AN APPLICATION OF REMOTLEY CONTROLLED
EXPERIMENTS TO PERFORM FEEDFORWARD AND
FEEDBACK DAMPING CONTROL OF AN ELECTRO
MECHANICAL SERVOMECHANISM
Andrew P. H. Weightman, Peter R. Culmer, Martin C. Levesley
School of Mechanical Engineering, The University of Leeds, Leeds, United Kingdom
Ben Hanson
School of Mechanical Engineering, University College London, London, United Kingdom
Keywords: e-Learning, Remotely Operated Experiments.
Abstract: This paper presents the findings of a pilot study to examine the possibility of delivering a control and
dynamics laboratory session, using equipment located at the University of Leeds, to students at University
College London (UCL), utilising the ReLOAD (remote access to real experimental equipment) system. The
ReLOAD system has been developed to provide improved access to experimental equipment, used as part of
undergraduate engineering degree programmes. Key to its operation is the availability of the system to serve
video clips that record the experiment in progress, providing essential visualisation plus an improved sense
of realism for the user. Students at UCL remotely performed an experiment to determine the effect of
varying the feedforward and feedback gain on the dynamics of an electro mechanical servomechanism.
Evaluation of the ReLOAD system, performed by the module leader at UCL, shows the system has huge
potential to allow more efficient and collaborative use of laboratory equipment across the global
engineering education community.
1 INTRODUCTION
Laboratory based teaching sessions are a key
element of teaching engineering during an
undergraduate degree program in the United
Kingdom. These laboratory sessions provide an ideal
opportunity to enthuse, motivate and inspire students
through face-to-face interaction with teaching staff
and real experimental equipment. However these
sessions are costly, both in terms of equipment and
time taken by teaching staff to deliver them. In the
UK, a typical face-to-face session may last 2-3 hours
and consist of a group of 20-30 students, working in
pairs or small groups within a laboratory
environment. Often multiple copies of experimental
equipment are required for these laboratory sessions,
limiting the quality of the equipment that can be
deployed. Even with group sizes of 20-30 students,
laboratory sessions often need be repeated many
times in a given year to ensure all students on a
particular program are exposed to the session. Even
with multiple copies of equipment and duplicate
laboratory sessions, students can find that access to
equipment is limited, as often the laboratory is
running a different class or teaching staff are
attending to other commitments. Students can
potentially miss out on laboratory sessions due to
illness, or be prevented from repeating a laboratory
session to collect more data because of such
restrictions. Laboratory sessions can be viewed as
being inefficient, since although expensive
equipment is used intensively for a period, it may
remain unused for most of the year.
In 2000, to overcome some of these limitations,
a system called ReLOAD (Real Labs Operated at
Distance) was developed within the School of
Mechanical Engineering at the University of Leeds,
which allows access to experimental equipment via a
web interface. Although the system cannot be used
to develop practical skills such as the use of
equipment, it does allow students to develop skills
such as planning experiments, collecting and
419
P. H. Weightman A., R. Culmer P., C. Levesley M. and Hanson B. (2007).
AN APPLICATION OF REMOTLEY CONTROLLED EXPERIMENTS TO PERFORM FEEDFORWARD AND FEEDBACK DAMPING CONTROL OF AN
ELECTRO MECHANICAL SERVOMECHANISM.
In Proceedings of the Third International Conference on Web Information Systems and Technologies - Society, e-Business and e-Government /
e-Learning, pages 419-426
DOI: 10.5220/0001278204190426
Copyright
c
SciTePress
analyzing data, and conducting further experiments
if required. It offers greater accessibility, can be run
24 hours a day, 365 days a year and allows many
students to access a single piece of high quality
equipment.
Other institutions across Europe the United
States, Australia, Asia etc. are also developing
similar systems to allow web based access to
experiments, many as part of consortia, to promote
sharing of equipment. Like the MIT iLabs system,
ReLOAD has been used recently to allow students
from one institution to use equipment in another,
many time zones away. This enables the system to
be used at times when normally it would be
underutilized. In some of the systems developed,
attempts have been made to reproduce face-to-face
laboratory sessions as closely as possible, in some
cases by employing technician staff to be available
in the laboratory at a prearranged time, to help
conduct the experiment remotely using
telecommunication links. In contrast to this
approach, ReLOAD aims to convince students that
this is a physical experiment and not just a
simulation, by providing highly visual experiments
with the content being visualized using video clips
taken from a webcam. It is an important aspect that
students should be able to perform the entire
experiment, without the need for technician
intervention, allowing the experiment to be
undertaken at a time that suits the student. This
approach is the same as that adopted by the State
University of New Jersey (RUTGERS), amongst
others.
