First Midterm For ECE374: 03/24/11 Solution!!

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

ECE374:

 First  Midterm                                                                                                                                                                                                                                        1  
 
First  Midterm  for  ECE374  
 03/24/11  
Solution!!  
 
Note:  In all written assignments, please show as much of your work as you can. Even if
you get a wrong answer, you can get partial credit if you show your work. If you make a
mistake, it will also help the grader show you where you made a mistake.  
 
Problem  1:  
In  modern  packet-­‐switched  networks  the  source  hosts  segments  long,  application-­‐
layer  messages  (for  example,  an  image  or  a  music  file)  into  smaller  packets  and  
sends  the  packets  into  the  network.  The  receiver  then  reassembles  the  packet  back  
into  the  original  message.  We  refer  to  this  process  as  message  segmentation.  Figure  
1  below  illustrates  the  end-­‐to-­‐end  transport  of  a  message  with  and  without  message  
segmentation.  Consider  a  message  that  is  8*106  bits  long  that  is  to  be  sent  from  
source  to  destination.    Suppose  each  link  has  a  maximum  capacity  of  2Mbps.  Ignore  
propagation,  queuing,  and  processing  delays.  
a. Consider  sending  the  message  from  source  to  destination  without  message  
segmentation.  How  long  does  it  take  to  move  the  message  from  the  source  
host  to  the  first  packet  switch?  Keeping  in  mind  that  each  switch  uses  store-­‐
and-­‐forward  packet  switching,  what  is  the  total  time  to  move  the  message  
from  source  host  to  destination  host?  
b. Now  suppose  that  the  message  is  segmented  into  4,000  packets,  with  each  
packet  being  2,000  bits  long.  How  long  does  it  take  to  move  the  first  packet  
from  source  host  to  the  first  switch?  When  the  first  packet  is  being  sent  from  
the  first  switch  to  the  second  switch,  the  second  packet  is  being  sent  from  the  
source  to  the  first  switch.  At  what  time  will  the  second  packet  be  fully  
received  at  the  first  switch?  
c. How  long  does  it  take  to  move  the  file  from  source  host  to  destination  host  
when  message  segmentation  is  used?  Compare  this  result  with  your  answer  
in  part  (a)  and  comment.  
d. Discuss  the  drawbacks  of  message  segmentation.  
 
 
 
 
Figure  1  Network  for  problem  1  

Solution:  
a) Time  to  send  message  from  source  host  to  first  packet  switch  =  

.  With  store-­‐and-­‐forward  switching,  the  total  time  to  move  


message  from  source  host  to  destination  host  =    
b) Time  to  send  1st  packet  from  source  host  to  first  packet  switch  =  .  

.  Time  at  which  2nd  packet  is  received  at  the  first  switch  =  
time  at  which  1st    packet  is  received  at  the  second  switch  =  
 
c) Time  at  which  1st  packet  is  received  at  the  destination  host  =  .  
.  After  this,  every  1msec  one  packet  will  be  received;  
thus  time  at  which  last  (4000th)  packet  is  received  =  
.  It  can  be  seen  that  delay  in  using  message  
segmentation  is  significantly  less  (almost  1/3rd).    
d) Drawbacks:  
i. Packets  have  to  be  put  in  sequence  at  the  destination.  
ii. Message  segmentation  results  in  many  smaller  packets.  Since  
header  size  is  usually  the  same  for  all  packets  regardless  of  
their  size,  with  message  segmentation  the  total  amount  of  
header  bytes  is  more.  
 
