Sample Questions for CIS 451, quizes and final. Questions from Tanenbaum you have handed in as homework are likely to be asked. Or minor modifications. Questions from Tanenbaum you have not handed in as homework are also likely to be asked, but not quite as likely as the ones above. Further sample questions: 1. Briefly describe the OSI Reference Model. (And when I say ``briefly'', I mean it!) (And when I forget to say ``briefly'', I still want it pretty compact!). 2. What does ``OSI'' stand for? Similarly: ISO, PCM, POTS, PSTN, VoIP, ... . 3. What do msec, mu.sec, nsec stand for/ How long a period does each denote? 4. What does LAN stand for? MAN ? WAN ? 5. Give an example of a LAN , MAN , WAN. 6. Briefly, describe what a store-and-forward transmission system is. 7. Give an example of a connection-oriented service. 8. Give an example of an application layer service. 9. Briefly, describe what Nyquist's theorem says. 10. Briefly, describe what Shannon's main result says. 11. Describe how ``signal to noise ratio'' is expressed in decibels. (Tanenbaum, p 89). Can the the signal-to-noise ration (in dB) be zero? negative? Hint: Think of the situation where the noise is stronger than the signal, but you can still somewhat make out the signal. Shannon's result still holds in this situation! 12. What is ``attenuation'' ? Be brief! 13. Describe how ``attenuation'' is expressed in decibels. (Tanenbaum, p 95). 14. For electromagnetic radiation, if the frequency is H Hz, what is the wavelength in vacuum? 15. Briefly describe the difference between ``Virtual Circuit'' packet switched and ``Full Address'' packet switched. 16. The ASCII representation of the character `a' is 1100001. What is the parity bit if you use even parity? 17. Suppose a packet of 1000 bytes (incl header) is sent from one computer to another over a 20 km coaxial cable with a bitrate of 1 Mbit/sec. How much time elapses from when transmission starts at the source until reception ends at the destination? Of this, what part is propagation delay? What part is serialization delay? In approximately the middle of the transmission, what fraction of the packet resides in the transmission link? Assume that the speed of light in vacuum is 300,000 km /sec, and that the speed of the signal in co-ax (2/3)*c. 18. Suppose a packet of 1000 bytes is sent over a 45 Mbit/sec link from an earthstation to a geostationary satelite. Distance = 35,000 km. (speed of signal is speed of light in vacuum). What is the propagation delay? What is the serialization delay? If the earthstation just keeps sending one packet after another, how many will be underway at the same time? 19. Approximately, what is the bitrate of a T1 link? 20. How many DS0-s in a T1? T1-s in a T3? 21. T1 uses byte interleaving, while T3 uses bit interleaving. What does that statement mean? 22. what does PCM stand for? 23. Briefly explain how PCM works. 24. Assume dogs can hear sound up to 18,000 Hz. If you had to design a digital Hi-Fi sound storage and transmission installation for dogs, and you had to stay within a transmission budget of 720 Kbits/sec, how many times a second would you sample, and how many bits per sample would you use? (No overhead for signalling, no other overheads). 25. Give an example of a Polar coding scheme. Also Bipolar, Biphase. 26. Briefly, describe Manchester coding. Draw the picture of the voltage applied for the sequence 0 0 0 1 1 1 0 1 . (If I ask a question like this I will probably add a hint: zero is downward, one is upward. Tanenbaum p 275, fig 4-16 b has it wrong.) 27. Do some examples with the Internet Checksum. (See homework). 28. Give other names for the ``stop and wait'' protocol. 29. In the ``stop and wait'' protocol as handled in class, the correct procedure is: If the destination receives Frame(seq) with seq == dest_hacked, the destination MUST send ACK(seq). If the source receives ACK(seq) with seq == source_hacked , the source MUST NOT re-send Frame(seq + 1). For all three other combinations, give an example of how eventually this leads to ``bad things'' (even if further the protocol does the right thing). For each, a picture and a few words should be enough! 30. Briefly (!) explain under what circumstances ``stop and wait'' (assuming a ``correct'' implementation) is a good protocol to use. Two pictures and a few words should be enough! 31. Give the basic idea of a ``sliding window'' protocol. 32. Suppose two computers S and D (Source and Destination) are exactly 100 km from each other. They talk to each other over a 48 Mbit/sec dedicated link (full duplex) using 1500 Byte data frames and 50 Byte acknowledgement frames. Suppose the signal propagates over the link at (2/3)*c. Practically speaking, S is always sending data frames and D is only sending acknowledgement frames. The two computers are using a sliding window flow control protocol. What size must the sliding window have (expressed in frames) to make sure these computers use the link efficiently? Hint: the best solution is 6 frames, not 4 or 5. This is because D can not send an ACK until it has the WHOLE data frame it is going to acknowledge. And S can not utilize an ACK to get the right to send more data until it has the WHOLE ACK. If the processors are slow, a sliding window of 7 or even 8 frames may even be better. 