Date on Master's Thesis

8-2025

Document Type

Master's Thesis

Degree Name

M. Eng.

Department

Computer Engineering and Computer Science

Committee Chair

Baidya, Sabur

Committee Member

Lauf, Adrian

Committee Member

Aqlan, Faisal

Author's Keywords

Rowhammer, cybersecurity, DRAM, memory, robotics

Abstract

This thesis investigates how low-level memory faults can undermine edge-assisted robotic systems that rely on memory optimization. As robots are utilized in real world applications, the ability to operate safely and successfully in mission critical deployment becomes important. To help achieve these goals, developers are increasingly starting to place computation nodes at network edges to meet latency and reliability requirements. Edge nodes, however, are resource-constrained and resources conservation techniques such as Kernel Same-page Merging (KSM) are enabled to deduplicate identical pages across processes or virtual machines. This thesis shows that this optimization technique quietly widens the attack surface and can allow a malicious process using read only privileges to access shared configuration data to use memory deduplication in conjunction with RowHammer attack to corrupt a co-located victim’s state without triggering copy-on-write or modifying any file on disk. Three attacks were implemented and evaluated on a DDR3 based edge platform with a Robot Operating System (ROS) controlled multi-robot setup. (Gazebo simulation offloaded to a separate workstation). Research Task 1 showed that targeted hammering on a single robot could reliably induce bit flips (1 bitflip/min) under realistic conditions, and when aimed at a speed-control variable, caused measurable velocity changes. Task 2 demonstrated a cross-process attack in a platooning setup: identical configuration data loaded by attacker and victim was merged by KSM, and re-hammering flipped a bit in the shared page, increasing victim speed from 1 m/s to 2 m/s and disrupting formation despite no visible configuration change. Task 3 introduced a vi “decoy” variant in which the attacker restored its own value post-flip, forcing an unmerge that left the victim corrupted; this caused sustained formation errors and a collision. Overall, while KSM improves memory efficiency, it can enable undetectable cross-tenant state corruption in edge robotics. Mitigations include disabling deduplication for safety-critical pages, hardening Rowhammer defenses, and adding in-memory integrity checks to balance performance with safety.

Download

Share

COinS