Advice for new PhD students

From rs_faq
Jump to: navigation, search

"How to get a PhD" originally written by Ben Smyth

Welcome to the next chapter of your lives. This article provides several pieces of invaluable information which you will ignore and inevitably regret in future years.

First, maintain a good bibliography! This will require you to start a fresh bibtex file (or Endnote file if you are so inclined). Every time you add an entry to this file make sure that it is correct. Personally, I highly recommend DBLP as a primary source of bibtex entries, but make sure that you check the details with the original source. Correctness is essential, looking up 100+ incorrect references, from several bibtex files, when you come to write your thesis is very painful (TRUST ME!). Also, be consistent with your references. For example, I use the style:

  @inproceedings{Smyth07:DAA,
  ...
  booktitle={ESAS'07}#proc#{4th European Workshop on Security and Privacy in Ad hoc and Sensor Networks},
  series   = lncs,
  publisher= sp,
  ...
  }

That is, my booktitle field is always of the form "ABBRV'YEAR: <1st International conference on> NAME" where the following definitions appear at the top of the bibtex file

  @string{sp = "Springer"}
  @string{acm = "ACM Press"}
  @string{ieee= "IEEE Computer Society"}
  @string{els = "Elsevier"}
  @string{ios = "IOS Press"}
  @string{crc = "CRC Press"}
  @string{ifip = "International Federation for Information Processing (IFIP)"}
  @string{lncs= "LNCS"}
  @string{proc= ": "}

I originally defined @string{proc= ": Proceedings of "}, but decided this was a little OTT!

The next important point is to annotate your bibtex file. I recently discovered that you can write plaintext around entries in a bibtex file. Take advantage of this to include notes about each paper you read. Personally, I would recommend "reviewing" each paper as if you were doing so for a conference, that is, write a brief 1-2 line overview, highlight the papers strengths and comment upon the papers weaknesses. This will be an invaluable source of information when you come to the paper again and cannot remember what it was about. (Again, this can be extremely painful when you are writing your thesis!) You may also like to highlight a couple of key quotations -- e.g., "the authors believe security protocols fail because ..." -- although I suspect this will provide limited future value (you won't quote something useful, but maybe it will help jog your memory of what the paper was about). In addition, I would include a literature review within your bibtex file, for example,

The first ECC-based DAA protocol was introduced by Brickell, Chen \& Li~\cite{Chen08:ECC-DAA,Chen09:DAA-security} using symmetric pairing and was extended by Chen, Morrissey \& Smart~\cite{Chen08:ECC-DAA-assymmetric-proof,Chen08:ECC-DAA-assymmetric} to an asymmetric setting to improve efficiency; however, Li discovered a vulnerability and Chen \& Li propose a fix~\cite{Chen10}. Chen, Morrissey \& Smart~\cite{Chen09} identified further attacks against~\cite{Chen08:ECC-DAA-assymmetric-proof,Chen08:ECC-DAA-assymmetric} and also found attacks against the symmetric pairing based protocol~\cite{Chen08:ECC-DAA,Chen09:DAA-security}; they propose fixes to both schemes~\cite{Chen09}. An optimisation of the fixed asymmetric scheme~\cite{Chen08:ECC-DAA-assymmetric-proof,Chen08:ECC-DAA-assymmetric,Chen09} has been proposed by Chen, Page \& Smart~\cite{Chen10:ECC-DAA-assymmetric-optimised}. Three further ECC-based DAA protocols have been defined by Chen \& Feng~\cite{CF08}, Brickell \& Li~\cite{BL09,BL09a} and Chen~\cite{Chen10:DAA-security-non-frameability-eprint,Chen09:DAA-security-non-frameability}.


This will be hugely beneficial to you and moreover, you could even turn your bibliography into a paper! (We call those literature reviews.)

Obviously, I followed none of this advice during my PhD... Most of you won't either. For those of you that do, and benefit from it, I want an acknowledgement in all the review papers you publish ;-)

Second, you are first year students which means you have exponential amounts of free time. This will decrease significantly in your second year when you find things to work on. By your third year you'll struggle to fit in a long lunch. Then, by your fourth year (statistics suggest you'll still be here!), you will forget what evenings and weekends are... Anyway, I have sidetracked, you are first years and have loads of time -- use it wisely! In my first year I studied an open university courses in Mathematics and Logic. I wish I had also spent some time learning how to write. I recommend:

  • Eats, Shoots and Leaves
  • The Elements of Style (I own, but haven't read)
  • Mark Lee (http://www.cs.bham.ac.uk/~mgl/guides/email_etiquette.html) would recommend Fowler's Guide to Modern English
  • Simon Peyton Jones: How to write a great research paper
  • Priya Narasimhan: How To Write a Good (no, Great) PhD Dissertation

Writing well is something that I aspire to and it will certainly make your supervisor very happy.

I mentioned that you will be here for in excess of three years... And yet your funding only lasts three years (or 3.5 in some cases, including mine). You need to budget for this!

Your website is essential! If I cannot email you (I will get your email address from your web page, which I will find using Google), how do you expect to get anywhere? (Replace *I* with *Rich employer* if necessary.)

Personally I recommend setting up your own personal web page (e.g., www.bensmyth.com) because you will exist after your PhD and you want a permanent base (although setup a minimal school page too e.g., http://www.cs.bham.ac.uk/~bas/). I also recommend a permanent email which you will only ever include in your publications -- that means a reader of your papers can contact you even if you paper was written 20 years ago. Alternatively, Aaron includes his URL rather than email address on papers. Your web page should be kept up to date and should include:

  • photo
  • outline of research
  • contact details
  • all of your papers
  • etc.
Personal tools