EL 604 Spring 2002

Solution to Homework 1
Instructions:
• Be sure to write your name on your submission.
• Complete the homework individually. If cheating is observed, you will be awarded 0 for this homework.
• Be neat. Unreadable solutions (by my standards) will not be graded.
• This homework contains questions for which material was not explicitly covered in class. This includes material you
are expected to know from your prior networking classes. If you are unfamiliar with something, please learn the
material needed from appropriate textbooks, and answer the questions. I have posted additional files in the review
folder in my.poly.edu on Dijkstra’s algorithm and IP.
• Total points: 35 points
• DUE DATE: Feb. 13, 2002, at the start of class; hand in homeworks to Xuan Zheng, Class TA; Westchester students,
please email your homework solutions to Xuan Zheng, zhxfifa@photon.poly.edu.

Question 1 (12 points): Assume the following parameters for a switching network :
Table 1: Parameters

Symbol Meaning Value

N Number of hops (links) between two given end systems 4

L Message length in bits 3200

B Data rate, in bits/second, on all links 9600

P Fixed packet payload size in bits (same in both CL and PS CO networks) 400

H1 Overhead (header) bits per packet in CL networks 16

H2 Overhead (header) bits per packet in PS CO networks 4

S End-to-end call setup time (circuit or virtual circuit) in seconds 0.2

D Propagation delay per hop in seconds 0.001

For low-load conditions,

1. Derive general expressions for the end-to-end delay incurred to send the message on the three
network types discussed in class: circuit-switched, PS CO, CL.

2. Compute end-to-end delays to send the message on the three networks using the values shown
for the parameters in Table 1.

Answer:

1. End-to-end delay in circuit-switched networks (1 for each term)

1
 First the circuit should be set up and then when data is sent, it incurs a propagation delay on each
link and a transmission delay for transferring the bits.
End-to-end delay in packet-switched CL network (1 for determining number of packets; 1 for tak-
ing into number of hops; 1 for pipelining)
( P + H1 ) L – 1 (---------------------
P + H1 )
TCL = ND + ---------------------N +  ---
B P  B
The first packet travels N hops and incurs a propagation delay on each hop and a transmission
delay. The remaining packets L ⁄ P – 1 packets only incur a one-hop transmission delay
because of pipelining. The propagation delay of all the remaining packets is also pipelined on the
last segment.
End-to-end delay in a packet-switched CO network (1: S; 1 for using H 2 ; 1 for transmission and
propagation delay)
( P + H2 ) ( P + H 2 )
T CO – PS = S + ND + ---------------------N +  -L-- – 1  ---------------------
B P B 

2. Numerical answers (1 each):

T CL = 0.481

T CO – PS = 0.667

Question 2 (5 points): The Internet community publishes Requests for Comments (RFCs) for various
protocols. Use your web browser to access the Internet Engineering Task Force (IETF) web page (http://
www.ietf.org). Find and retrieve the RFC entitled “Internet Official Protocol Standards.” This is RFC
2400. This RFC gives the state of standardization of the various Internet protocols. What is the latest RFC
number for the following protocols: IP, UDP, TCP, DNS, ARP?

Answer:

Protocol RFC number

IP 791

UDP 768

TCP 793

DNS 1034, 1035

ARP 826

2
Question 3 (3 points): Name two functions common to the Transport Layer and Data Link Layer. What is
the purpose served by each of these functions? List the different methods for performing these two func-
tions. (Keep your answers focussed and short).
Answer: Error control and Flow control (0.5 points). Error control is used to detect and correct
errors (0.5 points). Flow control is used to handle mismatches in the speed of the sender and the
speed of the receiver (0.5 points). Methods of error detection include parity checks, checksum,
CRCs (0.5 points); methods of error correction include ARQ (Automatic Repeat reQuest) and
FEC (Forward Error Correction) (0.5 points). Methods for flow control include rate control, ON/
OFF control and window control (0.5 points).

Question 4 (5 points): For the network shown below, assume that all links are bidirectional. The number
c
2 1
b 1
1 2
f
a e
1 1
2 1

d
indicated next to each link is the per-direction link weight, and the link weights for all links are the same
in both directions. Assume that the path cost is the sum of the weights of the links on the path. Answer the
following questions:

1. Show the steps of Dijkstra’s shortest path algorithm for the network shown above to obtain
the shortest paths from node c to all nodes? Make a table of iterations, showing the set N and
the distances to all nodes at each iteration as decribed in class.

2. At the end, show the routes from node c to each of the other nodes.

Answer: (4 points - one per row, roughly):

Iteration N Da Db Dc Dd De Df

0 {c} ∞ 2 0 ∞ 1 1

1 {c,e, f} ∞ 2 0 2 1 1

2 {c,e, f, d, b} 3 2 0 2 1 1

3 {c,e,f,d,b, a} 3 2 0 2 1 1

Routes from node c to all other nodes (1 point; okay if only one route is listed)
Table 3: Routes to all nodes from c

To Route

a c-b-a

3
 Table 3: Routes to all nodes from c

To Route

b c-b

d c-e-d

e c-e

f c-f

Question 5 (10 points): Assume that the nodes in the network of Question 4 are all IP routers, and the
links are all Ethernet LANs with the following subnet addresses.
Table 4: Subnet addresses

Link Subnet address

a-b 128.238.10

a-d 128.238.11

b-d 128.238.12

b-c 128.238.13

e-f 128.238.14

c-e 128.238.15

c-f 128.238.16

d-e 128.238.17

d-f 128.238.18

1. Using the routes computed in your answer to the Question 4, show the routing table at router
c, assuming it has a separate entry for each subnet in this network. Show a two-column table
with the first column listing all the subnet addresses and the second column listing the next-
hop node name (give the letter representing the node in all cases except when the subnet is on
an interface directly connected to router c, in which case, say “directly connected interface.”

2. If a portable unit with address 128.238.10.56 connects to the Ethernet LAN d-f, then how
should the routing table at node c change (assuming the routing table is updated for all porta-
ble unit moves)?
Answer: (9 points)
Table 5: Routing table at node c

Destination Next-hop

128.238.10 b

128.238.11 e or f

128.238.12 b

128.238.13 Directly connected interface

4
 Table 5: Routing table at node c

Destination Next-hop

128.238.14 e or f

128.238.15 Directly connected interface

128.238.16 Directly connected interface

128.238.17 e

128.238.18 f

Answer (1 point): The routing table at node c will need a new entry with a host specific route for
128.238.10.56 indicating that it is reachable through node f.