CS 640 - Final Exam Fall 1992

Questions 1-16 (True or False - 2 points each)
 
Questions 1-4 refer to 802.5
1. Automatic retransmission occurs when a frame returns to the sender with an 
incorrect FCS.
2. If the maximum bandwidth (in the standard) were increased, the minimum frame 
size would also increase.
3. Any station may become the active monitor.
4. Tokens are distinguished from frames by a single bit.

Questions 5-8 refer to 802.3
5. Every station reads every frame transmitted.
6. A station does not have to wait more than k slot times to transmit, where k 
is the number of stations attached to the bus. 
7. If the bandwidth (in the standard) were increased, the minimum frame size 
would also increase.
8. A sending station retransmits immediately after it detects a collision.

Questions 9-11 refer to FDDI
9. The receiving station removes the frame from the ring.
10. At most one token can be on the ring at any time.
11. Synchronous traffic cannot by a station if the Late_Count (LC) is greater 
than zero when it receives the token.

12. The 802.2 protocol provides either a connection-oriented or a connectionless 
service interface.
13. A field in the 802.2 header specifies the upper (network) layer protocol.
14. Bandwidth reservation is required for virtual circuits but not for datagrams.
15. The network layer manages buffer allocation and deallocation for the link layer.
16. The bandwidth delay product is a measure of how much pipelining is needed for 
efficient operation of a window oriented transport protocol.

Questions 17-38 (fill in the phrase that completes the statement - 2 points each)

One reason for having sending stations rather than receiving stations remove 
frames from an 802.5 ring is (17) _____  .

With CQL, buffers can be completely allocated to or shared between (18) _____ .

CQL prevents (19) _____ deadlock but is susceptible to (20) _____ deadlock.  

Hop level deadlocks are prevented when using virtual circuits by (21) _____  ,

A disadvantage of using VCs in a subnet is (22) _____ .

A disadvantage of using DGs in a subnet is (23) _____ .

Assume a 100 ms round-trip time and a network bandwidth of 400 megabits per second. 
The maximum TCP throughput in this environment is (24) _____ .

If a receiver agrees to use the TCP window scale option with its shift.cnt equal 
to 4, then its maximum receive window is (25) _____ .

Timestamps are used with the TCP echo option to help with (26) _____ and (27) _____ .

After receiving acknowledgments of six successive packets it has sent, a sender 
using "slow start" has a send window of at least (28) _____ .

Van Jacobson proposes that a TCP sender multiply its send window by a fraction less 
than one when (29) _____ .

IP datagram reassembly occurs at the  (30) _____ .

Subnetting is used with IP Class (31) _____ networks in order to (32) _____ .

An IP Internet Address  has two fields, a (33) _____ field and a (34) _____ field.

Class B and Class C IP Internet Addresses were introduced because (35) _____ . 

If an IP datagram is too large for the network protocol of the next network
if must traverse, then it is (36) _____ .

The IS-IS and RIP protocols are examples of  (37) _____ routing protocols.

A responsibility of the 802.5 active monitor station is to (38) _____.

39. (10 points) Assume the network of Figure 1 with links labeled by costs. Assume 
use of SPF. Show the order in which nodes are added to S where S contains nodes 
for which the shortest path from A has been found.  For each node in S, give 
its predecessor on the path from A, the source node.

40. (10 points) Assume the network of Figure 2 where all links have cost equal 
to one. Consider the routing tables entries at nodes B, C, D, E for destination 
A. Assume use of a distance vector algorithm where routing tables are always 
exchanged at the same time by all nodes. 

After the link B <----> A fails, the entries for destination A are:

	at B	infinity
	at C	B 2
	at D	C 3
	at E	B 2

a) Assume split horizon and hold-down are not used. Give the values for the  
routing table entries at B, C, D, E for destination A after each of the next 
two routing table exchanges.
b) Repeat a) using split horizon.
c) Repeat a) using hold-down and split horizon.

41. (9) Three parts; 3 points for each part; no partial credit.
a. Assume a generator polynomial G(x) = x**5 + x**4 + x + 1 and a frame 110100011. 
Compute the CRC for this G(x) and frame.
b. Give the shift register circuit for the G(x) of part (a).
c. Assume data strings of length 10. Give the single bit error correcting Hamming 
code for the data string 1011001111.