Teaching Computer Networking with the Help of

Personal Computer Networks

Rocky K. C. Chang
Department of Computing
The Hong Kong Polytechnic University
Hung Hom, Kowloon, Hong Kong
csrchang@comp.polyu.edu.hk

ABSTRACT coffee shops, just to name a few. As a result, there is a press-

This paper describes an “experiential” learning approach ing need to equip students on both undergraduate and grad-
to teaching a foundational course on Computer Network- uate levels with a solid foundation in the field in order to
ing. In additional to the traditional laboratory sessions further foster and flourish its development and applications.
and analytical problem-solving, I have introduced a class However, in many ways Computer Networking education
project based on personal computer networks for the last is still in an exploratory stage. For example, we have seen for
three years. Each project group sets up and owns an IP the last ten years many different approaches adopted by var-
private network throughout the course, and they perform ious textbooks—analytical [1], bottom-up [2], engineering
various experiments on it to verify and test the networking [3], system [4], balanced-view [5], visual [6], and top-down
principles learned from the textbooks and lectures. The stu- [7]. Whether these approaches can effectively facilitate stu-
dents’ feedbacks collected so far are extremely positive. Not dents’ learning are yet to be seen and evaluated.
only they can acquire practical skills during the process, they As the content of the field continues to increase in a very
are able to better understand the abstract networking con- rapid rate, it is also very important to identify a minimal
cepts and protocols through their working experience with set of core principles to teach undergraduate students, and
the personal computer networks. this has been discussed in the ACM Workshop on Network-
ing Education last year [8]. Moreover, there is increasing
effort on providing hands-on experience to students through
Categories and Subject Descriptors more traditional laboratory sessions/courses, implementa-
C.2.2 [Network Protocols]: Applications - Routing Pro- tion of networking hardware, simulation tools, and socket
tocols; C.2.6 [Internetworking]: Routers - Standards programming.
This paper considers a foundational course on Computer
General Terms Networking for junior students in a typical Computer Sci-
ence or Computer Engineering program. The main thesis
Algorithms, Measurement, Performance, Experimentation of this paper is that students’ learning in this subject can
be significantly enhanced if they can interact with an actual
Keywords computer network in parallel to other teaching and learn-
ing activities. In the next subsection, I will first present
TCP/IP, Experimental problem-solving, Personal computer
the difficulties connected to teaching and learning this sub-
networks
ject. In section 2, I describe my experience of using a class
project based on personal computer networks to enhance the
1. INTRODUCTION teaching and learning qualities. In section 3, I present the
Computer networks and the Internet have already become student survey results to assess the effectiveness of the ap-
the most critical infrastructure today for information dis- proach. Finally, I conclude with some future work in section
semination, business transactions, human communications, 4.
computer games, scientific computation, and even national
security. Moreover, computer networks are ubiquitously de- 1.1 Identifying learning and teaching
ployed in many other commercial sectors, such as automo- difficulties
biles, smart highways, smart clothing, smart appliances, and
“Why is it (not) so difficult to learn Computer Network-
ing?” is perhaps the first question that we as educators
should ask ourselves (and our students). One of the obvious
Permission to make digital or hard copies of all or part of this work for answers is of course to do with the rapid advances in the
personal or classroom use is granted without fee provided that copies are field. Only ten years ago, one undergraduate subject and
not made or distributed for profit or commercial advantage and that copies one postgraduate subject were perhaps sufficient to provide
bear this notice and the full citation on the first page. To copy otherwise, to a reasonable coverage. Today, however, an entire undergrad-
republish, to post on servers or to redistribute to lists, requires prior specific uate program can be devoted to Computer Networking, and
permission and/or a fee.
ITiCSE’04, June 28–30, 2004, Leeds, United Kingdom. its associated subjects, such as economic and social issues.
Copyright 2004 ACM 1-58113-836-9/04/0006 ...$5.00. Many Computer Networking topics have also quickly devel-

