Skip to content

Latest commit

 

History

History
33 lines (26 loc) · 1.64 KB

README.md

File metadata and controls

33 lines (26 loc) · 1.64 KB
Raw

PFSM building and PFSM-based deviation analysis

Requirement: install Synoptic for FSM construction

Ivan Beschastnikh, Yuriy Brun, Sigurd Schneider, Michael Sloan, and Michael D Ernst. 2011. Leveraging existing instrumentation to automatically infer invariantconstrained models. In Proceedings of the 19th ACM SIGSOFT symposium and the 13th European conference on Foundations of software engineering. 267–277.

The scripts are tested on Linux version 5.4.0-169-generic (Ubuntu 9.4.0-1ubuntu1~20.04.2) with python 3.7.6.

pip install networkx pydot numpy scipy matplotlib

Event traces

  • input: traces/log_routines_xx/
  • log processor: convertor_new.py: generate log files for Synoptic - uncontrolled dataset
  • convertor_trace_dataset.py: generate synthetic (for evaluting deviation scores) datasets and 5-fold training and testing sets

Build PFSM

bash synoptic.sh -o output/pfsm --dumpInvariants=True -r '(?<TYPE>.+),(?<DTIME>.+)$' -s '^------$' --outputCountLabels=True --outputProbLabels=True --outputSupportCount=True --ignoreNFbyInvs=True --supportCountThreshold=2 logs/trace_may1

Invariants: AlwaysFollowedBy and AlwaysPrecedes

Deviation measurement and detection

python3 state_machine_builder/state_machine_read.py output/pfsm traces/trace_may1
python3 state_machine_builder/state_machine_read.py output/pfsm traces/trace_train_may1
python3 state_machine_builder/state_machine_read.py output/pfsmtraces/trace_test_may1

output/pfsm: the generated FSM file from Synoptic

Others

  • synthetic_fn_analysis.py: edit event traces on routine dataset (add or alter)
  • train_test_split.py: convert routine logs to training and testing sets