Information for prospective PHD Students

 

How to pursue a PhD at the Distributed Systems Group

At first, make sure you want a PhD. If that is not your driving force, you will probably fail.
Having said that, there are several criteria for a good PhD thesis along with several tips how to pursue your research.

Criteria for a good PhD thesis

A PhD thesis is scientific work. Hence, you have to work hard to reach a scientific aim. In order to qualify for a scientific work, the main contribution has to fulfil the following criteria:

  • Motivation: You have to motivate the problem(s) you want to address in your PhD thesis. Explain what exactly is the problem (or set of problems) you want to address and why exactly you see them as a research challenge. Finding the right scope of the research problem you want to address is difficult and in many cases the effort needed to identify it, is underestimated.
  • Originality: You have to provide something novel. E.g., a new model, concept, architecture, algorithm or method. It can also be a new or alternate solution to a well-known problem, or the identification, specification, and solution of a new problem. Note, you contribution should be better in some way, not just different.
  • Relevance: Your work has to be general enough to be of interest for the broader audience of your research community.
  • Related Work: Identify how your approach and/or results differ from past and current related work. Be specific and do not simply list papers or books published, which deal with a similar and the same topic you address. Specify exactly, how your contribution differs. You really need to master the related work in the area you are pursuing the PhD in.
  • Proof and Evaluation: You have to prove and evaluate your work in a reproducible way. Among others, this can be a formal/mathematical proof (manual or automated) or an empirical proof. The latter can be based on systematic simulation, prototype implementation and evaluation, as well as well-defined experiments and systematic measurement of a real-life system. Combinations are also possible and often reasonable. If you produce software artifacts, making them available as open source further adds to reproducibility of your approach. Note that both the results and the method should be reproducible.

Tips on how to pursue your research

Research work is reading, thinking, trying, and writing - in any order and iteratively. Keep close contact to your advisor, and follow the steps below:

  1. Find a relevant problem. This can be from reading future work sections of papers or from open questions and tasks in ongoing research projects. Sometimes your advisor might have an idea. This is the basis for the motivation of your research.
  2. If needed, generalize the question to make it relevant for the broader audience, but keep it concise enough not to try to do it all. This should result in your research question ("the thesis"). Although not cast in stone, it serves as a guiding star through the darkness of complications and frustration which will accompany your research in some phases. If needed, refine the research question later on in accordance with your advisor.
  3. Relate the question to the state of the art. That involves a lot of literature research and reading of past and currently ongoing related work. Make short notes, you'll need them for the related work sections in your publications. During this process, you will identify "your" research community(ies). At the end, you should be able to describe, what's currently going on in your field of research. Relate your work to other disciplines as well, if applicable. Give a short presentation on that to your colleagues in the PhD seminar.
  4. Pursue your ideas (if you have none by now, you are in trouble) and try something out. Simulate or implement your ideas. Take care not to implement too much – your implementation is needed for evaluation purpose and as proof-of-concept mainly. Reflect on your assumptions, your methods, and your results in a critical way and refine your assumptions and methods if necessary.
  5. Write down your ideas and results in a research paper. Be honest and clear, don't promise too much, but show what you have achieved and mention the important points (problem, assumptions, contribution, novelty, relevance, results, proof) explicitly several times throughout your paper. Don't assume the reviewer likes your paper - assume he or she hates it and will make it as hard as possible to reject your work. Give a presentation to your colleagues in the PhD seminar before your international conference presentations.
  6. Repeat steps 4 and 5 thereby refining steps 2 and 3, until you have sufficiently and comprehensively answered your research question. As an ultimate goal, you should become an expert in your specific research field, well known to the few groups internationally that pursue similar research.
  7. Participate in reviewing and organising of workshops and conferences. You'll get to know the other side of the medal, which will help you, improve your own work. And after all, its service to the community you are part of.
  8. Write the thesis based on your most important research papers - you should typically have several high-quality papers published at that point in time. Refine the related work by double-checking complementary or concurrent ongoing work once more - by now you should know the most relevant people and research groups in your field anyway. Adhere to the formal guidelines for the thesis.
  9. Plan enough time for your advisor to read your thesis (four to six weeks is a good guess) and for yourself to revise it.

For a general overview on what question you should be able to answer/work out before seriously pursuing your research see also this presentation by Christoph Dorn