Handling the positional and chip-to-chip variations of side-channel monitoring

Abstract

Side-channel monitoring has the potential to offer a non-intrusive way to assess the behavior of an embedded device by monitoring physical signals such as power consumption or electromagnetic emissions by placing a small measuring probe close to the device. Although studies seem to indicate that it is effective in a laboratory environment with monitoring methods adjusted based on the specific target chip, the reliability of it in real world environments remains unclear due to several challenges such as imprecise probe placement and device variability.

This thesis aims to investigate the feasibility of using electromagnetic side-channel monitoring under these conditions in order to investigate the potential of autonomous monitoring. Using the Chipwhisperer Husky platform, Arduino targets, and a robotic arm to emulate an autonomous device, electromagnetic traces were collected from multiple chips and positions. Convolutional Neural Networks were then trained and evaluated on this data to determine which software was executing.

The results show that models trained on specific chips can attain some accuracy, but fail to obtain reliable accuracy again unseen chips, even when using data from multiple devices for training. However positional variation in training data improved the accuracy to random probe placement.

The findings suggest that side-channel monitoring is possible even with lack of accuracy in position, the variation between differing chips pose a large obstacle. Practical uses of the technology might therefore require chip-specific training data with positional variation.

Popular Abstract

“output” — 2026/1/26 — 13:48 — page i — #3
Popular Science Summary
The modern society is dependent on embedded systems, being used in most parts
of society, in everything from industrial plants, vehicles, critical infrastructure to
healthcare. Therefore, the security and ability of these systems to perform correct
operations is vital, and therefore it is important in many cases to properly monitor
that they function correctly. But adding direct software and hardware monitoring
on these devices can often be impractical, expensive, and a security risk, especially
on already established systems. An alternative approach is to monitor the systems
by observing the small physical signals they emit while running, such as the changes
in the electromagnetic field that one can observe from them. This is known as side-
channel monitoring.
In laboratory settings, side-channel monitoring has been shown to work, but
the question remains of how well they can function in real-life conditions. The
probes used to collect the electromagnetic readings can be difficult to place pre-
cisely, the environment can introduce additional noise, and different chips of the
same model can behave differently.
In this thesis we aim to investigate how some of these real-life conditions affect
the accuracy of monitoring through electromagnetic side-channels. By using a
robotic arm to simulate inaccuracies in probe placement and by collecting data
from several chips, we trained machine learning models to identify which software
these chips were running. We intend to compare how differences in position and
which chips is used for training affect the accuracy of such models.
The finding seems to suggest that there are many difficulties introduced by
variations in position and target chip and that side-channel monitoring might be
challenging if one cannot create a specialized model for a specific chip.

• author: Nilsson, Fredrick
• supervisor: Paul Stankovski
• organization: Department of Electrical and Information Technology
• course: EITM01 20242
• year: 2026
• type: H2 - Master's Degree (Two Years)
• subject: Technology and Engineering
• keywords: Side-channel, Electromagnetic, Security
• report number: LU/LTH-EIT 2026-1109
• language: English
• id: 9221038
• date added to LUP: 2026-01-28 13:47:24
• date last changed: 2026-01-28 13:47:24