Você está na página 1de 7

CS 6543 Computer Networks

Spring 2003 - Midterm


March 6, 2003
Name:

Score: ./20

1. (4 points) In this question, you will explain some concepts briefly.


a. Instead of sending a big packet, we can divide it into small packets and send them in a
pipelined manner. What are the advantages and disadvantages of this so-called packet
segmentation?

b. What is the key difference between cable modem and DSL (digital subscriber line)?

c. What information is used to identify a TCP (transport control protocol) connection


between a server and a client running on different hosts? Note that more than one
client might have a connection to the server process.

d. What is the goal of flow control? How is it achieved in TCP?

Name:.
2. (4 points) Consider the below network. Each link has the same propagation delay of 1msec.
The capacity of each link is shown in the figure (Mbps means mega bits per second). Assume
that packet processing time is zero at every node.
C
Pkt_X
10 Kbytes

40 Mbps
X

80 Mbps
A

Pkt_Y
10 Kbytes

40 Mbps

40 Mbps

40 Mbps

F
40 Mbps

B Y
80 Mbps

80 Mbps

D
Node A has two packets (Pkt_X and Pkt_Y) with the same sizes of 10 Kbytes (kilobytes) to send to
nodes X and Y, respectively. At node B, Pkt_X is forwarded to node C while Pkt_Y is forwarded to
node D. At the other nodes, packets are forwarded along the arrows. Finally, assume that if a
router gets/has two packets going to different next nodes, then the router can send these packets
simultaneously over the different links.
Using the following diagram, show important events (e.g., leaving A, transmission is done at A,
arrived B, queued at A for 3ms etc.) for each packet.

Leaving A
Pkt_X

Time (msec) 1
0

Pkt_Y

10

11

12

13

14

15

Name:..
3. (4 points) In this question you will deal with HTTP, local client caching, and web proxy
caching concepts. Consider the network below.
.
<img src=cnn.com/a.jpg>

CNN.COM
IP5.5.5.1

index.html
Internet

IP=1.2.2.1
Web proxy server for LAN 1.2.2.x

Client A
IP=1.2.2.2

Client B
IP=1.2.2.3

CNN.COM has a web page that consists of one object (a.jpg ) and updates a.jpg and index.html
every day at 5am. But, clients or web proxy server does not know that.
Assume that all clients and web proxy server know the IP address of CNN.COM. Initially there is
nothing in the cache of web proxy server or in the caches of clients. Finally, assume that HTTP
uses non-persistent connections.
Given that the following events are happening during Day1 and Day2, list every step (who is
requesting/sending what from/to who) until a client gets the up-to-date information whenever it
clicks the link. (see the example in the next page and continue from there).
Events
------------Day1, 9am:
Day1, 10am:
Day2, 8am:

client A clicks CNN.COM.


client B clicks CNN.COM.
client B clicks CNN.COM.

For example, (Problem 3 contd)


1.
Client A asks proxy server if it has cnn.com/index.html
2.
Since proxy server does not have this, it establishes a HTTP connection to
cnn.com (now you should continue)
3.

Name:
4. (4 points) You are asked to design a stop-and-wait like reliable data transfer protocol under
the following scenario.
Three hosts X, Y, and Z are connected via broadcast channel, i.e., when a host (say X) sends a
packet the other two will get that packet in a normal condition. But the packet might be
corrupted or lost independently before arriving to Y or Z. Now host X wants to send the same
packets to hosts Y and Z in a reliable manner. In other words, X will send one packet at a time
and make sure that this packet is correctly received by both Y and Z, then it will send the next
packet.
Draw the FSM (finite state machine) at the sender X and at the receivers Y and Z.

Name:---------------------5. Remember that we use Dijkstras algorithm to find the shortest path based on a single
additive cost parameter cij. Now assume that (instead of cij) we associate a probabilistic
reliability measure rij with every link (i,j), where 0 rij 1, and want to find the most
reliable paths from a given source node to all other nodes. The reliability measure of a path
is the multiplication of link reliability measures along that path, i.e., R(p)=(i,j)p rij
The most reliable path problem can be solved by simply modifying the Dijkstras algorithm.
a. (2pt) Using the pseudo code given in the next page, re-write the statements
indicated by dots. Note that the initialization step is already modified for you.
b. (1pt) Using this modified version of Dijkstra, find the most reliable paths
originating from node A in the below network. Do not enumerate all paths, but
follow the algorithm by showing all the changes in R and p.
c. (1pt) Based on the most reliable paths tree in (b), fill out the routing table at node A.

p(A): R(A): 1

0.5

p(B): -

p(C): -

R(B): 0

R(C): 0

B
0.2

0.1

0.4

0.8

0.6
D

0.2

p(D): -

p(E): -

R(D): 0

R(E): 0

Routing table at node A:


Destination

Next Node

Reliability (R)

B
C
D
E

Name:..
(Problem 5.a contd)
Original Dijkstra
Initialization:

Modified Dijkstra
Initialization:

a. N = {A}

a. N = {A}

b. for all nodes v

b. for all nodes v

i. D(v) = infinity
ii. P(v) = -1

c.

i. R(v) = 0
ii. P(v) = -1

D1(A) = 0

c. R(A) = 1

Loop

Loop
...
.

find w not in N such that D(w) is a minimum

add w to N

add w to N

for all v adjacent to w and not in N do

for all v adjacent to w and not in N do

if D(v) > D(w) + c(w,v) then

if .. .. then

D(v) = D(w)+c(w,v)

R(v)=.

p(v)=w

p(v)=w

add w to N

add w to N

end

end

until all nodes in N

until all nodes in N

Você também pode gostar