Guaranteeing High Availability of the „Secure Exam Environment“
(SEE)
Gabriele Frankl
1
, Peter Schartner
2
and Dietmar Jost
2
1
Department of eLearning, Alpen-Adria-Universität Klagenfurt, Universitätsstrasse 65-67, 9020 Klagenfurt, Austria
2
Department of Applied Informatics, Alpen-Adria-Universität Klagenfurt,
Universitätsstrasse 65-67, 9020 Klagenfurt, Austria
Keywords: Secure Online Testing, Secure Exam Environment, Security, High Availability, Monitoring.
Abstract: Online exams are an increasingly popular form of assessment. Compared to written exams they reduce the
marking workload and offer advantages such as enhanced objectivity, assessment that includes software
specific to the course and thus increased constructive alignment with teaching and learning processes. To
conduct large-scale online exams without the physical restrictions of (often extremely small) computer rooms,
we implemented the “Secure Exam Environment” (SEE) in 2011. The SEE enables online testing in any
lecture hall with electricity and LAN sockets using students’ own devices (and loan devices if needed) while
blocking access to unauthorized files or internet pages. Assessment is conducted via Moodle and additional
software (e.g. Eclipse or GeoGebra) can be used as well. As of November 2017 we have conducted 1,297
such online exams with 47,930 students and are able to test up to 220 students concurrently. To maintain
quality of service we developed a monitoring solution to control the growing complexity of the technical
infrastructure of the SEE. The monitoring solution aims to detect failures sufficiently early to guarantee a
high level of availability and to gather data to further improve the SEE.
1 INTRODUCTION
Student and teacher involvement in assessment,
including digitally-enhanced assessment, remains an
essential part of contemporary learning (Gibson &
Webb, 2015). Assessment channels students’
energies, heavily influences student behavior, shapes
students’ experiences and generates feedback with
opportunities for reflection (Marriot, 2009; Müller &
Bayer, 2007; Sharpe & Oliver, 2007). Despite a
growing number of alternative assessment strategies,
written exams and summative assessment continue to
be the primary methods of assessing factual
knowledge at schools, universities as well as in
several business areas, leading to a huge grading
workload if conducted with paper-and-pencil exams.
eExams result in noticeable time and money savings
(Anakwe, 2008) due to the automatic delivery,
storage and(semi-)automated scoring of (semi-
)standardized question types, along with the
improved readability, structure and clarity of typed
open-text answers. The greater efficiency of eExams
provides students with instant grading and – if
supported by lecturers – feedback (Hewson, 2012).
Furthermore, online exams provide greater flexibility
compared to traditional testing methods (Anakwe,
2008). Moreover, since today's students are more
used to typing than to extensive handwriting
(Hewson, 2012), online exams prevent hand pains
and bad handwriting related to paper-and-pencil
exams. In addition, eExams restrict the halo-effect
which occurs when different handwriting styles
influence the lecturer when grading. Online exams
enable each question to be evaluated on its merits
without being influenced by other answers provided
by the student and thus subjective construction
processes. Hence, online exams enhance objectivity.
Additionally, eExams bring further advantages such
as improved correction possibilities, the
establishment of a question pool or opportunities for
statistical analysis of questions, improving the quality
of questions over time.
Furthermore, the shuffling of questions and
answers, the automatic selection of questions out of a
sufficiently large enough question pool as well as the
opportunity to create questions including variables
which are assigned different values for each student
decreases the likelihood of cheating. Another and
very promising advantage of online exams is the
opportunity to include additional software and
130
Frankl, G., Schartner, P. and Jost, D.
Guaranteeing High Availability of the "Secure Exam Environment" (SEE).
DOI: 10.5220/0006751801300136
In Proceedings of the 10th International Conference on Computer Supported Education (CSEDU 2018), pages 130-136
ISBN: 978-989-758-291-2
Copyright
c
2019 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
multimedia in the examination environment.
According to Biggs and Tang (2011) and their
concept of “constructive alignment”, coherence
between all phases and elements of the learning
process is essential for high quality education.
Intended learning outcomes, teaching/learning
activities, assessment tasks as well as grading should
support one another (Biggs & Tang, 2011; Müller &
Schmidt, 2009). Thus, the software tools used for
teaching and learning – e.g. mathematical or
statistical calculations and analysis, programming,
literature essays - should be used during the
examination process as well. Being able to use
specific software and multimedia in electronic exam
environments paves the way to promising (hands-on)
performance assessments too.
