Enabling Moving Target Defense for Real-Time CPS Security

The Internet of Things (IoT) phenomenon has rather quickly enveloped modern society leading to regular interactions with several IoT devices and networks. Due to the exponential increase in the utilization of such devices, coupled with inherent limitations such as resource constraints and outdated firmware these devices are vulnerable and may be regularly subject to various threats. Most such systems lack any and all forms of encryption due to cost and resource constraints, as well as the general perception that such devices pose a low risk for attacks. Additionally, IoT networks are also present in several Cyber-Physical System (CPS) based applications such as water treatment facilities, where an entity can exploit these vulnerabilities to cause an attack with far-reaching consequences. With that in mind, a Moving Target Defense (MTD) is an emerging tactic to help shore up the defenses of such networks. This is done by regularly updating system parameters with the intention of putting up barriers for any malicious individual attempting to breach the network. This work deals with an attack scenario, wherein an attacker aims to take control of the sensors in a wastewater treatment facility, with the intention of deceiving the operators of said facility. An attack of this manner has the ability to tamper with the chemical composition of the treated wastewater which can end up devastating the ecosystems that this water gets released back into. This paper shows how an attacker can perform a Man-in-the-Middle (MiTM) attack using ARP (Address Resolution Protocol) spoofing and subsequently launch a Denial of Service (DoS) attack on the sensor network and disrupt the legitimate data stream. Based on this attack, an MTD approach is proposed, which is based on the routine updating of the IP and MAC addresses of the sensors with the aim to obfuscate as much of the network from the attacker as possible. The findings show that this method significantly improves the network’s ability to resist ARP spoofing and allows it to gracefully recover from a DoS attack. Additionally, this work experimentally showcases the overhead these techniques may impose. This would allow for system designers to provision for the resources needed without affecting the timeliness of time-sensitive systems, i.e., Real-Time Systems (RTS).