Evolving Attackers against Wireless Sensor Networks

Abstract

Recent technological improvements in wireless communication and electronics have enabled the development of small, low-cost wireless sensor nodes, capable of monitoring everything from human health to the performance of the electricity grid. A natural consequence is a desire to secure systems containing these nodes. Unfortunately, proving that systems are secure is beyond the current state of the art, and testing for security is problematic: test cases often miss attacks that have never previously been seen. In this paper, we use Genetic Programming (GP) to create attacks against Internet of Things devices, to help identify vulnerabilities before systems are attacked for real. To assess the effectiveness of each attacker, we used it against a wireless sensor network (WSN) with publish-subscribe communications, protected by a literature artificial immune intrusion detection system (IDS). The GP attackers succeeded in suppressing significantly more legitimate messages than a hand-coded attacker, whilst decreasing the likelihood of detection. As a consequence, it was possible to tune the IDS, improving its performance. Whilst these results are preliminary, they demonstrate GP holds significant potential for improving the protection of systems with large attack spaces.