Despite all the positive aspects of eExams
mentioned so far, we found a lack of technical
solutions for conducting secure online exams for
larger audiences. The problems we encountered were
computer rooms that were simply too small as well as
a lack of consideration for the security requirements
which inevitably arise in the context of (electronic)
exams: confidentiality, privacy, integrity,
authenticity, accountability, and availability. The first
five aspects are commonly addressed through
cryptography (e.g. encryption of transmitted and
stored data, network-based security mechanisms like
firewalls, and authentication of messages and users),
whereas the last one is provided by physical and
logical redundancy and continuous monitoring of the
IT system. This includes the continuous monitoring
of the infrastructure (hardware, software and
network) which is a preventive measure to help detect
issues before they cause any major problems.
To overcome the existing shortcomings, we
implemented the Secure Exam Environment (SEE).
After a depiction of the SEE, this paper will focus on
the low-cost monitoring solution (see Ratan, 2017 for
a list of open-source monitoring tools) that guarantees
high availability of our Secure Exam Environment. In
the case of the SEE, the monitoring checks not only
the availability of the service (i.e. Moodle-server) per
se, but also the quality of service (QoS) including
network-related parameters like available bandwidth
and latency (similar to the approach in Zeng et al.,
2009). Further information concerning the other
security requirements mentioned above can be found
in Frankl et al. (2017).
2 THE SECURE EXAM
ENVIRONMENT (SEE)
The Alpen-Adria-Universität Klagenfurt (AAU)
launched the Secure Exam Environment (SEE) for
online testing in 2011 (Frankl et al., 2011) with the
aim of supporting large class sizes, as well as modern
teaching and testing strategies, while working within
budgetary and organizational constraints. By making
use of the students’ existing personal computers
(laptops), the SEE increases efficiency since ordinary
lecture halls can be used for large scale online testing
as well as effectiveness since the students are
presumably familiar with their own devices.
The SEE disables access to students’ own files
and data as well as to other internet sites. Loan
devices are offered for those who do not own a laptop.
As a result, institutional asset requirements as well as
the associated maintenance costs are minimized. We
are currently able to test up to 220 students
simultaneously.
Furthermore, the SEE facilitates the integration of
different software tools and programmes, which are
increasingly used for teaching and learning (e.g.
Eclipse, GeoGebra), into the exam environment,
fostering pedagogical coherence (Biggs & Tang,
2011).
The actual exams are presented as quizzes, a key
component of the Moodle learning management
system (LMS) utilised by the AAU.
In contrast to other electronic exam environments
(e.g. SoftwareSecure, 2017), we avoid the use of
special equipment and encourage students to use their
own device. However, accessing the Moodle server
directly via a web browser running on the student’s
OS is an insecuret approach. In this case, blocking
connections to Wikipedia or other online resources
may be simple, but cheating by using materials stored
on the local hard drive is rather easy. Since we do not
want to force students to install additional software
(such as lockdown modules) on their personal
laptops, we have to use our own operating system
(OS) in order to restrict the access to the local
resources and programmes that are prohibited during
the exam. We decided to boot this OS via the Preboot
eXecution Environment (PXE) protocol over a local
area network (LAN), since the handling of USB sticks
or DVDs is very error-prone, time-consuming and
inflexible, especially when additional software is
needed, and the usage of WLANs is too insecure and
interference-prone. Clearly, this requires that the
client is able to boot via the network.
In order to support a very broad range of (private)
laptops, our solution is designed as a minimal Linux
Guaranteeing High Availability of the "Secure Exam Environment" (SEE)
131
system. At the moment, this OS is realized using
Fedora and Knoppix, which enables us to boot
Legacy or UEFI devices (both Apple and PC). In
order to restrict the access to external resources, we
implemented corresponding firewall rules. Since
Moodle as an LMS not only provides exam features
but also chatting capabilities and course related
material, a solution was needed to prevent access to
such resources and activities during exams. Running
an ordinary web browser in centOS – even when
restricted with firewall rules – would not have
completely solved the cheating problem. Fortunately,
the Safe Exam Browser (SEB – see Safe Exam
Browser, 2017) is fully supported by Moodle-core.
The SEB is more restrictive than an ordinary browser,
since it prevents students from opening other
programmes or additional web browser windows
during the exam and ignores certain key combinations
or clicks. So by limiting access to the exam page only,
cheating by exploiting Moodle’s features is no longer
possible. However, the SEB is only available for
Windows XP, Windows 7 and MAC OS X.