The ReLOAD system is now sufficiently
developed to allow the pilot study presented in this
paper to be conducted, which examines the system’s
suitability as an e-learning tool. Student feedback
has so far been very positive, with students
particularly appreciating the virtually unlimited
access and the ability to easily visualize the output
via the use of the video clips, simultaneously with
near real-time analogue data and the reinforcement
of the sense that they are accessing real experimental
data that these clips provide.
2 RELOAD SYSTEM OVERVIEW
The idea of remote access to experiments is that the
physical experimental equipment does not need to be
in the same location as the person performing the
experiments. Hence a greater flexibility in
performing experiments is gained, although there is
the danger the user may doubt the realism of the data
or feel detached from the experiment being
performed no matter how good the interface (Selmer
et al., 2005). The Internet or LAN network is ideal
for the transfer of data in such cases and though
several different methods can be employed to deliver
remote experiments, the basic structure adopted here
is as shown in figure 1 (Levesley, 2006).
Figure 1 illustrates that a client computer can
send a request to a central web server (the ReLOAD
web server) for experimental data across the Internet
from any location with an Internet connection. The
ReLOAD web server interprets the request and
redirects this request via a LAN to one of several
Experiment Server computers physically connected
to either single or multiple pieces of experimental
Internet
Client
computer
ReLOAD
Web Server
Experiment
Server 1
Equipment 1.1
Equipment 1.2
Hard Wired
Connection
Experiment
Server 2
Equipment 2.1
Equipment 2.2
Figure 1: ReLOAD Remote Experiment Structure.
LAN
WEBIST 2007 - International Conference on Web Information Systems and Technologies
420
equipment. The request is then converted into a
series of commands appropriate for the specific
equipment being run by the Experiment Server.
When the experiment requested is completed the
Experiment Server computer sends results in the
form of data images and video clips back to the user
on the client computer via the ReLOAD web server,
providing virtually the same level of information as
would be obtained by performing the experiment
locally.
2.1 The Web Server
The web sever is set-up to receive and process
requests from the client, and since the web server
can be a single computer, custom software can be
readily created and installed to monitor internet
requests. The client computer may be one of
hundreds of different PC's and the manner in which
data is transmitted and received from the client
computer has significant impact on the overall
functionality and versatility of the final system.
Several alternative methods have been proposed
previously. An Active-X control panel could be
installed on the client computer to send and receive
information (Hites, 2002) through an Active-X
enabled browser. Alternatively a Java applet could
be created to run on a standard Internet browser to
control the experiment (Rohrig & Jochheim, 1999,
Sanchez et al., 2000). The ReLOAD system
simplifies all aspects of the remote experiment to
create a reliable, maintainable system. A standard
web form is used at the client end to request
experiments and retrieve and view the associated
results. At the server end, LabVIEW software
performs the experiments while monitoring,
receiving and sending data across the internet to the
client computer. The client computer only requires
standard java support to view the video feedback.
2.2 Controlling Equipment
The Experiment Server computers perform the
experiments and communicates across a LAN and in
turn pass the data on via the internet to the client PC.
Using data acquisition hardware, LabVIEW
software is used to control the experimental
equipment. The use of a single programming
environment and language simplifies the
development process.
The front panel of the Experiment Server
computer displays all parameters used by the
experiment, together with a preview window for the
video camera and the results graph. Hence the
experiment server computer functions as a
demonstration display, allowing staff, students and
visitors within the School of Mechanical
Engineering in Leeds to observe the experiments
being requested remotely. A tabbed panel on the
computer displays either a log of the requests
submitted to the system or the setup panel. A
shutdown button is incorporated into the design such
that it ensures that any requests being processed are
first dealt with before the ‘shutdown’ state.
2.3 System Communication
LabVIEW allows two alternative methods to
communicate the required information between
computers to perform remote experiments LabVIEW
HTTP Server with Common Gateway Interface
(CGI) Scripts or embedded ActiveX control panels.