208
oped into separate subjects, notably Wireless and Mobile 5. Some of the networking problems are difficult to com-
Networks, Optical Networks, and Network Security. prehend and appreciate due to students’ common lack
From the educators’ point of view, it also becomes in- of practical experience. It is generally not difficult for
creasingly difficult to teach the subject effectively, partly undergraduate students to understand the access net-
because the Internet’s success has attracted students with work technology, because almost all of them have ex-
various backgrounds. Thus, it is hard to provide one class perience of accessing the Internet via ISPs and LANs.
for all. Even among students with similar educational back- However, it is relatively difficult to comprehend the
grounds, say Computer Engineering students, some of them scalability issues in inter-domain routing and the im-
may have taken professional examinations, such as Cisco’s portance of traffic engineering.
CCNA, before attending a first course on Computer Net-
working, while others are not even able to expand the term 6. The sequence of coverage can also affect students’ learn-
TCP/IP. ing significantly. Computer Science students are per-
Besides the issues related to the forever changing nature haps more comfortable with a top-down approach while
of the field (at least for now) and heterogeneity in students’ engineering students may find the pure bottom-up ap-
backgrounds, there are quite a number of obstacles to teach- proach more logical. In either case, it is bound to
ing and learning Computer Networking effectively, as out- have missing information, which is yet to be revealed
lined below. later, in order to complete the entire storytelling of
1. The principles underlying Computer Networking are the Computer Networking internals. Unfortunately,
intrinsically very complex. The layered model helps students may stumble on these missing information.
understand and manage the complexity. But very soon Therefore, a good sequence of coverage may follow a
students find out that this layering approach has its “nonlinear” path which is in contrast to a rigid top-
own inadequacy. For example, the layers are not re- down or bottom-up sequence.
ally independent of each other, and a network layer
could be degenerated into a data-link layer, e.g., IP
over ATM. Therefore, getting the whole picture cor- 2. PERSONAL COMPUTER NETWORKS
rectly is already a challenge to many students. One issue that clearly stands out from the discussion in
the last section is that a lack of practical and meaningful
2. Computer Networking concepts and protocols are also experience with an actual computer network could greatly
very abstract to many students. For one, they cannot hinder the understanding of the Computer Networking con-
see typical networking equipments, and visualize pack- cepts. Therefore, practical assignments are usually supple-
ets and protocols for themselves during lectures. For mented in a typical undergraduate course, e.g., socket pro-
example, talking about a LAN switch without seeing gramming assignments and laboratories.
one is already a hinderance for many students. Exam- Although socket programming assignments and labora-
ining the kernel code without a prior understanding tory sessions are useful, they do not allow students to in-
of how the protocols work helps very little. Computer teract with the “kernel” of a computer network. To fill this
animation alleviates the problem to some extent, but void, I initiated a class project three years ago to let students
it may not be able to equip students with the ability of experiment with a personal computer network throughout
conceptualizing other more abstract concepts on their the course—each group is expected to work on their own
own. network in parallel to the course’ progression. However, the
3. Unlike Computer Programming and Computer Archi- amount of resources required for supporting 170 students
tecture courses, for example, resource provision for is obviously very demanding, in terms of both equipments
hands-on practical experience in Computer Network- (PCs, hubs, NICs, etc) and space (space is extremely costly
ing is problematic. A Computer Networking labora- in Hong Kong). Finally, I have come to the conclusion that
tory, if available, usually has a ready-to-use computer the only solution to resolving the resource constraint is to
network on which students can conduct various exper- “bring the networking laboratory to where students are.”
iments. However, students should also be expected to That is, each group is responsible for finding their own ma-
know how to set up a computer network from scratch chines, OSes, networking equipment, and space. Although
after taking the course, very much like knowing how to it sounds like a very difficult task, all groups in the past
write programs after taking a Computer Programming three years were able to find all the resources and finish the
class. But it is not quite possible to provide adequate project.
resources to achieving that in many academic environ- The projects are performed on group basis. The entire
ments. project consists of several phases. The first one is to find
all the resources to form IP private networks which are usu-
4. Many terminologies and acronyms are introduced and ally situated in students’ homes. The network is set up in
used in the field, and some of them are very similar, a month’s time. Therefore, the students had to read ahead
e.g., ARP and ARQ. Worse yet, these terminologies and understand some of the basic networking concepts, such
are often not used consistently, especially in the indus- as IP subnets and NAT. The first phase is concluded with a
try, e.g., hubs, switches, switching hubs, port switching presentation on the network setup. The second phase is to
hubs, and segment switching hubs. Students, on the observe how network protocols work by conducting experi-
other hand, are more used to clear and formal defi- ments on the personal computer network. At the end of the
nitions. For example, many students in my class are project, each group is asked to submit a written report and
very perplexed by the term round-trip time when it is make a final presentation. A sample report is available in
first mentioned during the subject overview. [9].

