Date of Award


Degree Name

Master of Science


Computer Science


Deepak Tosh


With the introduction of IoT into ICS and smartgrid environments there has been a mod-ernization of communication protocols through the internet. This has led to the use of features such as TCP/IP but with it comes modernized attack vectors against these sys- tems. These attacks can be Man In the Middle (MITM), rogue device communication and device cloning. To prevent these attacks, this thesis deploys Radio Frequency Fingerprint- ing (RFF) techniques to verify the uniqueness and legitimacy of known devices. It is crucial to employ security measures within ICS that do not add to the network complexity as this effects the availability of critical resources. RFF aims to solve this by establishing itself a physical layer authentication method. It does not add network complexity as it focuses on the analysis of existing wireless transmissions amongst devices in the ICS network. RFF has improved significantly through Convolutional Neural Networks (CNN) and this the- sis presents a case study on the feasibility of deploying these new techniques on Remote Terminal Unit (RTU) devices. It has been found that a RFF CNN model can run alongside the normal duties of an RTU. Directly this thesis shows that the increased responsibility is possible on low end devices with a 64-bit architecture, which means that devices like the SIMANTIC S7-1500 controller can utilize RFF in the field. The trained accuracy of the CNN has a detection rate of 84% when handling the dataset gathered in this thesis. This is a promising result given the fact the computer intensive RFF mechanism is being executed on a resource constrained environment like a RTU.




Received from ProQuest

File Size

41 p.

File Format


Rights Holder

Evan White