Therefore, we boot a minimized Windows 7 as a
virtual machine on the minimized Linux system via
VirtualBox (see Virtual Box, 2017) (see Figure 1).
Additionally, proprietary software which only runs
on Windows systems is still widespread in the
educational sector. On the one hand, the reliance on a
virtual machine is a drawback in terms of
performance, on the other hand, it adds flexibility
regarding the management of the virtual machine
image. Furthermore, hardware driver management is
done completely in Linux, which is known for its
broad, driverless hardware support especially for
older devices. The selection of the allowed
programmes during the exam (in addition to the SEB)
is set via a configuration file, which is retrieved from
an Intranet Service. In the GUI of this service,
administrators are able to configure the exam (e.g.
only Calculator or Calculator and Excel or GeoGebra
or Eclipse and PDFs allowed).
Starting an online exam using the SEE begins by
booting a minimized Linux from the LAN, then the
minimized Linux automatically starts the Windows 7
virtual machine (VM), Window 7 automatically starts
the SEB, the SEB automatically connects to the
homepage of the AAU’s learning management
system Moodle, and finally users have to log in to
Moodle and select the exam.
Figure 1: The operating principle of the Secure Exam
Environment (SEE).
3 MAXIMIZING THE
AVAILABILITY OF THE SEE
The availability of an exam environment is an issue
of critical importance. Even a short downtime of the
SEE could prevent hundreds of students from taking
exams which might be urgently needed to get marks
or certificates, take new courses, finish modules,
classes or studies, get financial aid for higher
education studies or even get a new job. Furthermore,
students tend to be quite nervous before an exam and
a technical glitch would undoubtedly increase stress
and erode trust in the exam environment. Thus,
perception of the SEE’s reliability (from both for
examiner’s and examinees’ viewpoint) depends on
the availability of the (information) technology
during the exam (Sharpe & Oliver, 2007).
During the SEE boot process, the SEE-servers
(and the personal computers with which the exams
are written) have to operate properly as well as the
network including the switches in the lecture halls. At
the time of writing, the SEE depends on the online
connection between the SEB and the Moodle-server.
Thus, the availability of the SEE can be affected by
hardware failure, network drop outs or service
outages. Analyzing and identifying failures when a
breakdown occurs usually costs a lot of time, which
is at a premium while conducting an eExam. Thus, a
continuous monitoring solution of the various IT
components involved - e.g. servers and computer
networking technologies – to prevent failures and
optimize the availability of the SEE is mandatory,
particularly considering the SEE is based on various
hardware components which are administered by
different departments of the University.
CSEDU 2018 - 10th International Conference on Computer Supported Education
132
Drop outs of components or services or deviations
from thresholds within defined time intervals result in
alerts, allowing support staff to react to and resolve
issues immediately, leading to crucial time-savings
within the failure identification process. Monitored
components and services include the availability of
the SEE-servers (implemented with centOS)
including CPU and storage, as well as DHCP, NFS,
TFTP and HTTP services; the availability of the
administration backend of the SEE including the
corresponding HTTP service; the availability of
Moodle including HTTP-access, as well as end-to-
end-tests in the lecture halls with minimal computers
(Raspberry Pi); the availability of the network
(connection between SEE-server, clients and
Moodle), and end-to-end performance tests within the
network with low-cost probes (Raspberry Pi).
3.1 Monitoring the High-availability
SEE-host and Including Services
The availability of the server, providing the SEE for
network boot, as well as services like DHCP, NFS,
HTTP und TFTP is one of the key requirements of
online testing with the SEE.
We operate the SEE-server as a high availability
and stable system by running multiple redundant
SEE-servers. Using DRBD/Heartbeat or
Pacemaker/Corosync in a failover setup (to define
one server as the master server and the other one as
slave) enables us to switch from one server to the
other automatically in case of a failure or manually in
case of scheduled maintenance. Thus, a new update
can be safely implemented within the system by
installing it on one server and, after careful testing, on
the other and thus the production system.
While monitoring the services mentioned above,
we log CPU utilization, RAM and hard disc usage,
and the status as well as the utilization of the network
interface. Additionally, we periodically check for
pending updates, especially security updates, to
eliminate failures or prevent hack-attacks on the
system and improve performance. Controlling
upcoming updates enables us to schedule
maintenance periods efficiently around exams.
