Date of Award


Document Type

Thesis (Ph.D.)

Department or Program

Department of Computer Science

First Advisor

Sean W. Smith


Edge devices, or embedded devices installed along the periphery of a power grid SCADA network, pose a significant threat to the grid, as they give attackers a convenient entry point to access and cause damage to other essential equipment in substations and control centers. Grid defenders would like to protect these edge devices from being accessed and tampered with, but they are hindered by the grid defender's dilemma; more specifically, the range and nature of tamper events faced by the grid (particularly distributed events), the prioritization of grid availability, the high costs of improper responses, and the resource constraints of both grid networks and the defenders that run them makes prior work in the tamper and intrusion protection fields infeasible to apply. In this thesis, we give a detailed description of the grid defender's dilemma, and introduce TEDDI (Tamper Event Detection on Distributed Infrastructure), a distributed, sensor-based tamper protection system built to solve this dilemma. TEDDI's distributed architecture and use of a factor graph fusion algorithm gives grid defenders the power to detect and differentiate between tamper events, and also gives defenders the flexibility to tailor specific responses for each event. We also propose the TEDDI Generation Tool, which allows us to capture the defender's intuition about tamper events, and assists defenders in constructing a custom TEDDI system for their network. To evaluate TEDDI, we collected and constructed twelve different tamper scenarios, and show how TEDDI can detect all of these events and solve the grid defender's dilemma. In our experiments, TEDDI demonstrated an event detection accuracy level of over 99% at both the information and decision point levels, and could process a 99-node factor graph in under 233 microseconds. We also analyzed the time and resources needed to use TEDDI, and show how it requires less up-front configuration effort than current tamper protection solutions.


Originally posted in the Dartmouth College Computer Science Technical Report Series, number TR2016-804.