33. Suppose you are sending frames over a link. Suppose there is high quality error detection. Suppose the probability of the arriving frame containing an error is p. In average, how often is each frame transmitted before it arrives without error? DERIVE the formula you are using. 34. What are the purposes of having a sliding window-type protocol? (This includes stop-and-wait as special case: window = 1 frame.) 35. Describe the role of the Receive Window and the Send Window. In a connection using a sliding window protocol, does it make sense to have the Window larger than either the Send_Window or the Receive_Window? 36. What does ATM stand for? SAR? AAL? 37. What is the total size of an ATM cell? (header plus data). Of this, how much is header? How much is data? 38. Briefly, explain why in ATM the Virtual Path Identifier of a cell may change every time the cell goes through an ATM switch. 39. Suppose we have ``voice over ATM'' and 47 data bytes in the cells are indeed used for voice (one of the 48 data bytes is used for SAR), and suppose further the voice is PCM encoded, 64,000 bits/sec. What is the ``cellification delay''? 40. OC-1 is 51.84Mb/s (raw), of which 49.536Mb/s is available to the users. On an OC-3 connection, what is the serialization delay of an ATM cell? 41. What does QPSK stand for? PSK? QAM? 42. Make a drawing of the phase - amplitude diagram of 16-QAM with 8 phases and 2 amplitudes. 43. In 16-QAM, what is the highest number of bits per baud you can send? BRIEFLY explain. 43. Give an example of an error correcting code which sends 3 bits (total) for every data bit, and which can correct all single bit errors. 44. Give an example of an error correcting code which sends 5 parity bits for every 4 data bits, and which can correct all single bit errors. (Single bit errors per ``codeword''). 45. What is a RAID ? 47. BRIEFLY explain how in a RAID no data is lost if a single disk breaks down. 48. What does HDLC stand for? 49. In HDLC, the pattern 01111110 is used as delimiter. (SoF as well as EoF). Suppose you want to send the following pattern as data: 0011000111110011111110011111111111100 What do you actually send? 50. In the situation of the preceding question, unstuff the pattern 0011111011111001101111001111101111101100 51. BRIEFLY, explain why in HDLC there is only one address field in the frame header. Is this a source address or a destination address? Or ... ? 52. What does PPP stand for? PtP? What is the relationship / difference between PPP and PtP? 53. What does DSL stand for? ADSL? 54. What does MACA stand for? MACAW? 55. BRIEFLY describe the ``Hidden Station Problem''. Also the ``Exposed Station Problem''. 56. In the 802.11 MAC layer Frame there is space for 4 addresses. When could one need 4 addresses? 57. Suppose we have a PPP connection over modems, from your home to a telephone central office, less that 5 km away. Would you recommend using ``stop and wait'' or ``go back n''? If you recommend ``go back n'', what window size would you recommend? In PPP, the payload can be at most (2^16 - 1) = 65535 bytes long, so the PPP frame can be at most 65545 bytes long (incl flags), but often frames are 100 - 2000 bytes long. Assume the modem has a speed of less than 56Kb/s. 58. Suppose we have a PtP connection over a Geostationary Satelite (Earth - Satellite - Earth, with in the satellite only a dumb transponder). Suppose the framesize is at most 2000 bytes. Answer the same questions as in the preceding problem. The bandwidth is 100 Mb/s. Put in plausible numbers for propagation delays. 59. What does IMTS stand for? AMPS? D-AMPS? GSM? 60. Describe how D-AMPS uses both FDM AND TDM. 61. AMPS and D-AMPS have a bandwidth re-use factor of 7. What does that mean? 62. Give a BRIEF explanation of why you would expect AMPS and D-AMPS to have a bandwidth re-use factor of 7. 63. In AMPS there are 832 channels. If a single provider ``owns'' a territory, it has 21 control channels and 811 voice channels. If two providers equally share a territory, each has 21 control channels and 395 voice channels. Why is the number of control channels doubled when providers share a territory? 64. What does MSC stand for? MTSO? 65. Suppose you live in NJ and have bought cellphone service with roaming priviliges all over the USA and Canada. You are on an unexpected business trip to California and have not told anybody. You leave your phone on all the time. (But in the plane you turned it off, because the pilot told you to). Your aunt from Texas calls your cellphone number. Describe BRIEFLY how the connection reaches your cellphone in California. (How does the system find you in California?) 66. Describe BRIEFLY what is meant by the ``dogleg'' in mobile telephony. You can use the case in the previous problem as example. A small drawing may help. 67. Suppose your provider has implemented mobile telephony WITH dogleg. Describe what happens if your business contact in San Francisco calls you on your cellphone number while you are walking to her office. 