Exjobb project plan guidelines

About 2 weeks after you started your exjobb project, either at IDA or some industrial company, you will have to hand in a detailed project plan (thesis proposal) first to your supervisor, and after his/her approval, to your examiner (in this case, me).
This plan should, on 5-10 pages, contain the following:

• A formulation of the problem to be solved.

• A list of requirements that your solution has to fulfil.
  (Example: "do the operation ABC at least twice as fast, using the new UVW feature of the XY-2 processor").
  Assign priorities to these requirements.

• Sketch of an initial approach how you intend to attack the (main) problem.
  (Example: "by simulated annealing").
  Alternative approaches, if known.

• Sketch of an evaluation method how the requirements should be verified.
  (Example: "we will use the XY benchmark and compare to the figures in Z's article.")

• A schedule for the remainder of the project.
  Identify tasks, milestones, assign time resources.
  Identify potentially problematic dependences from others' work and from the availability of special resources that may be critical for your project.

• The result of your literature analysis:
  A survey of related work, about 2-3 pages with references. See also my writing guide for exjobb theses.

• A prospective plan for your thesis (main structure / table of contents)

The time for writing this document is well spent:

• The document becomes a kind of contract between your advisor (handledare), the examiner, and yourself on the goals and remaining tasks in the project and their prioritization.

• The document contains a detailed schedule for the remainder of your project. Hence, you can check whether you are still on schedule, and avoid delays. If you get stuck at some point, discuss with your advisor how to continue.

• Industry projects sometimes tend to be a moving target. The plan documents a fixed set of requirements that you can use to defend yourself against changing requirements. If some basic fact changes, such as a different hardware platform etc., inform your examinator about this.

• Dependences from other projects or availability of certain systems for experimentation and measurements can become problematic. Do a risk analysis and identify fall-back methods for various possible cases of failure.

• The text describing the result of your initial literature recherche can finally be integrated in the related work section of your final thesis.

• In the case that there should be major problems or errors in the plan, you will get feedback from the examinator. This can help you to avoid wrong directions or late fixing of problems in the final thesis, which would cause you a lot of unnecessary work.

Some further advice:

• Do a comprehensive literature recherche very early in the project, and continue searching and reading relevant papers while the project progresses. This time is well spent.

• Do not trust lecture material found somewhere in the internet, even if it is fast and convenient to look up definitions with search engines. Take a look into your old course books instead, and visit the library from time to time. Ask your advisor and your examinator which books you should read to prepare for the project.

• Experimental evaluation is important. Usually, time gets short at the end of an exjobb project, and therefore experimental evaluation usually gets very thin; the result is then not really meaningful, which is a pity if you consider all the other efforts that you put into the project. Hence, plan enough time for evaluation right from the beginning. If you do your project in an industrial company, make sure that you will have enough opportunities and resources for evaluation.

• For extraordinarily good master theses with meaningful results I usually recommend and support writing a paper and submitting it to some conference.

• Stay in contact with your examinator. Every few weeks, send a short e-mail message with a short status report. If you encounter fundamental problems e.g. with your chosen approach where your advisor cannot give you sufficient help, you should ask the examinator.