Problem  2:    
Suppose  within  your  Web  browser  you  click  on  a  link  to  obtain  a  Web  page.  The  IP  
address  for  the  associated  URL  is  not  cached  in  your  local  host,  so  a  DNS  lookup  is  
necessary  to  obtain  the  IP  address.  Suppose  that  n  DNS  servers  are  visited  before  
your  host  receives  the  IP  address  from  DNS;  the  successive  visits  incur  an  RTT  of  
RTT1,  …,  RTTn.  Further  suppose  that  the  Web  page  associated  with  the  link  contains  
exactly  one  object,  consisting  of  a  small  amount  of  HTML  text.  Let  RTT0  denote  the  
RTT  between  the  local  host  and  the  server  containing  the  object.  
a. Assuming  zero  transmission  time  of  the  object,  how  much  time  elapses  from  
when  the  client  clicks  on  the  link  until  the  client  receives  the  object?  
Now  suppose  the  HTML  file  references  eight  very  small  objects  on  the  same  server.  
Neglecting  transmission  times,  how  much  time  elapses  with  
b. Non-­‐persistent  HTTP  with  no  parallel  TCP  connections?  
c. Non-­‐persistent  HTTP  with  browser  configured  for  5  parallel  connections?  
d. Persistent  HTTP?  
e. Suppose  you  can  access  the  caches  in  the  local  DNS  servers  of  your  
department.  Can  you  propose  a  way  to  roughly  determine  the  Web  servers  
(outside  your  department)  that  are  most  popular  among  the  users  in  the  
department?  Explain.  
 
Solution:  
a. The  total  amount  of  time  to  get  the  IP  address  is   .  
Once  the  IP  address  is  known,    elapses  to  set  up  the  TCP  connection  
and  another    elapses  to  request  and  receive  the  small  object.  The  total  
response  time  is    
ECE374:  First  Midterm                                                                                                                                                                                                                                        3  
 
b. .
c.
d. .
 
Problem  3:  
Consider  sending  a  large  file  from  a  host  to  another  over  a  TCP  connection  that  has  
no  loss.  
a. Suppose  TCP  uses  AIMD  for  its  congestion  control  without  slow  start.  
Assuming  cwnd  increases  by  1  MSS  every  time  a  batch  of  ACKs  is  received  
and  assuming  approximately  constant  round-­‐trip  times,  how  long  does  it  
take  for  cwnd    to  increase  from  5  MSS  to  11  MSS  (assuming  no  loss  events)?  
b. What  is  the  average  throughput  (in  terms  of  MSS  and  RTT)  for  this  
connection  up  through  time  =  6  RTT?  
 
Solution:  
a. It  takes  1  RTT  to  increase  CongWin  to  6  MSS;  2  RTTs  to  increase  to  7  MSS;    
3  RTTs  to  increase  to  8  MSS;  4  RTTs  to  increase  to  9  MSS;  5  RTTs  to  
increase  to  10  MSS;  and  6  RTTs  to  increase  to  11  MSS.  
b. In  the  first  RTT  5  MSS  was  sent;  in  the  second  RTT  6  MSS  was  sent;  in  the  
third  RTT  7  MSS  was  sent;  in  the  forth  RTT  8  MSS  was  sent;  in  the  fifth  RTT,  
9  MSS  was  sent;  and  in  the  sixth  RTT,  10  MSS  was  sent.  Thus,  up  to  time  6  
RTT,  5+6+7+8+9+10  =  45  MSS  were  sent  (and  acknowledged).    Thus,  we  
can  say  that  the  average  throughput  up  to  time  6  RTT  was  (45  MSS)/(6  
RTT)  =  7.5  MSS/RTT.    
 
Problem  4:  
Consider  distributing  a  file  of  F  =  15  Gbits    to  N  peers.  The  server  has  an  upload  rate  
of  us  =  30  Mbps,  and  each  peer  has  a  download  rate  of  di  =  2  Mbps  and  an  upload  rate  
of  u.  For  N  =  10,  100,  and  1,000  and  u  =  300  Kbps,  700  Kbps,  and  2  Mbps,  prepare  a  
chart  giving  the  minimum  distribution  time  for  each  of  the  combination  of  N  and  u  
for  both  client-­‐server  distribution  and  P2P  distribution.  
 
Solution:  
For calculating the minimum distribution time for client-server distribution, we use the
following formula:

Dcs = max {NF/us, F/dmin}


Similarly, for calculating the minimum distribution time for P2P distribution, we use the
following formula:

Where, F = 15 Gbits = 15 * 1024 Mbits


us = 30 Mbps
dmin = di = 2 Mbps
Note, 300Kbps = 300/1024 Mbps.

