5. The average of response time
teacher. is measured repeating five times to acquire the exercises from
all nodes at the same time. The average times are summarized
As a result, the categories and the exercises are added to this
in Figure 4.
system normally. When the teacher logs in the system again,
the categories and the exercises are found in the system. 18000
Then, the P2P network is constructed according to the
following procedures. 16000

1) Five node of the N3 participate in the system, then the 14000

learners log in each node through the Student Interface. 12000

Each node is named node 51, node 52, ..., and node55
response time[ms]

in the order of participation respectively. 10000

2) After logging in the system, each learner requests the 8000

teacher who has already logged it in to send the exercise.
6000
3) Node 7 in which the Interface Agent corresponding to
node 51 and node 53 exists, after it leaves from the 4000

system, these learners request the exercises.

2000
As shown in Figure 3, the Interface Agent corresponding
to each learner is made on node 5, node 7, node 9, and node 0
0 5 10 15 20

12 respectively. The category name of the registered exercises number of node

are obtained respectively, and the category is displayed on Fig. 4. Mean time of response.

International Scholarly and Scientific Research & Innovation 3(7) 2009 1734 scholar.waset.org/1999.4/649
 World Academy of Science, Engineering and Technology
International Journal of Computer, Electrical, Automation, Control and Information Engineering Vol:3, No:7, 2009

The case where the node of the N1 is three is excluded, Kazunari Meguro was born in 1977. He received his B.Eng. and M.Eng.
as the number of nodes increases, the response time shortens. degrees from Hiroshima city University, Japan, in 2002 and 2004, respectively.
He is currently a Ph.D. student in Tottori University. His research interests
When the number of nodes of the N1 becomes 6 or more, the multi-agent systems and distributed systems.
width of a decrease at response time becomes small.

V. C ONCLUSION
A distributed e-Learning system in which different use types
can participate has proposed. The kind of the nodes which are
implemented the proposed system have been only participate
in the P2P network as a server, participate in the P2P
network and participate as a learner, only participate as a
learner with agents system requirements and only use as a
learner.
Shinichi Motomura was born in 1973. He received his B.Eng. and M.Eng.
Increasing the number of nodes that participate as a server, it degrees in Computer Engineering from Toyohashi University of Technology,
not only has become easy to maintain the node that composes Japan, in 1995, 1997, respectively. From 2004 to 2008, he joined Tottori
the server group but also has distributed the load to each node University as a research associate. In 2008, he received the D.Eng. degree
International Science Index, Computer and Information Engineering Vol:3, No:7, 2009 waset.org/Publication/649

from Tottori University, Japan. Since 2008 he had been in Tottori University as
that composes the P2P network, and the response of the entire an associate professor in the Information Media Center. His research interests
P2P network has improved. include multi-agent systems and distributed systems.
However, in the kind of a node only use as a learner, even
if the node where the Interface Agent was mapped doesnt
leave, the Interface Agent migrate with the change of the area
of DHT, then the Interface Agent is retrieved again. Therefore,
in the e-Learning system which the area of DHT changes
frequently, the node cannot exercise efficiently . In future
work, we should verify the usability with a larger-scale system.

R EFERENCES
[1] Nishita, T. and et el,: Development of a Web Based Training system Takao Kawamura was born in 1965. He obtained his B.Eng., M.Eng. and
and Courseware for Advanced Computer Graphics Courses Enhanced Ph.D. degrees in Computer Engineering from Kobe University, Japan, in 1988,
by Interactive Java Applets, Proceedings of International Conference 1990 and 2002, respectively. Since 1994 he had been in Tottori University as
on Geometry and Graphics, Vol. 2, pp. 123-128 (2002). a research associate and has been in the same University as an associate
[2] Homma, H. and Aoki, Y.: Creation of WBT Server on Digital Signal professor in the Faculty of Engineering since 2003. His current research
Processing, Proceedings of 4th International Conference on Information interests include mobile-agent systems and distributed systems. Dr. Kawamura
Technology Based Higher Education and Training, (2003). Marrakech, is a member of IPSJ and IEICE.
Morocco.
[3] Helic, D., Krottmaier, H., Maurer, H. and Scerbakov, N.: Enabling
Project-Based Learning in WBT Systems, International Journal on E-
Learning, Vol. 4, No. 4, pp. 445-461 (2005). Norfolk, VA.
[4] Kawamura, T. and Sugahara, K.: A Mobile Agent-Based P2P e-Learning
System, IPSJ Journal, Vol. 46, No. 1, pp. 222-225 (2005).
[5] Motomura, S., Nakatani, R., Kawamura, T. and Sugahara, K.: Distributed
e-Learning System Using P2P Technology, Proceedings of the 2nd
International Conference on Web Information Systems and Technologies,
pp.250-255 (2006). Setubal, Portugal.
[6] Napster, http://www.napster.com (1999).
[7] Gnutella, http://welcome.to/gnutella/ (2000).
[8] Clarke, I., Sandberg, O,. Wiley, B. and Hong, T. W.: Freenet: A
Distributed Anonymous Information Storage and Retrieval System,
http://freenetproject.org/papers/freenet.pdf (2000). Kazunori Sugahara received the B.Eng. degree from Yamanashi University,
[9] Ratnasamy, S., Francis, P., Handley, M., Karp, R. and Schenker, S.: Japan, in 1979 and M.Eng. degree from Tokyo Institute of Technology, Japan,
A scalable content-addressable network, Proceedings of the 2001 con- in 1981. In 1989, he received the D.Eng. degree from Kobe University, Japan.
ference on applications, technologies, architectures, and protocols for From 1981 to 1994, he was on the staff of the Department of Electronic
computer communications, ACM Press, pp. 161-172 (2001). Engineering, Kobe City College of Technology. In 1994, he joined Tottori
[10] Motomura, S., Kawamura, T. and Sugahara, K.: Logic-Based Mobile University as an associate professor of the Department of Electrical and
Agent Framework with a Concept of Field , IPSJ Journal, Vol. 47, Electronic Engineering and he is a professor of the department of Information
No. 4, pp. 1230-1238 (2006). and Knowledge Engineering. His current interest lies in the fields of computer
architectures and hardware realizations of image processing algorithms. Prof.
Sugahara is a member of IEEE, IPSJ and IEICE.

International Scholarly and Scientific Research & Innovation 3(7) 2009 1735 scholar.waset.org/1999.4/649