The LabVIEW Web Server and embedded panel
method has several shortcomings. Firstly, it would
be necessary to purchase as many licenses as the
number of users that may be expected on the
experiment. Secondly, the provision of video; since
continuous experimental interaction is possible it
would be consistent to use streaming video footage,
giving a ‘live’ indication to the state of the
experiment. However, the quality of the video
footage is dependant on the connection speed.
Finally, this method relies on proprietary plug-ins in
order to embed the VI’s panel in a web-page.
The HTTP Server with CGI scripts method does
not encounter these problems, and justifies its use
for this application. The HTTP server does not
require a license dependant upon the number of
users expected and only a widely supported, multi-
platform java plugin is required, in this case to
display the video. It is important to note that only
experiments not requiring continuous interaction are
possible using this method.
The LabVIEW Internet Toolkit, an add on to the
standard LabVIEW package, provides a HTTP
server and a set of related Virtual Instruments (VI’s).
VI’s is the term used to refer to units of code that
perform a particular function. Of particular
relevance to this application are the CGI script VI’s.
CGI scripts are typically used in association with a
form on a web-page. Forms are defined using
standard HTML elements and allow a user to enter
information into a number of fields and post it to a
server. The server receives the post and invokes the
relevant CGI script. The script then processes the
information and sends back a response, typically a
web-page, through the server to the user. Note, once
the user has posted a form the web-browser will
AN APPLICATION OF REMOTLEY CONTROLLED EXPERIMENTS TO PERFORM FEEDFORWARD AND
FEEDBACK DAMPING CONTROL OF AN ELECTRO MECHANICAL SERVOMECHANISM
421
await a response from the server and display it once
received.
It is possible to achieve the required interaction
between the user and the experiment by combining
the LabVIEW HTTP Server and the CGI script VI’s.
The experimental parameters can be entered into a
form on a web-page and posted to the LabVIEW
HTTP Server. This will then invoke a CGI script VI
that interprets these parameters and uses them to
conduct the experiment. The results will be returned
in the form of a web-page to the user when the
experiment is complete. This approach is illustrated
in figure 2. The CGI script VI’s can only be used in
conjunction with the LabVIEW HTTP Server. The
communication between the web browser and HTTP
Server can be over any network supporting HTTP
traffic, for example a departmental LAN or the
Internet. LabVIEW (8.0) contains tools aimed at
making it easier to control VI’s using a web-
browser. The front panel of a VI can be embedded
into a web-page. It takes the form of an ActiveX
component if Internet Explorer is being used, or a
plug-in for Netscape. The computer running the VI
must also run the LabVIEW Web Server which
controls communication between the embedded
front panel in the web-page and the actual front
panel of the VI. The user can view the front-panel or
request to control it. When viewing, the user sees a
‘live’ representation of the front panel. When
controlling, the user can do everything that would be
possible if using the actual VI, for example altering
controls or pressing buttons.
2.4 Experiment Request Control
As each experiment conducted will take a finite
amount of time, and multiple requests may be
submitted by different users at an instance in time, a
queuing system is used to ensure all experiments are
performed in an orderly manner.
Both incoming experiments and outgoing results
are stored in queues. When the server receives data
posted from the ‘input parameters’ web-page it is
added to the back of the request queue. When the
experimental apparatus is ready to conduct an
experiment it removes the oldest request from the
queue and uses the parameters given. The
experiment results are placed at the back of the
results queue once the experiment is completed.
Both queues operate on a first in first out basis. Each
item in the request queue contains a number of
elements. Firstly are the IP Address that the request
was submitted from and the CGI connection
information, such that the results can be sent back to
the user. The experiment name is sent as a text string
to provide for the possibility of running multiple
experiments on the same machine.
Finally, the parameters sent from the web-page
form are included as a keyed array. This is a cluster
of two arrays, the first containing the parameters
names (called keys) and the second their values. The
‘keys’ are simply the names given to the individual
fields in the web-page form. Each item in the
Results queue consists of two elements. Firstly the
CGI connection information required to send the
results back to the user. Secondly, a text stream
containing the results web-page location.
The queue size is limited to prevent large
numbers of requests amassing under heavy demand.