209
 KEITH-NOTEBOOK
(Windows XP)
eth0: 172.16.0.15
turtle
192.168.0.2 192.168.0.3 (Windows ME)
eth0: 172.16.0.17 ED_NOTEBOOK
(Windows XP)
eth0: 172.16.0.16
100Mbps Fast
Subnet 4
Ethernet switch
11Mbps Wireless (Ad-hoc) Ethernet
Network/Subnet ID: 192.168.1.0/24
SSID: PAUL
Hub WEP: 01:3b:4d:2b:3e:6a:7c:45:12:1d:2d:14:5e (104bits) Subnet 1
100BaseT Ethernet
Star Topology using Hubs
KEITH-NOTEBOOK Network/Subnet ID: 172.16.0.0/17
(Windows XP)
wlan0: 192.168.1.2
192.168.0.1

192.168.1.1
Router eth0: 172.16.0.1

192.168.1.2 wlan0: 192.168.1.1 Paul

(Redhat Linux 8)

ED_NOTEBOOK eth1: 172.16.128.2

(Windows XP)
wlan0: 192.168.1.3

100Mbps Fast 0516L

Ethernet switch (Redhat Linux 9)
Figure 1: A minimal personal computer network. eth0: 172.16.128.15

Subnet 2
100BaseT Ethernet
Star Topology using Ethernet Switch
Network/Subnet ID: 172.16.128.0/17

2.1 Configuring private IP networks

A minimum set up, as depicted in Fig. 1, includes two Network/Subnet ID: 158.132.45.0/24 eth0: 172.16.128.1
Gateway: 158.132.45.2
subnets—an Ethernet LAN and a PPP network. A Linux Internet eth2: 172.16.128.1
1742L
(Redhat Linux 8)
system is configured as a router that also performs Network
eth1: 192.168.0.1
Address Translation (NAT). Although the test network is Subnet 3
100BaseT Ethernet
small, it contains both point-to-point and broadcast net- Network/Subnet ID: 192.128.0.0/24
ken-hall
works. Moreover, the PPP’s MTU can be manually config- (Windows XP)
eth0: 192.168.0.15

ured to cater for experiments on IP fragmentation and TCP

throughput.
Advanced students are able to configure more complex Figure 2: A more elaborate personal computer net-
networks and services, such as that in Fig. 2. Starting from work.
last year I have asked students to configure an IPSec tunnel
between two such private networks. Therefore, two or more
project groups need to work among them to set up such 2. How the NAT router translates the internal IP ad-
tunnels. dresses and ports,
Through the process of setting up the network, students
pick up various invaluable practical skills of configuring net- 3. How a router fragments an IP packet,
works, diagnosing network problems, and finding solutions 4. The TCP three-way handshaking messages and the
to them. These skills cannot be taught adequately in tradi- protocol values, and
tional laboratory sessions. In particular, the students will be
familiarized with the followings after this stage of project: 5. The TCP’s persistent behavior in setting up a connec-
tion.
1. IP addressing scheme
Another set of experiments is related to the measurement
2. The concept of IP subnets and subnet masks
of the network performance. For example, based on the
3. Default router configurations measurement results, students can investigate

4. PPP network setup 1. The effect of frame collisions on an Ethernet network’s

performance,
5. NAT setup using iptables
2. The relationship between MTU and network through-
6. Network diagnosis tools, such as ipconfig, ping, and put,
netstat
3. The effect of IP fragmentation on the network through-
7. Setting up various network services, such as SSH, HTTP, put,
TELNET, SAMBA
4. The effect of packet fragmentation on the network per-
2.2 Network experimentation formance,
The second part is to reinforce the principles learned in
5. The relationship between buffer sizes and TCP through-
lectures by performing various experiments on the network.
put, and
In other words, the hands-on experiments draw students
closer to the Computer Networking principles, because they 6. The DNS delay in repeated queries.
can now visualize how protocols work for themselves. There
are several types of experiments. The first one is simply to The last set of experiments concerns network faults and
observe how network protocols work, e.g., they can observe misconfigurations. For example, students will observe that
1. How ARP resolves the target MAC addresses using 1. The states of a TCP connection are not affected by a
data-link broadcast, nonpersistent network outage,

