Two hours UNIVERSITY OF MANCHESTER SCHOOL OF COMPUTER SCIENCE. M.Sc. in Advanced Computer Science. Date: Monday 2 nd June 2008.

Transcription

1 COMP60242 Two hours UNIVERSITY OF MANCHESTER SCHOOL OF COMPUTER SCIENCE M.Sc. in Advanced Computer Science Mobile Computing Date: Monday 2 nd June 2008 Time: 09:45 11:45 Please answer any THREE Questions from the FIVE questions provided Please use a separate answer book for each Question. This is a CLOSED book examination The use of electronic calculators is permitted provided they are not programmable and do not store text [PTO]

2 Page 2 of 5 COMP a) How does WiMax (IEEE ) avoid interference from hidden and exposed nodes? Do WiMAX systems cause interference to other systems b) Is the Request To Send / Clear To Send (RTS/CTS) protocol better at avoiding hidden and exposed terminals than either Carrier Sense Multiple Access (CSMA), or as used in IEEE , CSMA with Collision Avoidance (CA)? Using simple diagrams explain why this is. c) Explain why error detection and forward error correction (FEC) are used simultaneously at the data-link layer on IEEE WLAN networks, whereas only error detection is generally used on wired networks. d) Why are convolutional rather than block codes currently used for FEC in wireless networks? If a convolutional coder has two generator functions expressed in octal as (13) and (15), what is the rate of the coder and what is its constraint length? Draw a diagram for the coder and calculate the first 8 bits of its output when the first 4 bits of the input are What is a systematic coder and could this particular convolutional coder be described as systematic? 2. a) Do you agree that today, it is not sensible to run a web server from a mobile device to support, for instance, browsing and purchasing goods? What needs to change in the network infrastructure and the protocols used in order for this scenario to become more sensible in the future? b) Briefly, describe the mechanisms in IPv4 and IPv6 for handover control of mobile network clients. Outline what needs to happen in a typical laptop computer today when the computer moves from one connection service to another. c) When must mobile computers change their IP addresses? How is this change in IP address achieved without user interaction? How do TCP and SCTP differ in how a transport connection responds to large changes in throughput or the loss of a connection such as may happen when a mobile device moves to a new network connection? What are the disadvantages and advantages of each approach?

3 Page 3 of 5 COMP a) A sensor network is being designed to allow monitoring of the temperature and humidity of fresh produce (fruit and vegetables) from the time it is picked until it is sold by a retailer by placing a wireless sensor in each sack. You may assume that a query attempts to collect data from each deployed sensor every five minutes and that sensors have very little processing power and only a small memory. Briefly, describe the role of both static and mobile network components in the implementation of this type of network. How will the device constraints mentioned in the Appendix affect the operation of the system? Using annotated diagrams and brief descriptions suggest a suitable architecture for this sensor network. Pay particular attention to the protocols required, where these protocols are located in the network and the network s components. (12 marks) Appendix for Question 3 Description of a typical sensor (from The MICA mote (sensor) uses an Atmel ATmega 128L processor running at 4 megahertz. The 128L is an 8-bit microcontroller that has 128 kilobytes of onboard flash memory to store the mote's program. This CPU is about as powerful as the 8088 CPU found in the original IBM PC (circa 1982). The big difference is that the ATmega consumes only 8 milliamps when it is running, and only 15 microamps in sleep mode. The mote has a 10-bit A/D converter so that sensor data can be digitized. The radio in the mote has a range of several hundred feet and can transmit approximately 40,000 bits per second. When it is off, the radio consumes less than one microamp. When receiving data, it consumes 10 milliamps. When transmitting, it consumes 25 milliamps. Conserving radio power is key to long battery life. This low power consumption allows a MICA mote to run for more than a year with two AA batteries. A typical AA battery can produce about 1,000 milliamp-hours. At 8 milliamps, the ATmega would operate for about 120 hours if it operated constantly. However, the programmer will typically write code so that the CPU is asleep much of the time, allowing it to extend battery life considerably. For example, the mote might sleep for 10 seconds, wake up and check status for a few microseconds, and then go back to sleep. (Question 3 continues on the following page)

4 Page 4 of 5 COMP60242 (Question 3 continues from the previous page) b) In the following three environments, what information about neighbouring network hosts can a typical IPv4 host collect? You should assume that the host is connected as a network leaf node. In each case state your assumptions about which protocols are used to collect the information. i) In a wired network attached to the network via a hub. ii) In a wired network attached to the network via a switch. iii) In an IEEE (WiFi) wireless network where no encryption is switched on. In an IPv6 network, how would your answers to i-iii above differ? Does mobility affect the data available? 4. a) With respect to the packetised transmission of speech over wired and wireless computer networks, how do the requirements of interactive voice telephony differ from those of streaming? Explain why the transfer protocol TCP is rarely used for streaming or telephony, and explain why RTP is more commonly used with its sister protocol RTCP. b) A VoIP telephone link packetises 30 ms segments of G711 encoded speech and operates between Manchester & New York. Jitter-buffers are employed at each end and it is found that the average packet loss occurring at the New York end is 0% whereas the average packet loss occurring at the Manchester end is 10%. How can the speech quality be improved at the Manchester end without increasing the overall round trip delay? (2 marks) c) What is meant by packet loss concealment as used in voice over IP (VoIP) telephony? (2 marks) d) Explain why voice over IP (VoIP) as widely used over wired computer networks is not ideally suited to wireless LANs (voice over WI-FI). What problems can contention mode access (CSMA) cause when battery powered mobile equipment is used for voice over WI-FI? What features of IEE standards could, in principle, be employed to reduce these problems?

5 Page 5 of 5 COMP a) Explain, with a simple diagram, the operation of a binary PSK modulator. Explain why sinc-like base-band pulse-shaping is necessary with PSK modulation and explain how the base-band pulse shape affects the modulated carrier. In theory, what is the maximum achievable band-width efficiency with binary PSK if inter-symbol interference (ISI) is to be eliminated? Why is this maximum bandwidth efficiency difficult to achieve in practice? b) Why are 'root-raised cosine' (RRC) sync-like pulses used in single carrier transmitters? Determine the channel band-width required for a bit-rate of 2 Mb/s using binary PSK with 50% RRC pulse shaping. c) Explain how the bit-rate of 2 Mb/s referred to in part (b) could be increased to 4 Mb/s over the same bandwidth using a vector-modulator and the same 50% RRC pulse-shaping.. d) Explain how the bit-rate may be further increased from 4 Mb/s to 8 Mb/s over the same bandwidth using 16-QAM. Give a constellation diagram for 16-QAM. Given that the received signal is affected by additive white Gaussian noise and that the signal to noise ratio at the detector is 30 db, what, in theory according to the Shannon-Hartley Law, is the maximum bit-rate that can be transmitted with an arbitrarily low bit-error rate? END OF EXAMINATION