# A Detection Mechanism Against Load-Redistribution Attacks in Smart Grids

## TL;DR

This paper introduces a real-time, physics-based detection mechanism for load-redistribution attacks in smart grids, improving detection speed and accuracy by leveraging domain knowledge and optimization algorithms.

## Contribution

It proposes a novel detection method based on power system physics and a greedy algorithm to identify sensitive attack points, enhancing existing bad data detection systems.

## Key findings

- Successfully applied to a 2383-bus Polish test system
- Achieved over 10x faster attack problem solving compared to traditional methods
- Enhanced detection capability by integrating with existing mechanisms

## Abstract

This paper presents a real-time non-probabilistic detection mechanism to detect load-redistribution (LR) attacks against energy management systems (EMSs). Prior studies have shown that certain LR attacks can bypass conventional bad data detectors (BDDs) and remain undetectable, which implies that presence of a reliable and intelligent detection mechanism to flag LR attacks, is imperative. Therefore, in this study a detection mechanism to enhance the existing BDDs is proposed based on the fundamental knowledge of the physics laws in the electric grid. A greedy algorithm, which can optimize the core LR attack problems, is presented to enable a fast mechanism to identify the most sensitive locations for critical assets. The main contribution of this detection mechanism is leveraging of power systems domain insight to identify an underlying exploitable structure for the core problem of LR attack problems, which enables the prediction of the attackers' behavior. Additional contribution includes the ability to combine this approach with other detection mechanisms to increase their likelihood of detection. The proposed approach is applied to 2383-bus Polish test system to demonstrate the scalability of the greedy algorithm, and it solved the attacker's problem more than 10x faster than a traditional linear optimization approach.