ECE/CSC 570 Section 001 Final test December 11, 2006 Questions 1 10 each carry 2 marks. Answer only by placing a check mark to indicate whether the statement is true of false in the appropriate box, and leaving the other box blank. Questions 11 20 each carry 5 marks. Answer in the space provided; provide reasonable explanations where called for, but make your answer brief. Questions 21 and 22 each carry 20 marks. Answer in the space provided, providing adequate intermediate steps or reasoning you have to use. Maximum score assigned will be 100 scores higher than this will be saturated to 100. This is a three hour test. First Name: Last Name: Student ID (9-digit): Pledge of Honor: I pledge on my honor that I have answered the questions in this test purely on my own individual ability, and have neither given help to nor received help from another person taking the test. Signature:
1. Layers in a networking stack may introduce headers and trailers, and may also change the bits of the payload. TRUE FALSE 2. A connection-oriented end-to-end service can only be implemented if the underlying network is a circuit-switched network. 3. The address 152.273.15.1 may be a valid IP address. 4. Ethernet, slotted ALOHA, and IEEE 802.3 do not all refer to exactly the same networking protocol. 5. A stronger Error Control Coding scheme always requires a longer ECC (more bits) than any weaker scheme requires. 6. Re-transmission of lost packets is a major function of the networking layer. 7. Virtual-circuit oriented networks have to utilize the same concept of store and forward as IP networks. 8. If all wireless links are symmetric (A can hear B implies B can hear A) in a wireless ad-hoc network, then the hidden station problem cannot occur. 9. The Internet Protocol (IP) is a forwarding protocol, not a routing protocol. 10. In real networks, propagation delays can always be ignored when compared to transmission delays.
11. A block-oriented Error Control Code operates as follows. The blocks are composed of 4 bits u 0, u 1, u 2, u 3. Three ECC bits are appended: r 0 = u 0 u 1 u 2 u 3, r 1 = u 0 u 1, and r 2 = u 0 u 2. State (with reasoning) what the minimum distance between valid codewords is for this code, and write down two codewords that have this minimum distance. 12. A datagram was fragmented multiple times during its journey from source to destination into an unknown number of total fragments. Four fragments were received by the destination. The offset and size fields of the IP header of the fragments were (75,400), (125, 600), (0,600) and (250,200). The fourth of these fragments also had the MF flag set to 0. (a) Is it possible for the destination to know how many other fragments may be expected for this datagram, and if so what is this number? (b) Is it possible for the destination to know the size of user data in the original fragment, and if so, what is the size? 13. In 802.11, what is the logic for keeping the IFS spacing to be used for ACKs shorter than the IFS delay to be used for direct access to the medium? 14. Describe briefly how the use of virtual circuits in a virtual circuit switched network reduce the forwarding overhead. 15. A corporate network purchases the Class B address space 152.1. It needs to use subnetting to distinguish between its 60 internal networks, each of which has 120 hosts. What are the different subnet masks it can use? Answer by writing down the actual masks, in /length notation.
16. Consider the rigorous analysis of idealized slotted ALOHA. A Markov model for a system with five stations is shown below, with the usual representation of the number of backlogged nodes as the state. 0 1 2 3 4 5 Argue whether any two of all the state transitions shown may have exactly the same probability. Assume that the probabilities of a new arrival q a and retransmission by a backlogged station q r are different. 17. When a collision occurs in an ALOHA like protocol, each station which had attempted to transmit retries after a backing off a random amount of time. Assume that the backoff time is deterministic rather than random. What is the problem that arises? 18. Consider a land wireless connection with a each-way propagation delay of 10 microseconds between the sending and receiving stations and a transmission delay of a frame of 1 microsecond. The transmission time for an acknowledgment is negligible. What is the minimum window size of Go Back N that would (over a period of time during which no errors happen to occur) result in an efficiency of 100%? 19. A large number of consecutive IP addresses are available starting at 198.16.0.0. When IP addresses are requests, they are always assigned at the earliest (lowest) possible open block. Suppose that four organizations, A, B, C and D, request 4000, 2000, 1000 and 8000 addresses respectively, and in that order. Would the blocks assigned to each organization have changed if the order of requests had been (a) A, C, B, D? (b) A, D, B, C?
20. Dijkstra s algorithm is being used to find the shortest path from node A to every other node in a network with 10 nodes, labeled A to J. At an intermediate step in the execution, it is found that the current best estimates of distance to the other nodes is as follows. The shaded boxes indicate that the nodes A, D, E and G are already in the confirmed set. To A To B To C To D To E To F To G To H To I To J 0 6 4 8 12 2 17 11 (a) What is the next node to be picked? Assuming this node has links to nodes C, E, I and J, which have link costs 4, 1, 9 and 10 respectively, which distances will be updated in the above table? (b) Is it possible to say whether the situation as shown above can ever occur in correct operation of Dijkstra s algorithm on a real network topology? 21. Consider a Selective Repeat ARQ system with W=5. Assume that the sender sends the first frame (sequence numbered 1) at time 0. The transmission time for all frames is the same and is equal to τ, ρ is the propagation delay for either frames or ACKs, and ρ = 4τ. The transmission time is effectively zero for ACKs. The sender uses a timeout of 10τ starting from the end of the frame transmission. (For purposes of avoiding simultaneity, it may be assumed that the timeout is marginally more than 10τ, not less.) Only ACKs are used, not NAKs. (a) Draw a timing diagram for this system, assuming that frame #2 is lost the first time it is transmitted, and frame #5 is lost both the first and the second time it is transmitted. No other frames and no ACKs are lost over the period of interest. Draw the diagram from time 0 to the time that frame #7 is correctly received and accepted at the receiver. (Use facing page.) (b) Show the contents of the sender s and receiver s buffer at every point of time when they change over the period of interest, indicating the time of change and the new contents.
22. The network shown below uses DV algorithm for routing. Assume that all the routers are booted up simultaneously, and at time 0 exchange the first round of distance vectors, in which every router advertises 0 distance to itself and to all others. As a consequence of this exchange, the routers also discover their neighbors and correctly measure the link costs as shown. Assume that from there on, every round of DV exchange proceeds exactly simultaneously (only the information from the previous round can be used, not from the current round). Show the DVs transmitted by each router in each round, until the DV algorithm converges. 2 C 4 3 D 3 F A 1 1 4 1 1 B 3 E 4 G