---------------
Referred Paper
---------------
Monitoring SIP traffic using support vector machines(Mohamed Nassar, Radu State, and Olivier Festor).

-------------
TCPDUMP
-------------
Input = tcpdump data placed in dumpfiles. eg:dumpfiles/try1

Perl interface for LIBPCAP library used :CPAN
call-flow-features.pl

Output : 38 features stored in temps.txt

command : perl call-flow-features.pl>temps.txt

---------
Dataset
---------
Replace tcpdump input with the dataset files (eg: asterisk)

---------------------------------
Converting temps.txt to SVM input
----------------------------------
Use C program convert.c to convert temps.txt features to SVM input.
Input: temps.txt
Output: svm.txt
Command:gcc convert.c

-----
SVM
-----
Input stored in svm_input folder for each of the dataset files.
Combined training & predict files stored in files named "train" and "predict"

LibSVM 3.1 is used.
Stored in libsvm-3.1 folder. (model file for training set also stored there)

Commands :
./svm-train -c 1000 /home/sarang/libpcap/train
./svm-predict /home/sarang/libpcap/predict ./train.model /home/sarang/libpcap/out

Output file stored in file named "out".