210
 Assignments Project for practical skills
Assignments Project for practical skills Projects for understanding Mid-term tests
Projects for understanding Mid-term tests
60
70 50
60 40
50
30
40
20
30
20 10
10 0
Strongly Agree Neutral Disagree Strongly
0 Agree Disagree
Strongly Agree Neutral Disagree Strongly
Agree Disagree Figure 4: Effectiveness of assessment components
Figure 3: Effectiveness of assessment components for a small class in 2002.
for a large class in 2001. Assignments Project for practical skills
Projects for understanding Mid-term tests

2. The protocol interactions as a result of configuring du- 60

plicate addresses on the same subnet, and
50
3. The effect of various hacking activities on the net- 40
works. 30
20
3. EVALUATION 10
This section presents students’ feedbacks received from
0
an undergraduate course on Computer Networking in the Strongly Agree Neutral Disagree Strongly
last three years. The class in 2001 was a large one, consist- Agree Disagree
ing of 168 students. The classes in 2002 and 2003, on the
other hand, were small, consisting of 54 and 46 students, Figure 5: Effectiveness of assessment components
respectively. The metrics of the evaluation are based on the for a small class in 2003.
impact of the group projects on the students’ learning, and
the change of students’ interests in the subject at the end of
project surpassed that for the assignments—more than 90%
the course.
of the respondents thought positively of the project on the
3.1 The learning outcomes practical skill aspect. Even with a large class, the project
was very useful to most of the students. Around 65% of
Besides the feedbacks on the group project, I also include
the responses (strongly) agreed that the class project helped
their feedbacks on the assignments and tests for the pur-
them acquire practical skills in Computer Networking, while
pose of comparison. Therefore, the following four statements
51% of responses (strongly) agreed that the class project
were presented to the students, and there were five possible
helped them understand the subject materials.
answers to each question: Strongly agree, agree, neutral,
The responses were quite understandable. Although un-
disagree, and strongly disagree. The feedbacks are charted
derstanding the networking principles was actually a major
in Figs. 3-5. The numbers of responses received were 133
objective of the project, students still perceived that ac-
(79%), 52 (96%), and 41 (89%) for 2001-2003, respectively.
quiring the practical skills throughout the process was more
• The assignments helped me understand the subject important. Based on students’ written comments, many of
materials. them felt that acquiring the skills of setting up networks
from ground up would help them land on jobs upon grad-
• The class project helped me acquire practical skills in uation. Another reason is that most of them had never set
Computer Networking. up a computer network before; therefore the practical work
itself seemed to be more rewarding and significant.
• The class project helped me understand the subject
materials. 3.2 Students’ interests in the subject
• The tests helped me find out how much I have under- The most important student feedback, in my opinion,
stood (or have not understood). is whether students are more interested in Computer Net-
working after taking the course. Therefore, students were
The responses to both statements regarding the class project asked whether they are interested in Computer Networking
were quite similar for the small class in 2002 in that around at two separate times: during the first class (indicated by
80% of the responses from the small class (strongly) agreed pre-teaching in Figs. 6-7) and during the last class (indi-
that the class project was helpful in the two aspects with cated by post-teaching in Figs. 6-7). The reason for doing
slightly more favorable responses on the practical skill. The this is to accurately assess the impact of this course on their
responses received from the small class in 2003 were even interests in the subject. Previously in the large class, stu-
more encouraging. For the first time, the feedbacks on the dents were asked whether they were more interested in the