A default value of 10 was used for the maximum
number of requests in the queue, an acceptable
compromise between allowing multiple users and
minimizing the delay before results are returned. If
the queue is full when a request is submitted, the
user is sent a ‘busy’ web-page in reply, the request is
then discarded. A further check can be performed to
check if the IP address of a submitted request is the
same as one currently in the queue. If this is the case
the request is discarded and a single user is
prevented from dominating the queue system.
3 EXPERIMENTAL EQUIPMENT
The experiment chosen for this pilot study of
ReLOAD, to be remotely delivered to students on
E467 Automatic Control at UCL, was one which
Results
Experiment
Parameters
http://www.leeds.ac.uk/
Remote Control
Experiment
Amplitude 5.00
Frequency 100
Submit
LabVIEW
HTTP
Server
CGI Script
VI’s
Figure 2: ‘Labview HTTP server with CGI script’
method.
WEBIST 2007 - International Conference on Web Information Systems and Technologies
422
enables investigation of the performance of a closed
loop position servomechanism, as the forward path
gain and the velocity feedback gain are varied. The
purpose of the servomechanism system is to keep
the angular position of the Output shaft (output dial)
in correspondence with that of a Reference or
Demand shaft. The purpose made equipment, used
for the remote experiment was designed and built by
staff and students of the School of Mechanical
Engineering at Leeds and is based on equipment
used in face-to-face laboratory sessions in the
second year of an undergraduate degree programme.
It comprises the Servomechanism (motor, position
potentiometer, tachogenerator etc) and Controller.
Figure 3 shows photographs of the original face-to-
face equipment (upper photograph) and the
adaptation of the controller (lower photograph) to
allow its use as a ReLOAD experiment.
A block diagram for the experimental system is
shown in figure 4. When a voltage Demand or Input
signal is applied to the system it passes to the
Forward path potentiometer with variable gain K
a
,
set by the user. The signal is then sent to an
amplifier which is then applied to the motor. A
tachogenerator measures the velocity of the motor;
within the VI the magnitude of the velocity feedback
is calculated dependant on the value of a gain K
t
, set
by the user. The position of the motor is measured
via a potentiometer, with gain K
p
. This experimental
system allows the user to vary the forward path gain,
K
a
, and the feedback gain, K
t
, and observe the
response of the system in terms of position and
velocity of the servomechanism, hence illustrating
important control concepts. The magnitude of the
position feedback gain is kept fixed at 1 and can not
be altered by the user.
The experiment control software used was
LabVIEW 8.0, the internet server software was
LabVIEW HTTP server 8.0. and the experiment
server was a Pentium 4 cube PC.
Figure 3: The face-to-face experimental rig (upper) and
adaptation for the ReLOAD enabled version (lower).
4 THE CLIENT COMPUTER
The only requirement for the client computer is that
it has appropriate Internet browser software
installed, as only relatively small amounts of data
will be sent to the ReLOAD web server. Upon
entering the experiment internet address, the user is
Output dial
Forward path
potentiometer
Servomechanism
Connections to
Data acquisition
card
Velocity Feedback
Potentiometer
Demand / Input
signal
K
a
K
g
1
s
K
p
volts
volts
speed
position
Output Position
θ
(s)
Tachogenerator
Input Voltage
V(s)
K
t
)s1(
m
m
τ
+
K
Integrator
Potentiometer
(Position Feedback)
Gear Box
Amplifier
& Motor
position
volts
volts
Potentiometer
+
-
+
-
Figure 4: The experiment block diagram.
AN APPLICATION OF REMOTLEY CONTROLLED EXPERIMENTS TO PERFORM FEEDFORWARD AND
FEEDBACK DAMPING CONTROL OF AN ELECTRO MECHANICAL SERVOMECHANISM
423
presented with the Parameters Input web page as
shown in figure 5.
A standard HTML form is used to submit
information to the ReLOAD Web server, with
parameters entered into the text boxes, and
submission activated using the submit button.
Usually a submit button will directly post the
information in a form to the defined address.
However, to prevent the user repeatedly pressing the
submit button, a java-script function first checks to
see if the user has already pressed the submit button.
The form will be submitted only if the function
returns a Boolean ‘true’. Note it would be possible
to remove this java script and use the server
computer to delete duplicate experiment requests.
The results web-page presents the results from
the experiment to the user in a number of ways. A
graph is displayed showing the input signal to the
servomechanism, its displacement, via the voltage
output from a potentiometer, and its velocity,
represented by the voltage output of a tachometer.
