A Multi-Agent Adaptive Architecture for Smart-Grid-Intrusion Detection and Prevention

Abstract

The present paper deals with selected aspects of energy prosumers’ security needs. The analysis reported aim to illustrate the concept of the implementation of intrusion-detection systems (IDS)/intrusion-prevention systems (IPS), as supporting agent systems for smart grids. The contribu-tion proposes the architecture of an agent system aimed at collecting, processing, monitoring, and possibly reacting to changes in the smart grid. Furthermore, an algorithm is proposed to support the construction of a smart-grid-operating profile, based on a set of parameters describing the devices. Its application is presented in the example of data collected from the network, indicating the process of building a device-operation profile and a possible mechanism for detecting its changes. The proposed algorithm for building the operating profile of devices in the smart grid, based on the mechanism of continuous learning by the system, allows for detecting network malfunctions not only in terms of individual events but also regarding limits of the scope of system alerts, by determining the typical behavior of devices in the smart grid. The paper gives recommendations to a software-agent system development, which is dedicated to detecting and preventing anomalies in smart grids.