67. Having no dogleg may save some transmission resources. Do you think it may have consequences for privacy? (BTW: I do not know whether in the USA there is a dogleg. I am moderately sure that in Europe there IS a dogleg). 68. BRIEFLY describe how, in cellular telephony, smaller cells allow lower transmission power, and how the combination of lower power and smaller cells together allow better bandwidth utilization. 69. In mobile telephony with powercontrol the base station lets the cell phone (indirectly) know how strong a signal it receives, and the cellphone lets the base station (indirectly) know how strong a signal it receives. Then both adapt output power to roughly the minimum needed for good quality transmission. BRIEFLY describe how powercontrol improves bandwidth utilization. BTW (not done in class) yet another way of improving bandwidth utilization is for base stations to have directional antennas. (Using a so-called phased array, the base station transmits in a beam directed at the mobile unit.) BRIEFLY describe how powercontrol together with directional transmissions further improves bandwidth utilization. BTW: for the time being, mobile units can not do directional transmission. 70. BRIEFLY explain what Secrecy, Authentication, Integrity, Non-Repudiation stand for (in the context of network security). 71. Suppose we have a perfect single key (symmetric key) encryption/ decryption scheme. BRIEFLY, explain which of the four results in the previous problem it garantees. Are there additional requirements? 72. We have a perfect public key encryption system. Person A has public key U_A and private (secret) key R_A. Similarly, person B has U_B and R_B. For convenience, we use the same symbols for the keys U_A etc and the maps (transformations) generated by U_A etc. Suppose A wants to send a message P (plaintext) to B that generates secrecy, authentication, and non-repudiation. (And with some care also integity, not to be worried about in this problem). Which of the following messages does A send to B? a. R_B(U_A(P)), b. U_A(R_B(P)), c. U_B(R_A(P)), d. R_A(U_B(P)). WARNING: this is NOT how it is really done, because of ``man in the middle'' attacks, people advertising fake public keys pretending to be somebody else, etc. This excercise illustrates the basic ideas but does not go far enough. See Tanenbaum sections 8.3, 8.4 (homework for the Christmas break!). Model response to problem 72: a and b are nonsense: they require A to know R_B. c and d both look plausible. However, deeper study shows c has a flaw. I recommend you re-do the solution below in handwriting, with subscripts (U.sub.A instead of U_A). That way it will look much clearer! First the plausibility: B receives the message. In c: B computes R_B(U_B(R_A(P))) = R_A(P) and keeps it. Then B computes U_A(R_A(P)) = P. B has the plaintext. B has P as well as R_A(P) and only A could have done that transformation: authentication. In d: B computes U_A(R_A(U_B(P))) = U_B(P). Then B computes R_B(U_B(P)) = P. B has the plaintext. B has U_B(P) as well as R_A(U_B(P)) and only A could have done that transformation: authentication. Both also ``garantee'' confidentiality. (R_B is needed in the decryption). Now the non-repudiation. In c, in order to prove in court that the message must have come from A, B must show the court that R_B applied to U_B(R_A(P)) indeed gives R_A(P), and then that U_A applied to R_A(P) indeed gives P. However, the first step is possible only by making R_B public. In the next 5 minutes, ``everybody'' generates messages ``provably'' from B. AND decrypts all recent messages to B encrypted with U_B only. In d, B keeps P and R_A(U_B(P)). In case of a court case, B shows that U_B applied to P and U_A applied to R_A(U_B(P)) give the same result. This proves A must be the sender, without exposing B to further attacks. For a slightly better introduction, see the sections in Tanenbaum. This stuff is really used! (But in a more sophisticated way than described above). Aside: Public Key encryption is very good but thus far too slow for routine encryption of large messages. A typical modus operandi seems to be to use public key encryption to encrypt and transmit keys for symmetric key systems, and then use the symmetric key systems for encryption, transport, and decryption of actual messages. 73. Message digests will not be asked on the final. But read Tanenbaum sections 8.3 and 8.4 over the Christmas break! 74. In the Irridium Satellite communication system, ``Routing and Switching'' is done in space. What did this (true!) statement mean? 75. In the Globalstar Satellite communication system, the satellite link is strictly a ``bent pipe'', and all routing and switching is done terrestrially. What did this statement (assuming it is true) mean? (This statement used to be true. It may still be true, I am not entirely sure.) More to be added. I will add a few questions on ``security and related topics'' and on the final there will be at least one question on ``security and related topics''. As before: minor variations on questions asked in previous quizes are likely! Minor variations on homework questions are likely!