An embedded video clip of the experiment and a set
of standard controls allowing replays, rewinding and
fast-forwarding are also displayed. The video is
displayed using a java applet providing a consistent
experience across different browsers and does not
require the user to download any special plugins. A
link to the results Comma Separated Values (CSV)
file is provided allowing them to be imported into
spreadsheet applications such as Excel and a copy to
clipboard function is also included, which serves a
similar purpose. The results web-page is shown in
figure 6.
5 THE EXPERIMENT
Both the face-to-face and ReLOAD delivered
experiments involve two components both utilising a
square wave input signal to drive the
servomechanism. The ReLOAD interface enables the
period of time over which experimental data is
collected to be set as well as the option of a video
clip captured using a standard USB webcam (60fps
320x240). Visualisation of the servomechanisms
motion is a crucial component of understanding the
mathematical parameters used to characterise
dynamic systems.
The first component of the experiment
investigates the effect of varying the forward path
gain, K
a
, with no velocity feedback, that is K
t
set to
zero. The user is required to determine the
magnitude of K
a
which gives the fastest possible
settle time, that is the critically damped state. The
user then chooses two values of K
a
above and two
values below this critical state and repeats the
experiment. The user should observe that as K
a
is
decreased below the critical value, the settle time
increases, there is no overshoot and the damping is
greater than one. Whilst as K
a
is increased above the
critical value the settle time and overshoot increase
whilst the damping decreases. Students should
observe that varying K
a
does not alter the magnitude
of the steady state gain. Figure 7 (upper three
graphs) illustrates the response of the
Figure 6: The Results Web-Page.
Figure 5: The Parameter Input Web-Page.
WEBIST 2007 - International Conference on Web Information Systems and Technologies
424
servomechanism system, in terms of displacement,
to increasing values of K
a
when K
t
is zero. The
figure shows two and a half seconds of data with the
first second omitted so as not to show the initial
transient state exhibited by the system, hence
enabling a clearer comparison.
1 1.5 2 2.5 3 3.5
-2
0
2
1 1.5 2 2.5 3 3.5
-2
0
2
1 1.5 2 2.5 3 3.5
-2
0
2
1 1.5 2 2.5 3 3.5
-2
0
2
1 1.5 2 2.5 3 3.5
-2
0
2
1 1.5 2 2.5 3 3.5
-2
0
2
The second component of the experiment
investigates the effect of varying the velocity
feedback gain K
t
with the forward path gain, K
a
, set
to a magnitude of 0.5. The user is required, as in the
first component, to determine the magnitude of K
t
which gives the fastest settle time without any
overshoot, the critically damped state. The user
again chooses two values of K
t
above and below and
the critical state and observes the percentage
overshoot, settle time, frequency of oscillation and
damping ratio. The students should observe below
the critical damped state as K
t
is decreased the level
of damping decreases, the settle time, overshoot and
the frequency of oscillation increase. Whilst as the
magnitude of K
t
is increased above the critical value
the level of damping increases, the settle time
increases and there is no oscillation and the settle
time increases. Figure 7 (lower three graphs)
illustrates the response of the servomechanism
system, in terms of displacement, to increasing
values of K
t
when K
a
is set to 0.5.
6 EVALUATION
The particular remote experiment described in this
paper has been used successfully since 2005 as an
alternative to an almost identical face-to-face
version of the Position Servomechanism experiment.
During 2006 the ReLOAD Position Servomechanism
experiment was utilised by number of students at
UCL School of Mechanical Engineering as part of
an assignment.
A servo-motor lab has run at UCL for at least 20
years; 6 sets of experimental apparatus were
originally purchased, and these have been used by
groups of 2 or 3 students at a time. Since its original
purchase, the experimental procedure had become
somewhat outdated: students were asked to change
gains on the controller by interchanging large wire-
wound resistors. Considerable time has been
required to set up the experiments and return them to
storage each year, in addition to the regular
maintenance required to repair natural wear and tear.
In 2006 there was a further pressure of space, as the
control laboratories were being refurbished and
refitted. The combination of these factors meant that
the internet-based laboratory was a very attractive
alternative.
Compared to the traditional laboratory
experiment, the internet-based version has been
continually updated in order to demonstrate the most
current experimental procedure – this was tailored to
be specifically appropriate to the particular course
being taught.
