Mid Term Review

Mid Term Wed, April 4 –Open Book –Open papers –Open laptop In class exam –Duration 1 hour 15 minutes Exam level –Not too easy, not too hard –20% of total grade

–Wireless Networking CSMA, TDMA –Wireless MAC , Multi-channel, rate, antenna … –Mobile Computing Social apps, localization, mobility User interfaces Energy Location privacy –Routing Ad Hoc, opportunistic Syllabus

Approximate Exam Format Around 5 questions -- you will need to answer any 4 Q1. -- Set of true/false statements –You will need to justify your answer in one sentence Q2 to Q4 -- Topic questions with sub-parts –You may need to work out small problems –You may need to argue and/or identify flaws –You may need to give examples/counter-examples Q5. -- Design Question –Open ended –You will need to sketch the design of a protocol

Approximate Exam Format Questions should not require you to know paper details –You will not be asked to derive interference function in “Paritally Overlapping Channels not considered harmful” paper … –However, need to understand what is the need to compute an interference function

Some Focus Topics Understand Basics (Undergrad review) –Transmission time, propagation delay, Q delay –CSMA/CD Vs CSMA/CA, slotted/unslotted ALOHA –Layering –Dijkstra’s, bellman ford routing on graphs –UDP/TCP, Transport schemes (stop&wait, GBN, SRQ …) Understand well –Carrier sensing, Backoff, Hidden/Exposed terminals –RTS/CTS, ACK, NAV, Spatial Reuse Understand tradeoffs between CSMA/TDMA –Graph coloring as TDMA abstraction –Distributed TDMA

Some Focus Topics Directional Antennas, Multiple Channels –Beams, channels - their impact on spatial reuse –Deafness, HT, Capture, Overlapping channels Rate Control (RC) –Channel characteristic, Tradeoffs –Thrghput/Time fairness, Problem w/ ARF, RBAR, OAR –Need for decoupling collisions and fading

Some Focus Topics Routing –Understand DSR, LAR, Directed Diffusion –Routing metrics - impact on performance –Reactive vs proactive routing - when good/bad –Why ETX? Why ExOR? Tradeoffs Delay Tolerance –Understand the use of mobility, tradeoffs

Some Focus Topics Mobile Computing –Design choices in localization. Which is better when? –Energy inefficiencies in the WiFi –Impact of mobility in various aspects –Grammars in PhonePen –Difficulty with location privacy. Issues with path cloaking

Some Example Questions

Channel Utilization (T/F): For , let X be saturation throughput when a single user transmits to an AP. With increasing number of users, the aggregate throughput will first increase and then decrease, relative to X. –(assume that all users transmit identical traffic and transmit at a specified data rate).

Rate Adaptation (T/F): When the channel quality is constant, ARF performs better than RBAR –(I am open to explaining to you ARF or RBAR, if necessary)

IEEE Carrier Sense Some say that IEEE solves hidden terminal problems (HTP) by introducing the RTS/CTS mechanism. –1. List at least 2 counter-arguments. Use figures if necessary. –2. Calculate the carrier sensing threshold that must be used to ensure that a node that does not carrier sense an ongoing communication will also not interfere with the communication. (Assume SINR threshold as β = 1, path loss factor as α, and communication radius as R). –3. Assume that in Jupiter, signal power degrades linearly with distance (as opposed to exponentially with factor α). Will that be better or worse for the spatial reuse of the wireless network? Discuss in 2 or 3 sentences.

To Channel or Not to Channel You want to schedule traffic from laptops to a single AP You have two choices : (1)Make 2 channels, each 10 Mbps, and divide 50 people in each channel (2)Make a single channel of 20 Mbps, and have all 100 in the same channel. Which will you choose? Why?

Open Q: Location Proof You were supposed to meet John at Twinnies. You went there a bit late, and could not find John around. Later, John does not believe that you were late -- he accuses you of not showing up at all. You need to prove to John that you did go to Twinnies. Can you design a scheme that allows you to prove your location.

Open Q: 3G problems The 3G network is severly strained and WiFi hotspots are quite underutilized in general. –Design a scheme that will reduce the load on 3G networks by exploiting WiFi hotspots better.

Open Q: SurroundSense SurroundSense requires that all locations be fingerprinted first. This is a hurdle, and may affect quick deployment of the system. –Design a scheme to quickly create a database of ambience fingerprints for Durham. –Discuss why your scheme will work –Discuss where it may not work (identify assumptions you are making)

Some Non-Examples What is the size of RTS/CTS packets? When may DRAND not converge to a schedule? What fingerpriting technique used in SurroundSense? What is the interference function in the paper on “partially overlapping channels …” Describe the operations of DSR How is the expected zone defined in LAR protocol? These type of questions will not be asked

Please me your class ppt slides if you have not done so already

Any Questions? …

Some Example Questions A DTN protocol X is described. You are asked: –Order which mobility model will perform best, worst Random walk, random waypoint, manhattan, brownian (I am open to explaining the mobility models if needed) –What is the problem when X runs over Rand Waypoint –Modify X to make it suitable for RW You are given a scenario and application. You are asked to choose a routing protocol between DSR and AODV. –Justify your decision briefly.

Location Service

Some Example Questions You are given a graph, G. Also you are told that p_i > p_j if i>j (I.e., p1 < p2 < p3 …). You are also told which node is the gossip originator. –(a) You will be asked to assign most efficient probabilities to each node in G –(b) You will have to kill 3 nodes. Which 3 will you kill and retain max reliability? –(c) Write an algorithm that takes as input N and G p, and shows which N nodes should be killed.

Propagation and Transmission You are trying to determine the bottleneck link on a particular route. One technique used is known as packet-pair estimation which is as follows: (i) The source sends two equal-length packets back to back on its outgoing link. (ii) When the packets are received, they are found to be separated by some duration T (time duration from the start of the first packet to the start of the second packet). (iii) Observing T, and knowing the lengths of the packets (L), it is enough to calculate the bottleneck bandwidth. Assume no other traffic in network –1. Can packets ever reach the destination back to back? If no, explain why not. If yes, give an example. –2. Let bottleneck bandwidth be B. Calculate B as a function of T and L. –3. Why is it necessary to assume that there is no other traffic in the network? Answer very briefly. –4. Modify this scheme to achieve a good estimate of bottleneck bandwidth under presence of other network traffic.