Client Server
N
10 100 1000
300 Kbps 7680 51200 512000
u 700 Kbps 7680 51200 512000
2 Mbps 7680 51200 512000

Peer to Peer
N
10 100 1000
300 Kbps 7680 25904 47559
u 700 Kbps 7680 15616 21525
2 Mbps 7680 7680 7680
 
 
Problem  5:  
Compare  GBN,  SR,  and  TCP  (no  delayed  ACK).  Assume  that  the  timeout  values  for  all  
three  protocols  are  sufficiently  long  such  that  5  consecutive  data  segments  and  their  
corresponding  ACKs  can  be  received  (if  not  lost  in  the  channel)  by  the  receiving  host  
(Host  B)  and  the  sending  host  (Host  A)  respectively.  Suppose  Host  A  sends  5  data  
segments  to  Host  B,  and  the  2nd  segment  (sent  from  A)  is  lost.  In  the  end,  all  5  data  
segments  have  been  correctly  received  by  Host  B.  
a. How  many  segments  has  Host  A  sent  in  total  and  how  many  ACKs  has  Host  B  
sent  in  total?  What  are  their  sequence  numbers?  Answer  this  question  for  all  
three  protocols.  
b. If  the  timeout  values  for  all  three  protocols  are  much  longer  than  5  *  RTT,  
then  which  protocol  successfully  delivers  all  five  data  segments  in  shortest  
time  interval?  
 
Solution:  
a. GoBackN:
A sends 9 segments in total. They are initially sent segments 1, 2, 3, 4, 5 and later
re-sent segments 2, 3, 4, and 5. B sends 8 ACKs. They are 4 ACKS with sequence
number 1, and 4 ACKS with sequence numbers 2, 3, 4, and 5.

Selective Repeat:
A sends 6 segments in total. They are initially sent segments 1, 2, 3, 4, 5 and later
re-sent segments 2. B sends 5 ACKs. They are 4 ACKS with sequence number 1,
3, 4, 5. And there is one ACK with sequence number 2.

TCP:
A sends 6 segments in total. They are initially sent segments 1, 2, 3, 4, 5 and later
ECE374:  First  Midterm                                                                                                                                                                                                                                        5  
 
re-sent segments 2. B sends 5 ACKs. They are 4 ACKS with sequence number 2.
There is one ACK with sequence numbers 6. Note that TCP always send an ACK
with expected sequence number.
b. TCP. This is because TCP uses fast retransmit without waiting until time out.
 
Problem  6:  
For  this  problem  you  should  familiarize  yourself  with  Figure  2  first.    Now  assume  
that  in  the  network  shown  in  Figure  2  two  parallel  TCP  transmissions  are  
performed.  TCP1  is  a  transmission  between  Source  A  and  Sink  A  that  uses  TCP  
Tahoe.  TCP2  is  a  transmission  between  Source  B  and  Sink  B  that  uses  TCP  Reno.  
Initial  ssthresh  for  both  TCP  transmissions  is  set  to  32.  In  this  specific  scenario  no  
additional  delay  through  forwarding  is  introduced.  Thus,  the  RTT  is  only  composed  
of  the  sums  of  the  delay  indicated  on  each  link,  times  two.  
 
a. For  the  TCP  1  transmission,  draw  the  resulting  congestion  window,  assuming  
that  a  packet  loss  (triple  duplicate  ACKs)  is  detected  at  time  t=900ms.  
b. For  the  TCP  2  transmission,  draw  the  resulting  congestion  window,  assuming  
that  a  packet  loss  (triple  duplicate  ACKs)  is  detected  at  time  t=650ms.  
c. Describe  the  benefit  of  TCP  Reno  over  TCP  Tahoe.  
 
 
 
 
 
 
 
 
 
 
 
 

Figure  2  Network  layout  for  problem  5.  


 
Figure  3  Solution  

 
Solution:  
ECE374:  First  Midterm                                                                                                                                                                                                                                        7  
 

 
 

You might also like