ECEN 5032 - Communication Networks

Timothy X Brown, Spring 2006

Semester Project

• Choose a research question.
• Perform background research on the question.
• Work on the research question.
• Present the results of your research.

• We have industry standard simulation tools such as Qualnet or Opnet. Use one of these tools to show the effects of different protocol modifications. For instance, the behavior of TCP is strongly dependent on the backoff mechanism used.
• Networks are subject to various kinds of intentional and unintentional "jamming". What are the effects of this jamming on different communication layers.
• A modern theme in networking is that the distribution of file transfers, the number of links to web pages, and the number of connections to an AS all exhibit a power law distribution. This implies that the connectivity and performance of the network are dominated by outliers. Pick one of these metrics and test whether the distribution is well matched by a power law.
• 802.16 provides a mechanism for dividing connection resources between time and frequency resources. This allows for interesting bandwidth management where you can provide users different quality of service. Develop a mechanism to achieve this bandwidth management via MPLS or GMPLS.
• 802.11 is a fairly complex protocol and there are many subtle feature interactions that are not well understood. Some examples: The protocol can be unfair, some users can get a disproportionate amount of bandwidth depending on proximity to an access point or who gets the media first. When users communicate with an access point at different rates, some studies suggest the overall throughput is governed by the minimum data rate. Under what conditions is it possible to communicate with an access point in receive-only mode.
• Weighted fair queueing (WFQ) is a mechanism for dividing bandwidth resources between different classes of traffic on a packet network link. Write a WFQ schedular and show that it can protect one class of traffic from bursty or overloaded other classes. You can use the Click modular router software.
• There are many free tools which you can download for Linux. Netstumbler, or Airsnort can be used for wardriving to capture access point location information. Ethereal is a protocol analyzer that lets users study packets from a wide variety of wired and wireless protocols. Wepcrack breaks WEP protected 802.11 networks.
• The Click modular router is a program language that allows users to easily write complex packet protocols. Write a Mobile IP implementation, or write a program that serves up different webpages depending on location and other context information.
• There is quite some interest in war gaming these days. The war gaming simulator OTB 2.0 can manage troop (i.e., node) deployments, movement and status, and interact with Qualnet, which then does the traffic flow simulation. The combination of the two simulators would enable a thorough investigation of mobility control schemes to enhance wireless communications in sparsely connected wireless ad-hoc networks.

• Introduction: You should first describe your research question. State it as a question: "In this project we attempted to answer the question: ...". Throughout the introduction you will attempt to do the following with respect to your question. First be specific. clearly define the question so that the reader and you would agree on what is the question. More importantly, you need to describe well enough the criteria that you will use in the answer to the question so that, after seeing the results, you and the reader would agree on the answer. As part of being specific you should scope your question. What are you not answering? What pieces are you specifically ignoring? Second show the question is significant. Why should we (or anyone) care about the answer? Third show the question is contestable. By this, we mean that you should show that the answer to the question is not obvious or already well known.
  
  Showing your paper is specific will involve defining different terms and giving background on what you are working on. The background should be just enough for another student in this class to understand it. If you are doing a paper on WLANS, I do not expect or want a description of the 802.11 MAC. We know this, and a reference to the standard is more than sufficient. But, if you are using a more obscure part of the MAC that would not be understood by the class, then a description of this may be necessary. Remember, that your goal is to make sure that you and the reader agree on the question and answer. One of the best ways to show the question is significant and contestable is through a review of prior work in the literature. Such a review is a necessary part of your final report. If you have any questions on these concepts. See section 3.D in the Capstone Introduction.
  
  The introduction in a technical paper should not leave the reader in suspense. It is appropriate to have a single paragraph that summarizes the rest of the paper including your answer to the question.
• Methodology: Here is where you describe what you did (or will do) to answer the question. You should try to put the minimum of what you need to do to answer the question. Research has many false starts and dead ends. The readers are not interested in all of these tracks. They want the linear path from question to answer. If you feel you have an interesting side track to report, it goes in the discussion. As you describe your experiments, analysis techniques, etc. ask yourself at each step how this helps you to answer the question. In other words, clearly state how the experiments you do will lead you on that linear path to the conclusion.
• Results: Describe in neutral or non-committal tones any experimental, analytical, or research results that you have obtained. Again, these should be the results on the linear path to your conclusion. Graphs, tables, and figures are very important here. There should be very few graphs. If you have more than 5 graphs you probably have too many. Your goal in every graph is to tell a story. You are inviting the reader to compare two different lines or to compare one line as the x-axis varies. In order to make this comparison, the graph should be simple and uncluttered. Three dimensional graphs are almost always a bad idea. They are a red flag that the authors are sloppy, lazy, or trying to mislead the reader. Similarly: Multicolor backgrounds--bad; unlabeled axis--bad; changing scales between related figures--bad; dense grid lines--bad; and too many digits of precision--bad. Tables should also tell a story. As with graphs, simpler is better. Put the data you need and no more. Put the data in the cleanest form possible. If it is numeric measurements of some value then it is unlikely that you have 6 digits of precision. The precision of the printed data should match the precision of the actual data. Figures should again be simple and clean. Avoid cut and pastes from the web or other documents. First, it is a copyright violation and considered plagiarism. Second, a figure done yourself will be sure to tell your story and not someone else's. You are also more likely to edit a figure done yourself to better suit your purposes as you work on your paper. Third, it is usually not more than a half hour job to create your own similar figure. In graphs, tables, and figures always ask yourself if they tell the story that you want.
  
  The instructions here about editing and telling a story with the data are in no way suggesting that you should lie about the data or throw out data that does not support your conclusion. The data should be chosen and presented in a form that answers the question as clearly as possible. For example, if your question is whether RTS/CTS improves throughput in congested 802.11 BSS, then your data should allow the reader to clearly see the change in throughput with and without the RTS/CTS mechanism.
• Discussion: Here is where you describe how the results relate to the research question. Did they clearly answer the question? If not, what might have been the problem? How might you be able to answer the question better? What are the implications of the results?
• Conclusion: Here is where you should restate the research question, the answer, list highlights from the results, and present suggestions for future work.