211
 Pre-teaching Post-teaching
hand, the percentage of agreed cases decreased significantly
80 from 72% at the beginning of the course to 43% at the end
70 of the course. The other cases also increased at the end of
60 the course. Similar results were observed for another small
50 class in 2003, as shown in Fig. 7.
40
30 4. CONCLUSIONS AND FUTURE WORK
20 In this paper I have presented a group project based on
10 personal computer networks. The main advantage of this
0 learning approach is to let students practise and verify what
Strongly Agree Neutral Disagree Strongly they have learnt from lectures and textbooks as the course
Agree Disagree
advances. By doing so, the abstract networking concepts can
be made more concrete to the students. At the same time,
Figure 6: Students’ interests in Computer Network- they are taught practical skills in configuring and diagnosing
ing before and after taking the course in 2002. an actual computer network.
One possible improvement is to further tighten the in-
Pre-teaching Post-teaching tegration of the class project with the instructional part.
60 Currently, I start giving out the project approximately a
month after the beginning of the course. In the next year, I
50 am planning to give out the class project in the first class,
40 and students are expected to “experience” a certain net-
working topic on their personal computer network as soon
30 as the topic is covered in lectures. This approach is some-
20 what similar to the experiential learning approach discussed
in the education sector. In contrast to cognitive learning, ex-
10
periential learning attempts to provide incentives to learn,
0 and in my case the incentive is to build a personal computer
Strongly Agree Neutral Disagree Strongly network after completing the course.
Agree Disagree

Figure 7: Students’ interests in Computer Network- Acknowledgment

ing before and after taking the course in 2003.
This work was partially supported by a Teaching Develop-
ment Grant from The Hong Kong Polytechnic University.
subject after taking the course, but the responses did not The network setup in Fig. 2 is part of the project report by
clearly indicate the absolute levels of interests. Edmond Chan, Joey Chan, Keith Lee, and Paul Chu [9].
Although the group project is just one of the many activ-
ities that would affect the students’ interests in the subject, 5. REFERENCES
I believe that the project’s contribution is much more sig- [1] D. Bertsekas and R. Gallager, Data Networks, Second
nificant than others. It is quite easy to observe a positive Edition, Prentice-Hall, 1992.
correlation between a student’s interest in the subject and [2] A. Tanenbaum, Computer Networks, Fourth Edition,
his interest in the project. Some students were very excited Prentice Hall PTR, 2003.
about setting up a private network and observing how proto- [3] S. Keshav, An Engineering Approach to Computer
cols work, and in turn they were willing to spend more time Networking, Addison Wesley Longman, 1997.
on the project. On the other hand, other students would
[4] L. Peterson and B. Davie, Computer Networks: A
perceive the project just another burden to bear, and this
Systems Approach, Second Edition, Morgan Kaufmann,
attitude in turn decreased their interests in the subject.
2000.
The feedbacks are very revealing. For the 2002 class, for
[5] A. Leon-Garcia and I. Widjaja, Communication
example, around 90% of the respondents (strongly) agreed
Networks: Fundamental Concepts and Key
that they are interested in Computer Networking in the first
Architectures, Second Edition, McGraw-Hill, 2004.
class. This result is also predictable, because of the impor-
tance of the Internet. Moreover, some students wrongly be- [6] B. Forouzan, Data Communications and Networking,
lieved that Computer Networking is equivalent to WWW, or Second Edition, McGraw-Hill, 2000.
other network applications. Therefore, after learning the ac- [7] J. Kurose and K. Ross, Computer Networking: A
tual Computer Networking principles, many students’ per- Top-Down Approach Featuring the Internet, Second
ceptions about the subject changed. Some became aware Edition, Addison Wesley Longman, 2003.
that Computer Networking was not for them, because the [8] ACM Workshop on Networking Education, Aug. 2002.
content may be a little too difficult to comprehend. How- Available from http://www-
ever, others found the subject much more interesting than net.cs.umass.edu/sigcomm/education/workshop1.html.
before, because both the extent and depth of the subject far [9] E. Chan, J. Chan, K. Lee, and P. Chu, A report on
exceeded what they knew before. As a result, we see a higher “Networking in an Unsafe World!”, available from
variance in the post-teaching responses. The percentage of http://www4.comp.polyu.edu.hk/∼csrchang/Edmond.pdf,
strongly-agreed cases jumped from 19% at the beginning of May 2003.
the course to 32% at the end of the course. On the other

212