Section: New Software and Platforms

AHMA

Automatic Malware Hardware Analysis

Keywords: Side-channel - Deep learning - Malware

Functional Description: This framework is composed of several parts, each one of them taking in charge the generation and the processing of the data at different levels. Drivers have been developed to automatically control the different oscilloscopes we are working with (picoscope 6407 et infiniium keysight). We use signal processing tools on the raw data to feed a deep neural network which is in charge of classifying the observed malwares. We are using two different approaches to manage the infection of the system. The first one is to reinitialize it each time we make a measurement to ensure its integrity. We have proposed a method allowing to speed the procedure up a lot. Besides, we developed several malwares, to make our experiments in a controlled environment, to avoid the necessity of cleaning the system up after each measurement.

• Contact: Annelie Heuser