Duke University

CS / ECE 356 Final

Fall 2021
DECEMBER 8, 2021, 9:00 – 12:00, Reuben-Cooke 126

Name (print): First: Last:

Student ID number:

Number of exam pages: 20

Exam type: closed book/notes
Additional materials allowed: one sheet of your own notes

Marking scheme:
Question Score
Q1: Network control plane / 30 + 2 bonus
Q2: Link layer / 23
Q3: Wireless networking / 21
Q4: Security in computer networking / 26
Q5: Enterprise networking and multimedia / 11
communications
Total: /111+ 2 bonus

1
Q1: 30 pts: Network Control Plane
Routing algorithms: Link State (LS), Distance Vector (DV), both, neither? Circle the correct
answer.
(1 pt) Multiple commonly deployed intra-domain routing protocols use this routing algorithm:
LS/DV/BOTH/NEITHER
(1 pt) In this algorithm, each node talks to all other nodes: LS/DV/BOTH/NEITHER

(1 pt) This algorithm is susceptible to the count-to-infinity problem: LS/DV/BOTH/NEITHER

(1 pt) In this routing algorithm, the shortest paths on the network can be computed using Dijkstra’s
algorithm: LS/DV/BOTH/NEITHER
(1 pt) This routing algorithm can be additionally secured through the inclusion of a reliable
flooding mechanism: LS/DV/BOTH/NEITHER

Name the protocol:

(1 pt) This protocol is sometimes lovingly referred to as “the duct tape of the Internet”:
__________________
(1 pt) The often-used ping utility is implemented via the Request and Reply messages of this
protocol: _________________

(4 pts) Consider the following network. With the indicated link costs, use Dijkstra’s shortest-
path algorithm to compute the shortest path from node X to all other nodes. Show how the
algorithm works by filling out the table below.

Y U

X W

2
Step D(y), p(y) D(v), p(v) D(z), p(z) D(w), p(w) D(u), p(u)
0
1
2

(4 pts) We are calculating the shortest paths in the following network, with the link costs indicated.

Assume that the nodes send their distance vector updates at the same time. That is, all nodes send
their updates first at t=0, then at t=1, then at t=2, and so on.
The following is the information the nodes (vertical column) advertise about other nodes in the
network (top row) at t=0.
t=0
A B C D E
A -- B, 7 -- -- E, 1
B A, 7 -- C, 1 -- E, 8
C -- B, 1 -- D, 2 --
D -- -- C, 2 -- E, 2
E A, 1 B, 8 -- D, 2 --

What is the information the nodes advertise at t=1, t=2, and t=3? Fill out the tables below.
t=1
A B C D E
A
B
C
D
E

3
t=2
A B C D E
A
B
C
D
E

t=3
A B C D E
A
B
C
D
E

What if the link between the nodes A and E fails? Recompute the distance vectors in the table at
t=3. Fill out the table below.

A B C D E
A
B
C
D
E

4
(3 pts) The BGP is referred to as a “path vector” protocol. Explain why (2 pts). Use the following
network, in which AS1, AS2, and AS3 are exchanging information about themselves and network
X, to provide a specific example to substantiate your explanation (1 pt).

(4 pts) We have been referring to BGP as a policy-driven protocol. Provide a specific example of
a case where BGP’s behavior is driven by policy, rather than performance. Draw a specific network
topology that corresponds to your example (2pts). Explain how specifically the network behavior
you are describing is advantageous from the perspective of policy (1pt) and how specifically it
may hurt performance (1pt).

5
(4 pts) Ease of network management has been noted as an advantage of Software-Defined
Networking (SDN) in our routing lectures, and in the invited lectures of the two speakers who
work on enterprise and planetary-scale networking. Explain, in detail, two specific ways in which
SDNs are easier to manage than traditional networks (2 pts each).

1:

2:

(4 pts) Name and briefly describe two types of instructions (2pts each) that an OpenFlow SDN
controller could be sending to an SDN switch.

