Guest Editorial: Privacy and Security in Smart Grids

A smart grid is an advanced electricity network that tightly couples cyber and physical aspects through advanced monitoring, information processing and bi-directional communication technologies. While advanced monitoring and communication technologies make the grid ‘smarter’, they also introduce new vulnerabilities. Smart grids are complex and extensive critical infrastructures and they are under constant threat from a wide range of attacks, including data theft, false data injection, denial of service attacks, malware attacks, energy theft, as well as coordinated and distributed versions of these attacks. Successful attacks on smart grids can cause significant economic and social damage through power outages, damage to infrastructure, energy theft and waste. Therefore, it is essential to detect physical as well as cyber attacks before they result in cascading failures in the network. Privacy is another growing concern for the future of smart grids. The smart aspect of the grid requires near real time monitoring of the grid state, including consumer demand and response behaviour through smart meters, to increase the resilience and reliability of the grid. However, with advances in machine learning and data mining technologies, such fine-grained information about user activity can easily provide sensitive information about individual consumers, threatening their privacy. Consumer privacy must be protected not only against third-party attackers, but also against the legitimate energy providers and grid operators, making it a highly difficult challenge. This Special Section brings together papers that study security and privacy issues in smart grids with the goal of developing information technologies that can provide the necessary security and reliability to the grid without sacrificing consumer privacy. with the nature and privacy in bring papers from range of research areas including power systems, machine statistical estimation and optimisation.

A smart grid is an advanced electricity network that tightly couples cyber and physical aspects through advanced monitoring, information processing and bi-directional communication technologies. While advanced monitoring and communication technologies make the grid 'smarter', they also introduce new vulnerabilities. Smart grids are complex and extensive critical infrastructures and they are under constant threat from a wide range of attacks, including data theft, false data injection, denial of service attacks, malware attacks, energy theft, as well as coordinated and distributed versions of these attacks. Successful attacks on smart grids can cause significant economic and social damage through power outages, damage to infrastructure, energy theft and waste. Therefore, it is essential to detect physical as well as cyber attacks before they result in cascading failures in the network. Privacy is another growing concern for the future of smart grids. The smart aspect of the grid requires near real time monitoring of the grid state, including consumer demand and response behaviour through smart meters, to increase the resilience and reliability of the grid. However, with advances in machine learning and data mining technologies, such fine-grained information about user activity can easily provide sensitive information about individual consumers, threatening their privacy. Consumer privacy must be protected not only against third-party attackers, but also against the legitimate energy providers and grid operators, making it a highly difficult challenge.
This Special Section brings together papers that study security and privacy issues in smart grids with the goal of developing information technologies that can provide the necessary security and reliability to the grid without sacrificing consumer privacy. Aligned with the interdisciplinary nature of security and privacy challenges in smart grids, we bring together papers from a wide range of research areas including power systems, machine learning, statistical estimation and optimisation.

Papers in this Special Section
In 'Detecting load redistribution attacks via support vector models', Chu et al. investigate an emerging and important problem in smart grids, the detection of cyberattacks. While traditional model-based detection techniques have been proven inadequate in identifying stealthy type attacks, the authors propose a machinelearning based detection framework by combining a support vector regression load predictor with a support vector machine attack detector, which is demonstrated to be effective in detecting both random and intelligently designed load redistribution attacks. Along with other recent publications, this paper highlights the opportunity of applying machine learning-based methods for cybersecurity enhancement in smart grids.
In the paper 'Modern power system reliability assessment with cyber-intrusion on heat pump systems', Gunduz and Jayaweera develop a novel framework to extend traditional physical-only power system reliability assessment to explicitly consider the growing interactions between cyber and physical networks. Case studies demonstrate that the cyber-intrusion, detection, and recovery processes can significantly affect the overall system reliability, which highlights the need to include cyber-physical interactions in the quantitative assessment of system reliability in smart grid.
Another paper dealing with the limitations of classical physicsbased data detection methods in detecting cyber attacks on smart grids is 'Ensemble CorrDet with adaptive statistics for bad data detection' by Nagaraj et al. Existing data-driven methods are based on learning the statistics of a 'normal' grid operation to detect anomalies. However, these statistics continuously change in smart grids with the variations of loads and generation. To tackle these variations, they propose an adaptive data-driven anomaly detection method, by using a IEEE118-bus system. Their experimental results show that the proposed adaptive technique outperforms the current state of the art for bad data detection.
In 'Survey of machine learning methods for detecting false data injection attacks in power systems', Sayghe et al. focus on false data injection attacks that can compromise state estimation in smart grid monitoring systems, and provide a comprehensive overview of recent machine learning based methods for detecting false data injection attacks. The paper provides background on smart grid state estimation, common false data injection attack, and their potential impacts on power systems. The authors also identify limitations of the existing machine learning techniques, and provide future research directions to remedy these limitations.
In 'Privacy-cost trade-offs in smart electricity metering systems', Giaconi et al. consider exploiting an energy storage device for the dual purposes of energy cost optimisation and privacy by demand shaping. While the goal in cost optimisation is to shift the demand to off-peak periods with lower energy cost, privacy is achieved by obtaining a fixed demand profile that is as independent as possible from the real consumption profile. An analytical optimisation framework is introduced to solve for the optimal energy management policy, and numerical results are presented characterising the optimal cost-privacy trade-off under various assumptions on the available information about the future energy demand profile.
This Special Section also includes a comprehensive survey paper on smart meter privacy by Farokhi titled 'Review of results on smart-meter privacy by data manipulation, demand shaping, and load scheduling'. In this survey paper, Farokhi classifies mechanisms to ensure smart-meter privacy into three categories: data manipulation, demand shaping, and load scheduling. The first category refers to methods that focus on manipulating smart meter readings before being communicated to utility providers and retailers. These include non-stochastic methods, such as aggregation, binning, and down sampling, as well as stochastic noise addition. Demand shaping and load scheduling refer to physical privacy enabling techniques, where smart-meter readings are communicated without any manipulation, but instead the consumption is altered by exploiting renewable energy sources, batteries, or by shifting loads with the intention to confuse an attacker that may try to infer consumer behaviour based on smart meter data. This survey paper not only reviews a comprehensive list of privacy-enabling techniques within each of these categories, but also provides a thorough survey and comparison of smart meter privacy measures adopted in the literature.
We hope that this collection of papers contributes to the ongoing research activities in this active research area. We also envisage that the papers in this Special Section will find applications in industry and contribute to policy discussions. A secure and reliable smart grid is essential for the future of our society, and we hope that this collection will help us achieve this critical goal.