Metior

This is the official demonstration code for Metior, a comprehensive model to evaluate obfuscating side-channel defense schemes.

This repository contains the code required to reproduce Figure 7 of Case Study I. Specifically, it contains code that:

1. Models varying secure cache architectures under different configurations.
2. Performs Monte Carlo simulations using these models to derive the probabilistic observations under a cache-occupancy attack scenario.
3. Derives the Maximal Leakage for each attack strategy considered, plotting the space sweep using matplotlib.

Setup

Due to the large number of experiments required, we suggest running this demonstration on a many-core (48+) server.

After cloning this repository, install the requisite Python 3 dependencies by running pip3 install -r requirements.txt.

Replicating Figure 7

To replicate Figure 7, first run python3 runSweep.py, which will use all cores on the machine to run the Monte-Carlo simulation. This will write the results to results.pkl, which can be then plotted by running python3 plotResults.py results.pkl.

Code Organization

A brief overview of the source code included in this repository (found in src/) is as follows:

• primeProbe.py: Implements the routines required for the attacker to prime and probe the victim cache during a cache occupancy attack.
• batch_primeProbe.py: Organizes arguments, then calls the Prime+Probe cache occupancy attack routines.
• cacheObj.py: Defines default parameters for the cache architecture being studied (including number of cache lines, line size, etc.).
• dag.py/dagGen.py: Internal DAG representations for the attacker/victim's access patterns.
• calculateLeakage.py*: Computes the Maximal Leakage for an attacker's observations. The input is a dictionary, where the keys are the victim modulation pattern, and the values are lists of attacker observations.
• caches/: Includes models for varying secure cache architectures.