Evaluating a Process-Aware IDS for Smart Grids on Distributed Hardware

Abstract

Recent incidents clearly identify the need for improved (cyber) security in the power distribution grid. The communication infrastructure of a power grid (the Supervisory Control and Data Acquisition (SCADA) network) is often a lucrative target for cyber-attacks and manipulations. In a recent line of work, a process-aware approach was proposed to locally monitor the communicated data and detect anomalies and inconsistencies. Recently, that approach was extended to a neighborhood level and tested in a simulation environment. This paper takes the extended approach closer to practice and shows its feasibility on distributed hardware. We evaluate the hardware capacity, the chosen communication protocol, and the real-time capability with respect to performance on a Raspberry Pis cluster and compare it to the originally centralized test cases. Further, requirements for transforming the distributed prototype into a real-world application are discussed.