3.2 Measuring Network Performance
Measuring the run-time of the network including the
connection between the SEE-server, clients and
Moodle during an eExam in real time generates
significant data about the latency and utilization of
the network. The open source software SmokePing is
a suitable tool for measuring and visualising the
round-trip-time (RTT) of Linux-based systems by
defining the specific hosts as well as relevant external
hosts which are reachable via ICMP. By default,
every five minutes twenty ICMP-packages are
transmitted to each specified host and used to
calculate RTTs. The median for each interval of
measurement is shown in Fig. 2, with green indicating
no packages were lost while red indicates 19 out of
20 packages were lost. Each single RTT is shown as
smoke in the graph.
Figure 2: RTTs of a host, measured with SmokePing.
Package loss is a signal for capacity overload of
the main host or related hosts, or for a failure or an
erroneous configuration of a network device. Black
‘smoke’ at an interval of measurement shows the
range of fluctuation of the RTT. Increased smoke
indicates a high variation of the RTT per ping and
thus capacity overload of the network. The
combination of SmokePing and probes (Raspberry
Pi’s) placed in the SEE-network enables us to monitor
all servers and network devices and thus to recognise
network bottlenecks and failures at an early stage.
3.3 Maximizing Availability of the
Network Connection
In order to maximise network availability, we only
use wired LAN connections at this point in time.
Despite recent developments, WLAN remains too
error prone and, additionally, a malicious user could
easily perform a denial-of-service (DoS) attack on the
WLAN access points and hence prevent all users
from taking the exam. To achieve such an attack, a
battery-powered pocket-sized WiFi jammer could be
mounted close to or in the room where the eExams
take place.
To ensure the maximum stability of the network
system, the network department of our university
provides high redundancy within the network-core,
distribution-switches, firewalls and the border-router,
Guaranteeing High Availability of the "Secure Exam Environment" (SEE)
133
as well as load sharing via the Border Gateway
Protocol (BGP) in a multihomed environment and
redundant cables. In addition, the equipment used in
the core and distribution layer are high-end
components.
3.4 Infrastructure
The availability of our Secure Exam Environment
(SEE) is affected by the infrastructure in which the
SEE components are embedded. One critical issue is
an Uninterruptible Power Supply (UPS) for the SEE-
server as well as for the network to protect the system
from power failures. The UPS also guards against
over and undervoltage and is backed by means of
batteries (short-term power failures) and a diesel
generator (long-term power failures).
Another important topic is the geographical
distribution of the (redundant) hardware components.
The two SEE-servers are located in different areas of
the university and thus, in the case of an extended
power failure, fire or flooding, it is unlikely that both
servers will be affected.
3.5 Backups
One indirect approach to guarantee the availability of
the SEE-servers, and thus the SEE, is frequent, well
organized backups. In case of an outage like hardware
failure, the SEE-server must be restored to the most
recent valid state. An up-to-date, functioning backup
reduces the mean time to repair (MTTR). A well-
organized backup-strategy includes the evaluation of
functionality of the frequently executed backups as
well as the documentation and frequent testing of the
backups and training of the responsible staff.
Furthermore, it should be guaranteed that spare
hardware (like hot-swappable harddisks and power
supplies and spare network components) is on hand
in case of serious hardware failure.
3.6 Monitoring the Availability of the
Administration Backend of the SEE
Including the Corresponding
HTTP Service
The administration backend is another key
component of the SEE, offered via web interface and
used by the supporting staff to activate any additional
software (e.g. a calculator or Eclipse) for an exam.
The administration backend is accessible only via a
URL https://backend.spu.aau.at. A periodical check
of the HTTP server’s reachability is performed which
monitors the HTTP status code. If the wrong status
code is returned from the backend, an alarm is sent to
the service team. Additionally, it is possible to check
the server’s response times. Longer response times
could be an indicator of network outages or a server
problem.
3.7 Centralized Monitoring of All SEE-
Components and Services
Deviations from threshold values of all components
and services of the SEE are reported at regular
intervals. Every outage triggers an alarm (via e-mail
or SMS) which, together with centralized monitoring,
helps the service team to rapidly identify the cause of
a failure, saving additional time.
3.8 Optimizing the SEE based on
Monitoring Data
The constant monitoring of all components and
services of the SEE offers the opportunity for (trend)
analysis (also see section 5.1 “Further
developments”) as a basis for the continuous
optimization of the systems’ performance.
3.9 Loan Devices
Loan devices serve two purposes within the SEE:
Firstly, it cannot be assumed that all students have a
portable device, and secondly, they may substitute a
student’s personal device in the case of technical
problems or breakdowns during the exam. The AAU
currently has approximately 100 laptops serving as
loan devices for students.