1:

2:

6
(2 bonus points) Despite the benefits of SDNs, enterprise adoption has been relatively slow.
Explain how traditional telecom providers such as Cisco and Juniper have played a role in this.

Q2: 23 pts: Link Layer and Medium Access Control

TRUE or FALSE? Circle the correct answer.
(1 pt) The two “taking turns” MAC protocols we discussed, polling-based and token-passing-
based, have a single point of failure: TRUE / FALSE
(1 pt) Ethernet is connection-oriented: TRUE / FALSE
(1 pt) Ethernet uses acknowledgments and retransmissions to achieve reliable data delivery:
TRUE / FALSE
(1 pt) In the “bus” Ethernet network topology shown below, all nodes are in the same collision
domain: TRUE / FALSE

(1 pt) In the “star” Ethernet network topology shown below, all nodes are in the same collision
domain: TRUE / FALSE

7
Fill in the blanks:
(1pt) In the example below, the channel partitioning approach is based on
_____________________________________.

(1pt) ALOHA, Ethernet, and IEEE 802.11 are examples of _______________________ MAC
protocols.
(2pts) The diagram below shows 2 types of devices that store and forward packets on a network,
1 and 2. What types of devices are these?

1: ___________________________
2: ___________________________

(4 pts) In Ethernet, the nodes listen to the channel before transmitting. However, it is still possible
for nodes’ transmissions to collide.
Why are Ethernet collisions possible, despite carrier sensing? Provide a specific example
of a case where Ethernet node transmissions collide (2 pts).

8
 What is the benefit of collision detection in Ethernet? Provide a specific example of
performance improvement achieved through collision detection (2 pts).

(6 pts) The next 4 questions refer to the following network diagram, in which two networks are
connected through a router R.

(1 pt) Assume that node A’s ARP table is empty. A needs to send a datagram to a
node with IP address 111.111.111.112. What are the source (0.5pt) and destination
(0.5pt) MAC addresses of the ARP request A sends out?

(1 pt) What are the source (0.5pt) and destination (0.5pt) MAC addresses in the
ARP Reply packet node A receives?

9
 (2 pts) Assume that A knows that to get to C and D, it needs to go through router
R, and that A has R’s information in its ARP cache. A crafts a datagram destined
for node C. What are the source (0.5pt) and destination (0.5pt) IP addresses in this
datagram? What are the source (0.5pt) and destination (0.5pt) MAC addresses?

(2 pts) R is now forwarding A’s datagram to C. Assume that R has all nodes’
information in its ARP table. What are the source (0.5pt) and destination (0.5pt) IP
addresses in the datagram R sends out? What are the source (0.5pt) and destination
(0.5pt) MAC addresses?

(4pts) The next 4 questions refer to the following network diagram, where a 4-interface switch is
connecting 4 nodes. Assume that this switch’s forwarding table is empty. The switch has
received a frame from node A, destined to node B.

(1 pt) Where does the switch send this frame?

10
(1pt) What changes does the switch make to the MAC addresses in this frame?

(1pt) Having transmitted the frame, what entry does it include in its switching table,
shown below?

(1pt) Node B is now generating a reply, that needs to get to node A. What does
the switch do with this reply?

11
Q3: 21 pts: Wireless Networking
True or False? Circle the correct answer:
(1 pt) Commonly deployed 802.11 protocols, e.g., 802.11a/b/g, operate in unlicensed spectrum
bands: TRUE / FALSE
(1 pt) 3G operates in unlicensed spectrum bands: TRUE / FALSE
(1 pt) 4G/LTE operates in unlicensed spectrum bands: TRUE / FALSE

Fill in the blanks:

(1 pt) IEEE 802.11 MAC incorporates a combination of carrier sensing and collision
_________________________________________________________________________
(1 pt) In IEEE 802.11, when choosing which AP to associate with, the host usually chooses the
AP which has _____________________________________________________________.

Name the “G”: The following 2 questions refer to the cellular “generations”: 2G, 3G, 4G, and
5G.
(1 pt) ___G is the first “G” designed for a diverse set of connected devices.

