[Discuss-gnuradio] GSOC '16: Signal Intelligence (gr-sigint)

[Richardson, Christopher (richarc2)]

Hi everyone,

I'm very interested in working on the Signal Intelligence (gr-sigint)
project for the Google Summer of Code.

I'm currently a PhD student at Lancaster University, UK, studying attack 
detection 
in a privacy preserving manner.

I achieved an MSc in Bristol, UK, making use of machine learning techniques to 
detect viruses - http://www.lancaster.ac.uk/pg/richarc2/dissertation.pdf. 
As mentioned in the idea suggested by Mr Rajendran "Another approach is to use 
available waterfall images and run some image comparison algorithms", 
I am curious if I could make use of such machine learning techniques to achieve 
this.

I am also especially interested in how the performance of such classifiers 
could be measured through conducting real-world experiments, 
with 2 SDRs (one for transmission and one for reception) at a range of 
increasing distances, potentially making use of
techniques such as Receiver Operating Characteristic (ROC) curves and the Area 
Under Curve (AUC) as a metric for quantifying 
the performance of a classifier.

I'm currently reading more about algorithms to detect cyclostationary features 
along with a survey on Automatic Modulation Recognition.
I'm also looking at existing GNU Radio modules such as gr-specest.

If anyone could point me at further reading material or suggestions for the 
proposal, that would be great!

Kind Regards

Christopher Richardson