Wednesday, January 6, 2010

The length of the Emperor of China's nose and a few simulation stories

After getting in to IIT and in two years getting a campus placement; is not uncommon. On one side appreciations from people, asking for treats etc etc. On the other side it is a puzzle if I complete MS in time or not and you will kill your time lost in thoughts.
One funny thing facing me is that one year back I proposed a way to improve the existing wireless system. I could envisage it beautifully. I was so happy at that time that it is something unique. Now it is time for Annual Progress Report submission and I am puzzled again on what to write. I never implemented it. In fact I am not even convinced that there will be any use of such a "theoretically beautiful" protocol to be implemented.
If you are not wondering why such a situation; surely you might have done research. Anyway; let me try to explain my situation. The proposal came after a series of articles and scientific publications on the area I am working on. It is a 10 year old technology and still results are in simulation. People propose millions of ideas; simulate it and says "oh my way; look at it!!! it is awsome... it improves performance by 200%". Then another one says "oh yeah I have changed parameter x to parameter y, reason being a probabilistic modelling using ...blah ..blah . and improved results by 300%". I being stupid or what; believed such things will actually work in real hardware. Thus my proposal was based on previous results of my great predecessors.

I tried to implement my solution in real hardware. Learned a new operating system, learned to work with a system which does not even have a monitor or a display to tell me if an error occurred. I got initial results. Weired!!!! It does not even give me 10% of what it is capable of. If I implement my method; that will give another overhead and will reduce performance. Huh!!!! Now the puzzle comes back to me. Do I need to believe that someone will find it fascinating and will implement. We are researchers; we just create models.

I felt the whole research is like the story of "length of Emperor of China's nose". Nobody was permitted to see the Emperor of China, and the question was, What is the length of the Emperor of China's nose? To find out, you go all over the country asking people what they think the length of the Emperor of China's nose is, and you average it. And that would be very "accurate" because you averaged so many people. But it's no way to find anything out; when you have a very wide range of people who contribute without looking carefully at it, you don't improve your knowledge of the situation by averaging. It is not that I can also tell, mine is more accurate by adding survey with another set of people and says "oh mine is 300% more accurate ".

If you try to build the system; it is more funny. You need to average at least for 100 nodes, because predecessors did it for 10000 or even more. Purchase of 100 nodes will take its own time. Then you realize that off the shelf devices does not give any performance you wanted. Design new one. Another purchase!!! Funniest thing is that it is always a story of re-inventing the wheel to come up with a design. Once the prototype is built; which costs 1/10 or even lesser than that of imported one and perform better; you try to scale it to 200 nodes. One node will have say 4 components to be assembled, manufactured by four companies. You need to get quotes at least 3 for each. You will try to call up persons in India. If not you need to figure out how to ship from country X to India including customs clearance and taxes. Once it is done, after 2 weeks reply will come from purchase that; quote is no longer valid. Then you try for re-quote. It take another month to give Purchase Order. Each component may take their own time from 1 to 3 weeks. If you are lucky; it will stay in customs office for another month because there is additional free cd came with that consignment which was not mentioned.
Meanwhile you try to find people for making custom boards. It take a month to get and you will see that the connection is reversed. It goes for revision. Once it comes out well; you place the devices and try to do measurement; it fails for no reason. You debug the circuit and see that component x and y do not inter-operate. You go for another revision. Then integrate 200 of them and verification will take weeks. Then you need to implement your algorithms which might not have considered the models we integrated. That will take another fortnight. To deploy and test with one setup takes hours in ground. You need at least 5 people with average of 20 nodes per person. To have a common time for them to measure one setup will be once in a week. Assuming you have 3 parameters to change; you need all permutations and combinations of them to measure. Each one needs update of 200 nodes and do the measurement. It can eat up months.

And if it works; your results will be pathetic compared to supermodels in simulation. By this time you will see hardware advancement and your design is no longer valid. Now you can at least think yourself; how to find the length of Emperor china's nose.


Phani said...

Interesting points presented in an abstracted sense. Excellent :-)

Nesmel said...

the truth behind research and prototypes.. nicely put..!

renjithdwg said...

You just presented wonderfully the glimpse of the research scenario in this century..
just a briefing again:
Say one of the universally accepted equation looks like this:
Viscosity,m (Poise) = (0.00022 .t — 0.18 / t )
A person will just take it, do some experiments and come up with a lot of manipulated results and say like this: "I FOUND FROM THE EXPERIMENT THAT THE CONSTANT 0.00022 IS NOT CORRECT, IT IS 0.00023" and claims that its a great invention and publishes a paper...

Manish said...

This post ranks right up there on the top ( statistically speaking ;) ) ....

Anonymous said...

If you got your quote from Surely your joking Mr. Feynman. You should give a little credit to his version/book.

Gireesh said...

How do you say that I took it from Feynman's book?
i like feynman... i know he used this in his book. But it is not his quote, it is a general story.
So no credits to Feynman

riswan said...

@ gireesh sir,
really watever u wrote is absolutely true... its a pathetic situation here in most of the engineering colleges where v take up some other's work n just change the active voice of the paper into passive voice and submit a project. i should say true research is not encouraged.
n i recently happened to be at iit-m for exebit n had a chance to meet u der. its really awesome work all of u ppl doin der..