(1 pt) The cellular network architecture shown above is the ____G architecture.

12
(2 pt) In wireless communications, we can improve the SNR by increasing transmission power,
up to the regulatory limits. Name 2 disadvantages of this approach to improving node’s bit error
rate (1 pt each).

1:

2:

(4 pt) Describe the channel reservation mechanism in 802.11. Specify which frames are
exchanged. Explain the role of the access point in this mechanism. Draw a diagram if it helps with
your explanation.

13
(8pts) Compare and contrast Ethernet and IEEE 802.11. Which one is more complex? (1pt) What
is the underlying reason for this? (1pt) State and briefly explain 2 similarities (2 pts) and 4
differences (4pts) in their capabilities and the services they provide. Be specific. Please number
the similarities and the differences you are listing, to help us parse your explanations.

14
Q4: 26 pts: Security
TRUE or FALSE? Circle the correct answer.
(1 pt) DES is more secure than AES: TRUE / FALSE
(1 pt) An SSL handshake starts before the TCP handshake is completed: TRUE / FALSE

Fill in the blanks:

(1 pt) The key advantage of symmetric key cryptography over public/private key cryptography
is: _____________________________________
(1 pt) The key disadvantage of symmetric key cryptography over public/private key
cryptography is: _______________________________________________________________
(1 pt) In public key cryptography, entity’s public key is known to ______________________.
(2 pts) Following the notation used in the lectures, in public/private key cryptography, p,q are
large prime numbers, n = p*q, z = (p-1)(q-1), e<n and has no common factors with z, and ed mod
z = 1. Then, to encrypt the message m with the public key, we compute

c= _________________

To decrypt the resulting ciphertext c and recover the original message m, we compute:

m = _______________

(2 pts) In the process of generation of a public key certificate for entity E, the
__________________________________ signs the certificate with its _________________ key.
(2 pts) To create a digital signature of his message, Bob encrypts the _____________________
of the message with his ______________________ key.
(1 pt) Traditional firewalls examine the Data Link, Network, and Transport layers of the OSI
model. Next-generation firewalls examine these layers, as well as the _________________ layer.

15
(6 pts) SSL uses a Master Secret to derive several keys. Why? How many keys are derived?
Explain how each key is used.

(4pts) Explain how authentication protocols use nonces. Draw a diagram if it helps with your
explanation (2pts). Explain how this prevents replay attacks. (2pts)

16
(4 pts) A stateless packet filter-based firewall implements the following set of access control
policies:

What is the IP address of the network that is being protected? (1pt)

Give a specific example of an everyday activity that the users on this network are able to
do. (1pt)

Give a specific example of an activity that the users of this network are not able to do. (1pt)

Is the provided set of policies likely to be used in typical enterprise networks? Explain
why/why not. (1pt)

17
Q5: 11 pts: Multimedia Communications
(4 pts) Why do we call multimedia transmission streaming? (1 pt) Name 3 differences between
multimedia streaming and file transfer (e.g., via FTP, or as part of web browsing) (1 pt each). To
help us parse your explanations, please number the differences you are listing.

Fill in the blanks:

(1 pt) When streaming live events (e.g., sports matches, new events), we can tolerate delays up
to approximately ____________.
(1 pt) In VoIP communications, packets that arrive after 400 ms are usually _______________.

18
(2 pts) We discussed the following Forward Error Correction (FEC) scheme, where additional
redundancy is added to VoIP transmissions to allow for partial recovery of lost information.

Name and briefly explain the specific “redundancy” that is used in this scheme.

(3 pt) The following table refers to different approaches to providing multimedia support at the
network (rather than the application) layer. In this table, the column “Deployed?” refers to the
extent of the deployment of the three different approaches.

19
What are 1, 2, and 3? Provide a description of the extent of the deployment, rather than a
“yes” or a “no”:
1: ____________________________________________________________
2: ____________________________________________________________
3: ____________________________________________________________

20