3.10 Reliability
At the time of writing, the SEE depends on the online
connection between the SEB and the Moodle-server.
As a result, users cannot save current results or
proceed to the next question during a network failure.
Thus, the temporary storage of the answers (during
network failures) remains a problem. Fortunately,
Moodle saves the last answer received and the
progress of each examinee. Therefore, the examinee
may simply continue the exam from the point where
the error occurred after potential network problems
are solved. In the worst-case scenario, the last answer
of the examinee is lost. Similarly, laptop failure is not
a severe problem because all previous answers are
stored on the server and the student can simply
continue his or her exam on one of our loan devices.
CSEDU 2018 - 10th International Conference on Computer Supported Education
134
4 TECHNICAL OBSTACLES AND
CHALLENGES
One of the current restrictions of online exams is the
necessity of a network connection. As WLAN is still
prone to failure, LAN is the best option, especially for
larger groups of students. This results in another
challenging aspect, namely that lecture halls require
LAN and power sockets near at least every second
seat. Unfortunately, not all lecture halls fulfill these
requirements and retrofitting is extremely expensive.
The obstacle with the LAN sockets could be
overcome with access points, however running
laptops purely on battery power is risky.
New generations of laptops, requiring continuous
adaptation of the SEE, remain a persistent challenge.
For example, we had to invest significant effort to
support UEFI as a new interface between the
hardware and the OS. Moreover, some manufacturers
no longer offer PXE or Net-Boots on their devices,
forcing us to find workarounds. Furthermore, as many
new laptops come without Ethernet-sockets, we must
support adapters within the SEE.
5 EXPERIENCES WITH EEXAMS
AT THE AAU AND FURTHER
DEVELOPMENTS
In June 2011, we began offering online exams with
the SEE. Table 1 shows the growth of eExams
conducted with the SEE at the AAU over the last six
years.
Table 1: The progression of eExams with the Secure Exam
Environment (SEE), * in progress.
201
1
2012 2013 2014 2015 2016
2017
*
Total
288 2,717 7,475 7,082 8,954 10,391
11,02
3
47,930
5.1 Further Developments
Further developments in monitoring will include the
integration of the students’ devices and the loan
devices into the monitoring concept and predictive
maintenance (for details refer to Sasisekharan et al.,
1994; Susto et al., 2015; Hashemian & Bean, 2011)).
In more detail, we will pursue the following ideas:
· By gathering and analysing the devices’ log-
files whenever they are connected to the SEE,
students’ devices and the loan devices may be
directly integrated into the monitoring system.
This will help to keep the loan devices up-to-
date, because a problem detected on a single
device (currently in use) can (automatically) be
fixed on all other instances of the same model.
A similar process can be applied for the
students’ devices: a problem detected with one
device can either trigger an update of the SEE
(e.g. with respect to drivers) or a warning for
other students using the same model. In the
long-term, the log-data may be included in a
predictive model.
·
The goal of Predictive Maintenance is to
determine the condition of equipment (servers,
laptops, and network-infrastructure like
switches and cabling) in order to predict when
maintenance should be performed in order to
avoid failures. This is contrary to the classical
approach, where maintenance is either triggered
by a concrete failure (aka the break-fix model
[20]) or on an interval-based approach which
often causes unnecessary costs. In short,
predictive maintenance promises time and cost
savings and a higher level of availability.
Currently, we are only able to execute one eExam
with specific settings, e.g. additional software, at the
same time. Therefore, we are developing a boot
environment which enables us to run eExams with
various additional software simultaneously by
recognizing the identity of the student and
transmitting the proper exam environment.
Furthermore, the support of newer devices without
LAN ports is in development. In the future, we also
intend to provide a WLAN access point. Finally, like
every software solution, the SEE needs constant
security and compatibility updates.
6 CONCLUSION
eExams extend the possibilities for assessment in
terms of quality and especially efficiency. However,
the transition from paper-based to electronic exams
raises “new” security-related problems. Traditional
paper-based exams handled requirements like
confidentiality, privacy, integrity, authenticity,
accountability, and availability in a straight-forward
manner: simply preventing access to the empty and
completed exams guarantees their confidentiality, the
paper and well established organizational and
Guaranteeing High Availability of the "Secure Exam Environment" (SEE)
135
personnel processes do the rest (privacy, integrity,
authenticity, accountability, and availability). For
eExams all the aforementioned aspects have to
covered by complex mechanisms, particularly
technical ones. In this paper, we first briefly presented
the secure exam environment (SEE) used at the
Alpen-Adria-Universität Klagenfurt (AAU) and then
presented our low-cost monitoring system that helps
us to achieve a high quality of service level with
respect to the availability of the SEE. We also
discussed technical obstacles and challenges of the
SEE and possible future work concerning the
monitoring system.
