Engineering postgrad.

Tar.Aldarion

Hello. Just a few questions to those of you that may be in the know.
I am interested in applying for a phd in dcu, have just gotten a degree in Electronic and Computer Engineering, this one interests me but I would like to learn more about it if any of you know:
http://www.dcu.ie/engineering_and_computing/research/PG/Optical210408.shtml

Here is what I have learned about it so far, if this is accurate:

The project focuses on a new type of optical network called an Optical Burst Switched (OBS) network which can give a very high bandwidth transport over optical fiber compared to current technologies. These OBS networks have not yet been deployed as there are a number of outstanding questions regarding how exactly they should be designed. One such question relates to what capabilities each switch in the network should have so that the whole network performs as best it can. The main reason for potential performance problems in OBS is when multiple packets arrive at a switch at the same time and can't all be routed out of an output port at the same time - some must be dropped. There are extra capabilities that can be added to a switch to allow it to resolve this type of contention, however, the hardware adds expense so there's a trade off between expense and performance. Furthermore, because traffic passes through multiple switches from source to destination nodes in a random fashion, the relationship between the resources needed at each switch becomes complicated and needs some mathematics to be applied in order to get the best (lowest cost / highest performance) solution. Because the traffic is random (packets arrive to switches at random times) we need to describe and understand the nature of the randomness and how it effects the performance at the switch, so probability theory is needed here. Once we have equations that desribe how switches will perform under certain types of traffic we can then describe mathematically how switches will effect each other and try to get the best placement of switching resources across the network. This is a 'mathematical optimisation problem' which relates to a whole branch of maths itself. Furthermore the problem is non-linear so is quite difficult to solve and normally special approximations (e.g. using genetic algorithms) need to be used to find solutions. In summary, the maths that would be involed would be probability theory, queuing theory, markov chain analysis, non-linear optimisation. These are not topics normally taught at undergraduate level (except possibly some probability and queuing theory) and so a PhD student would study these as needed during the studentship. Regarding software, part of the project would involve simulation to verify that the mathematical models are correct and we use a package called OPNET for this. Most of the work in OPNET involves C or C++ code. There are a number of software libraries available for solving queuing mathematical models and optimisation problems, most of which are written in C or C++. So, the software tools required for the project are quite specialised and again would be learned as needed as the project progresses.

Just interested to see what I would be mainly doing and if the above is correct.



I would also be interested in these:
http://dgrayweb.computing.dcu.ie/studentships.jsp
Although they say they want a strong background in the following:

1. The first studentship will address the correctness of the cryptography used in the IBS schemes. Applicants should have a strong mathematical or computer science background and ideally, have experience of probability theory, complexity theory or provable cryptography.
2. The second studentship will address the correctness of protocols associated with the IBS scheme. Applicants should have a strong computer science or software engineering background and ideally, have experience in formal methods, process algebras, model checking, cryptography or security protocols.

I'm not sure if any Engineering degree would give anybody a background in these! Any information about these phds would be much appreciated, thanks.

Shiny

I actually had OBS on my list of projects I wanted to do in DCU.

I did electronic engineering myself and I can only assume that
your Electronic and Computer Engineering course was pretty
similar.

You will have the background for it but it is pretty tough.
While I was working on my project I often saw the guy who
chose OBS, working on his project and the majority of it
seemed fine but he was pissed off with small annoying errors
in his code that only the supervisor could identify.

Tar.Aldarion

Cheers, and what one did you do in the end?

Shiny

I did "Ultra Wide Band Radio over Optical Fibre" in the end.

Basically UWB is used in the upcoming Wireless USB and Bluetooth3.0 standards.

Tar.Aldarion

Any fun?
I had a choice of uwb or rfid for my fourth year project, chose rfid.

Any suggestions as to which of the current phds sound good? hehe

Shiny

I usually pick my project based on how interesting it is
and who the supervisor is.

If you are doing a PhD, you don't want to get treated like
**** for 3 years.


...Oh and it was fun, absolutely loved the project and found it
more and more interesting as I did the research. Got a 1.1 in it too