538B Distributed Systems: Project proposal instructions

Winter 2019

Your final project report should be no longer than 10 pages (not including references). The 10 pages should include the title page and any appendixes. I prefer that you style your paper in ACM latex format.

Detailed instructions.

• Your report should have all of the essential sections that a research paper has: title/authors/abstract/introduction/methodology-evaluation/related work/conclusion/references.
• The report may have other, optional sections, that we've seen in research papers: motivation/design/implications/discussion/future work/etc
• The report must properly motivate the research problem you have attempted to answer. Spend at least 1-2 pages motivating the problem (just look at any of the paper we have read this term). You should do this in the introduction, though many papers also have a separate "motivation" section or similar to motivate/introduce the problem in more detail.
• The report must motivate/introduce a problem that you actually solve. Make sure you are not motivating a different problem than what you solved (a common mistake). It helps to connect the problem to a broader context from the very start (see last bullet below).
• You must somehow argue that you have solved the problem. There are many ways to do this. If your problem was algorithmic/theoretical then you may have a proof. If your problem was empirical (e.g., what fraction of traffic on campus is due to BitTorrent) then you should have measurement study/results. If your problem requires a new system, then you should have built a system and demonstrated that it is the right system. Etc.
• Any empirical evaluation results must have a proper methodology to introduce the results. What was the goal of the evaluation? How did you translate the research question into evaluation questions -- why did you measure what you measured? How did you select the subject programs/networks/people/devices/etc and why? Typically, the more information you provide to describe the experiments, the better. But, it requires careful judgment to report just the important details.
• You must have some related work. The project for this class does not need to be a novel research project. But, it helps to position your work in the context of existing research. What is the closest work to your project? How is your work different? I don't expect you to do an exhaustive literature survey, but I would like to see at least half a page of related work discussion. Aim for maximum a page of related work.
• I recommend that you find 1-2 papers that we read that tackle a similar problem or use similar methodology. Then, consider how these papers structured the motivation and the presentation of the solution/argument. Don't hesitate to adopt/copy their structure/form.
• A good project report usually, at some point, takes a step back from the low-level details of the work and considers the broader context. It would be great if you can do this in your work. Show me that you can think more broadly, beyond your project and that you can identify implications of your work for other systems technologies/contexts/problems. Think of it this way: if the reader is generally interested in distributed systems and not in the concrete problem that you solved, then what can they take away from your work? Typically this kind of discussion starts in the introduction, may appear in brief snippets throughout the paper (to connect the low-level details to a larger focus), and then picks up again in the discussion or the implications section.