REFERENCES
Anakwe, B., 2008. Comparison of Student Performance in
Paper-Based Versus Computer-Based Testing. In:
Journal of Education for Business (October), 13-18.
Biggs, J., Tang, C., 2011. Teaching for Quality Learning at
University. McGraw Hill, Berkshire.
Fluck, A., 2011. eExaminations Strategic Project Final
Report for Academic Senate, University of Tasmania.
Meeting 1/2011, cited in Fluck, A., Hillier, M.:
Innovative assessment with eExams. Paper presented at
the Australian Council for Computers in Education
(ACCE) conference, 29 September – 2 October,
Brisbane, Queensland (2016).
Frankl, G., Schartner, P., Jost, D., 2017. The „Secure Exam
Environment“: E-Testing with Students’ Own Devices.
In: Tomorrow’s learning: Involving everyone learning
with and about technologies and computing, Springer,
179-188.
Frankl, G., Schartner, P., Zebeding, G., 2011. The "Secure
Exam Environment" for Online Testing at the Alpen-
Adria-Universität Klagenfurt/Austria. In: World
Conference on E-Learning in Corporate, Government,
Healthcare, and Higher Education. Association for the
Advancement of Computing in Education (AACE),
Hawaii.
General Electric Company, 2013. Beyond the break-fix
model: predictive services to leverage GE’s record
$229 billion backlog. GE Reports, October 18,http:
//www.gereports.com/beyond-the-break-fix-model.
Gibson, D. C., Webb, Mary, E., 2015. Data science in
educational assessment. In: Education and Information
Technologies (20, Issue 4), 697–71.
Hashemian, H.M., Bean, W.C., 2011. State-of-the-Art
Predictive Maintenance Techniques, In: IEEE
Transactions on Instrumentation and Measurement,
Vol. 60, No. 10.
Hewson, C., 2012. Can Online Course-Based Assessment
Methods Be Fair and Equitable? Relationships between
Students' Preferences and Performance within Online
and Offline Assessments. In: Journal of Computer
Assisted Learning, (28, Issue 5), 488-498.
Marriott, P., 2009. Students' Evaluation of the Use of
Online Summative Assessment on an Undergraduate
Financial Accounting Module. In: British Journal of
Educational Technology, (40, Issue 2), pp. 237-254.
Müller, F. H., Bayer C., 2007. Prüfungen: Vorbereitung -
Durchführung - Bewertung. In: Hawelka, B., Hammerl,
M., Gruber, H. (eds.), Förderung von Kompetenzen in
der Hochschullehre. Asanger, Kröning, 223-237.
Müller, A., Schmidt, B., 2009. Prüfungen als Lernchance:
Sinn, Ziele und Formen von Hochschulprüfungen. In:
Zeitschrift für Hochschulentwicklung (4 No.1), 23-45.
Ratan, V., 2017. An Overview of Open Source Tools for
Network Monitoring,
http://opensourceforu.com/2017/04/overview-open-
source-tools-network-monitoring
Safe Exam Browser (SEB),
http://www.safeexambrowser.org/news_en.html, last
accessed 2017/08/16.
Sasisekharan, R., Seshadri, V., Weiss, S.M., 1994.
Proactive Network Maintenance Using Machine
Learning, in Workshop on Knowledge Discovery in
Databases (KDD94), pp. 453-462.
Sharpe, R., Oliver, M., 2007. Designing courses for e-
learning. In: Beetham, H., Sharpe, R. (eds.): Rethinking
Pedagogy for a Digital Age. Designing and delivering
e-learning. Routledge, London and New York.
SoftwareSecure, http://www.softwaresecure.com/, last
accessed 2017/08/16.
Susto G.A., Schirru, A., Pampuri, S., McLoone, S., Beghi,
A., 2015. Machine Learning for Predictive
Maintenance: A Multiple Classifier Approach, In: IEEE
Transactions on Industrial Informatic, Vol. 11, No. 3.
Virtual Box, http://ww.virtualbox.org, last accessed
2017/08/16.
Zeng, W., Wang, Y., 2009. Design and Implementation of
Server Monitoring System Based on SNMP,
International Joint Conference on Artificial Intelligence
2009
CSEDU 2018 - 10th International Conference on Computer Supported Education
136