250 likes | 558 Views
Intrusion Detection System. Bruno Melo Diego da Silva Matheus Finatti Vinicio Meira. Advised by Dr. Xiang Fu. Intrusion Detection System. Monitor system processes Detect Analyze Block. Malicious Activities. System Architecture. CLUSTER. IDS. Support Vector Machine - SVM.
E N D
Intrusion Detection System Bruno Melo Diego da Silva Matheus Finatti Vinicio Meira Advised by Dr. Xiang Fu
Intrusion Detection System • Monitor system processes • Detect • Analyze • Block Malicious Activities
System Architecture • CLUSTER • IDS
Support Vector Machine - SVM • Analyze Data • Recognize Patterns • Classify Data
Classified Data SVM
SVM Interface for IDS • Interface LIBSVM • SVM • IDS
Cluster Operation Modes Training mode: $ python clustey.py --train -c <logfolder> -w <function name> Predict mode: $ python cluster.py --judge -r <modelfile> * Test data is in the environment variable called “request”
Main Module - Java™ Program • Generate C++ Wrappers • LD_PRELOAD • LD_LIBRARY_PATH • dlsym() • Intercept and log Apache library calls • Monitor Apache library calls • How to generate wrappers?
IDS Operation Modes - Train • Parse Configuration File • Generate and compile wrapper • Start Apache • Intercept calls and generate log files
IDS Operation Modes – Complete Train • Stop Apache • Send log files to cluster’s training mode
IDS Operation Modes – Monitor • Parse configuration file • Generate and compile wrapper • Start listening server to communicate with wrapper • Start Apache • Intercept calls and send to listening server • Send response to C++ wrapper • Send log entry to cluster to analyze • If rejected, ask user if Apache should be killed
Using IDS Modes • Training mode: • # java –jar ids.jar -c <configfile> -o <outfile> -mode train [-v|-i] • Complete train mode: • # java –jar ids.jar –p <logpath> -mode completetrain • Monitor mode: • # java –jar ids.jar -c <configfile> -o <outfile> -mode monitor