Operating over the internet gave students the
flexibility to study alone or in groups. The lecturer
offered a “surgery session” in a computer cluster,
after a scheduled computer-based class where
students had the opportunity for personal feedback
and guidance in running the experiment. The
queuing system described above worked flawlessly,
and the brief duration of the experiment (approx.10
seconds) meant that students rarely had to wait more
than 20 seconds for their results.
Outside this scheduled session, the system had
the benefits of increasing available experimental
time for the more interested students, and also
increased participation among the less-motivated
students. The experiment was available over the
Figure 7: The servomechanism response to variation of
the forward path and feedback gains.
a
b
c
d
e
f
K
a
= 0.015
K
a
= 0.030
K
a
= 0.050
K
t
= 0.1
K
t
= 0.4
K
t
= 0.6
Effect of varying the forward path gain, K
a
, excluding
negative velocity feedback, k
t
= 0.
Effect of varying the negative velocity feedback gain K
t
,
with K
a
= 0.5.
AN APPLICATION OF REMOTLEY CONTROLLED EXPERIMENTS TO PERFORM FEEDFORWARD AND
FEEDBACK DAMPING CONTROL OF AN ELECTRO MECHANICAL SERVOMECHANISM
425
Christmas holiday period, and this gave an
opportunity for students to repeat and check their
results as they wrote their reports. Students
particularly appreciated this opportunity to act on
feedback and guidance from their tutors before
submission of their reports.
An important aspect of the evaluation of this
form of remote experimentation is whether the
learning outcomes of those students undertaking the
experiments remotely are different form those
undertaking face-to-face experiments. A study to
investigate this in detail, would need careful design
to ensure valid comparison. To date, the experiments
performed at UCL and Leeds, though using the same
basic equipment and sharing the same broad
objectives, have been tailored to the individual
institution’s curriculum needs. This has resulted in a
the two institutions setting different learning
objectives for their sessions. This in some way
illustrates the flexibility of the ReLOAD system, but
makes a comparative study problematic. Further
experiments using this system are planned, that it is
hoped will allow this type of direct comparison to be
made.
7 CONCLUSIONS
This paper has illustrated the operating principles of
the ReLOAD system in general and for the remote
delivery of a position servomechanism experiment,
taught face-to-face at The University of Leeds, to a
number of students at University College London.
The pilot study has proved the efficacy of the
ReLOAD system, with both students and lecturers
finding a useful educational tool.
Further development of the ReLOAD system is
underway with a view to improved visualisation and
the delivery of alternative experiments using the
same experimental equipment, alongside the
development of new equipment. Furthermore, the
Higher Education Academy Engineering Subject
Centre has provided funds to build a demonstration
area of the ReLOAD website to raise awareness
amongst UK academics of this new mechanism for
delivering experiments. This will include an
interactive section and teaching materials developed
for use with the experiment.
ACKNOWLEDGEMENTS
The authors would like to thank National
Instruments and the Higher Education Academy
Engineering Subject Centre for their support.
REFERENCES
Selmer, A., Goodson, M., Kraft, M., Sen, S., McNeill, V.,
Johnston, B. S., Colton, C. K., 2005, Process
Dynamics and Control Experiments Performed Across
the Atlantic. Workshop on Internet Accessible
Laboratory Experiments in Chemical Engineering
Education, Cambridge, UK.
Hites, M., 2002, Creating and running undergraduate
experiments controlled through the internet. American
society for engineering education. 2002 IL/IN
sectional conference.
Rohrig, C., Jochheim, A., 1999, The virtual Lab for
controlling real experiments via Internet. Proceedings
of the 1999 IEEE International symposium on
computer aided control system design. Hawai, USA,
pp 279-284.
Sanchez, J., Morilla, F., Dormido, S., Aranda, J.,
Ruiperez, P., 2000, Virtual and remote control Labs
using Java: A qualitative approach. IEEE Control
systems magazine. pp 8 – 20.
Levesley, M.C., Culmer, P., and Cripton, P. 2006. An
Application of Remotely Controlled Experiments to
Perform Feedback-Damping Control of a Vibrating
Beam. Proceedings of the 2
nd
IASTED International
Conference on Education and Technology, Calgary,
Canada. pp233-238.
WEBIST 2007 - International Conference on Web Information Systems and Technologies
426