CS/EE 5516, Spring 1992, Exam 1 Name: ________________________

Prof. Abrams

Use your own paper to solve the exam; scrap paper is available in the front of the room. Start each answer on a new page. Arrange the answer pages in problem number order, and staple this sheet to the front of your exam. Write your name on each page; staples sometimes fail.

Don’t spend more than 15 minutes on any problem until you have tried all the problems! The exam is open book (Bertsekas and Gallager, 2nd edition, only) and lasts for 75 minutes. Good luck!

1. Bertsekas and Gallager 2.30

2. Draw a timing diagram for a go-back-4 DLC protocol similar to Figure 2.25 for the first 13 time units of the following scenario. End points A and B communicate over a bidirectional channel. DLC sender A accepts 10 packets from the network layer at time 0; DLC sender B accepts no packets prior to time 0.5 and accepts 10 packets at time 0.5. Both senders never again accept packets from the network layer. The transmission delay for each frame is 1 time unit. The propagation delay for each frame is 2 time units. All acknowledgements are piggy-backed on reverse traffic. Each frame has its own timer of duration 6 time units that is started when frame transmission starts. The second frame from A to B is lost.

Be sure to label the time axis and show in your diagram the values of SN and RN for A and B, the sender window sizes at A and B, and when each packet is delivered to the network layer by A and B.

3. Consider aggregating the traffic from 10 sessions over a point to point communication link with a data rate of 75,000 bits per second. All sessions are statistically identical and independent Poisson processes. Each session generates packets at a rate of 2.5 per second. Packet lengths are exponentially distributed with a mean of 1,500 bits. Find the average delay per packet, the average number of packets in the system, and the average number of packets queued in the system when the line is allocated to the sessions using (a) statistical multiplexing and (b) assigning each session to each of 10 equal capacity time-division multiplexed channels. Show your work and draw a box around the final six numerical answers. (c) Does case (a) or (b) perform better, and what is the intuitive reason for this? (Hint: If you need a calculator, your answer is wrong.)

4. Consider bit-oriented framing. What is the consequence of corruption of the end-of-frame flag? How can this be detected?

5. "The president of the Specialty Paint Corp. gets the idea to work together with a local beer brewer for the purpose of producing an invisible beer can (as an anti-litter measure). The president tells his legal department to look into it, and they in turn ask engineering for help. As a result, the chief engineer calls his counterpart at the other company to discuss the technical aspects of the project. The engineers then report back to their respective legal departments, which then confer by telephone to arrange the legal aspects. Finally the two corporate presidents discuss the financial side of the deal. Is this an example of a multilayer protocol in the sense of the OSI model?" [problem due to A. Tanenbaum]

Justify your answer. If your answer is "yes," draw a diagram similar to the OSI model and label the diagram components by the entities in this problem (i.e., president, etc.). Label some part of your diagram as "virtual," as is done in Bertsekas and Gallager Fig. 1.8, if something analogous occurs in the example. If your answer is "no," draw the OSI model and explain how it is violated.

6. Bit stuffing uses a flag of 01n0, where n=6. Suppose a colleague proposed using n=216. Would you tell him/her that n=216 is a good or a bad idea? Give a good (but brief) technical justification for your answer.

Extra credit:

[A correct solution to part (a) and part (b) will each add 1/2% to you final, cumulative grade in the class. (The final, cumulative grade is based on a 100% scale.)]

Consider the selective repeat protocol in which the receiver always tells the sender which packets are missing and need to be repeated. The sending DLC is guaranteed to find out wheter a given transmitted packet was successfully recieved during or before the bth frame transmission following the frame in which the given packet was transmitted.

(a) What rule can be added to the sender and/or receiver to guarantee that the sending DLC never save more than b+1 unacknowledged packets? (You may want to state you rule in terms of the go-back-n algorithm on pages 75-76 to state the modification more precisely.)

(b) Outline a proof that that the rule(s) added in (a) guarantees the property stated in (a).