Noses and how we detect fragrances is actually done on an atomic level, basically by testing the reverberation of the bonds between molecules. So when two things smell the same they actually have the same bond strength, but can be completely different chemically. This is why creating artificial scents is so hit or miss.

The science and discovery of all this is really fascinating, I recommend The Secret of Scent by Turin: http://www.amazon.com/Secret-Scent-Adventures-Perfume-Scienc...

Noses are actually alarmingly complicated organs, and are also very poorly studied compared to other sensory organs. Progress on electronic noses has been slow to date, and existing devices tend to be fragile and expensive.

http://en.wikipedia.org/wiki/Electronic_nose

The Syft (1) Mass Spectrometer - the Voice 200 (2) - can sniff air and determine minute amounts of compounds present in real time, sending the output in standard form to a computer. It's the size of a big washing machine, and a lot heavier, and is portable only in vans. One application is food and flavour detection.(3) Others are sea container contamination testing (a very real problem - 20% of so are toxic when opened), ambient air testing for pollutants, medical breath testing and so on.

(1): http://www.syft.com (2): http://www.syft.com/products (3): http://www.syft.com/solutions/food-and-flavor-analysis

(disclosure - I have a very small investment in the company)

A gas chromatograph? I'd imagine it would need a decent database to compare to as well, and it certainly wouldn't be too trivial a task.