Skip to content

Latest commit

 

History

History
41 lines (31 loc) · 2.79 KB

README.md

File metadata and controls

41 lines (31 loc) · 2.79 KB

You can access the paper on Arxiv

There are two main directories.

measurement_code

  • commands.txt : Contains the command to get gcov running and simulating noide
  • compile.sh : run it to compile the neccessary programs
  • generic_query.c : you can run ./generic_query "db_name" "SQL command" to run the query on your database
  • parse_traces.m, parse_traces_nonuniform_simulator.m : The matlab code to parse the traces and extract the volume found in each trace, the "simulator.m" is for simulating the non-uniform query.
  • range_query.c : The main code to run range queries, make sure you edit line 109 to have the correct name of your table in the db file. To simulate the non-uniform query, first gather large amount of traces (in our case 200 traces per query) and then simulate your query using the parse_traces_nonuniform_simulator.m file. To do this comment line 102 and uncomment line 103 with desired number in line 103.
  • run_fr.sh, run_query.sh : in one terminal run run_query.sh and in the other terminal run run_fr.sh. This will simulate running range queries on one terminal and running flush and reload on the other terminal.
  • db/ : Put your test database here
  • Mastik../ : The codes for running the Flush and Reload based on Mastik library
  • mat_files/ : Once you run the .m files the traces inside Traces/Test will be parsed and the result of mat workspace will be saved here. If you have Traces/Test1 Traces/Test2 each containing different experiment there will be two workspace saved there.
  • Traces/ : The traces gathered using run_fr.sh will be saved in Traces/Test

attack_code

Make sure you have networkx package installed. The main program can be run as follows.

python main.py MODE NOISE_BUDGET DROP_NUM SEED
  • MODE can be 'clique' if you just want to run noisy clique finding or else it can be anything else
  • NOISE_BUDGET is the value of noise budget for example 0.002 if you want to run the clique finding with noise budget of 0.002
  • DROP_NUM and SEED are for future experiments and DROP_NUM has to be set to 0 for now

The following run noisy clique finding with noise budget of 0.002

python main.py 'clique' 0.002 0 0

The following run Match & Extend with noise budget of 0.002

python main.py 'me' 0.002 0 0

To run the experiment for random dropping volumes run drop_test.sh. The files run 10 different experiments for each case where 1, 2, 4, 6 and 8 volumes are dropped. To check the scenario where you want to drop high volumes go to file main_drop.py and uncomment line 575 to 577 and on line 575 set VOLUME_UPPER_BOUND*0.7 if you want to drop all volumes greater than 70% of the total size of database.

  • CVP.ipynb : The code to run closest vector problem to reduce the noise in some of the experiments.