Resilience enhancement of renewable cyber–physical power system against malware attacks

Highlights

• Malware attacks and renewable energy uncertainty are considered simultaneously in CPPS..

• Prevention & restoration methods are proposed to enhance the resilience of renewable CPPS.

• Factors affecting resilience are further discussed to help build resilient renewable CPPS.

Abstract

The widespread application of cyber technologies and Renewable Energy Sources (RESs) has gradually transformed traditional power systems into renewable Cyber–Physical Power Systems (CPPSs), which are increasingly threatened by malware attacks. In this paper, we develop an original and novel simulation framework with consideration of both malware attack process and the uncertainty of RESs to quantitatively analyze the resilience of renewable CPPS. Based on the framework, we consider different stages of the resilience process and propose corresponding prevention and restoration methods to enhance the resilience of the system. The efficacy and applicability of our proposed methods are demonstrated in a test renewable CPPS formed by coupling the IEEE 39 Bus System with a scale-free communication network. Experimental results reveal that accurate RES output forecast and a disassortative communication network can improve the resilience of renewable CPPS against malware attacks.

Introduction

In modern society, the electric power system is one of the most critical infrastructures that closely related to our daily life. Due to the importance of the system, scholars and engineers have been working together over the past few decades to make our power system works more stably, reliably and efficiently. Therefore, a number of pioneering technologies such as sensing, communication and control are developed and integrated in the power system, making it gradually evolves into a Cyber–Physical Power System (CPPS) [1]. The integration of these new technologies makes CPPS more intelligent than the traditional power system, but at the same time, it also increases the system complexity and brings the new problem of cyber attacks, which poses serious challenge to the reliability and safety of the system [2], [3].

Malware attack is a new kind of cyber attack method against CPPS that has emerged in recent years. As CPPS relies heavily on cyber facilities for wide-area monitoring and control [4], this type of attack often leads to catastrophic consequences [5], [6]. The 2015 blackout in Ukraine [7], for example, was a typical result of such attack, in which hackers forcibly disconnected circuit breakers of the country’s power system by implanting malware in its communication network, resulting in thousands of customers to be deprived of power supply. Since the Ukraine case, malware attacks have been a hot topic in CPPS research. In Ref. [8], Zhang et al. proposed a stochastic state transition model to characterize the malware-induced cascading failures in CPPS, and on this basis they studied the failure propagation patterns in the system. In Ref. [9], Wang et al. analyzed the interaction between malware propagation in the communication network and failure propagation in the power network, and then proposed an interactive cascading failure model to simulate malware attacks in CPPS. Based on these works, the effect of malware incubation period was analyzed in [10] and an optimal coupling strategy was proposed in [11] to further improve the robustness of CPPS. The above works established a fundamental framework for analyzing malware attacks in CPPS, however, most of them only concentrate on modeling the malware attack process, while prevention and restoration methods for such attacks are neglected.

In reality, prevention and restoration methods are a critical part of building a reliable CPPS, enabling the system to obtain the ability to bounce back from malware-caused damage and maintain essential services. In the field of reliability engineering, such ability is called resilience, which refers to the capability of a system to withstand, accommodate and recover from disruptions [12], [13]. Specific to the resilience of power systems, there are a lot of studies that have been conducted. For instance, Senkel et al. [14] developed a simulation framework incorporating the intrinsic dynamics of the power system to assess its resilience. Shen et al. [15] studied the key factors affecting power system’s resilience and found that the power system is less resilient in more urbanized areas. Additionally, in terms of resilience-enhancing methods, Zhang et al. [16] developed a double-loop optimization method to help power systems better recover from external attacks. Jalilpoor et al. [17] developed a two-stage optimization method to strengthen and optimize the configuration of power system components. Wu et al. [18] established a defense–attack–recovery tri-level game optimization method to allocate defensive and restorative resources against consecutive attacks. However, all the above studies are about standalone power systems, which cannot well characterize the CPPS.

Regarding the resilience of CPPS, the interdependence between the power network and the communication network is a key factor to be considered. In the literature, Ouyang et al. [19] and Almoghathawi et al. [20] studied cascading failures in interdependent power infrastructures and proposed joint restoration strategies to reinforce the resilience of the system. Mana et al. [21] presented a co-simulation model that took into account the operational mechanisms of both power and communication networks to evaluate the impact of different communication technologies on CPPS resilience. Similarly, Wu et al. [22] modeled the cascading failures in CPPS and optimized the repair sequence of the power and communication components to achieve a better recovery of the entire system. The aforementioned works are insightful, however, most of them only focus on studying one aspect of the resilience process (i.e., recovery), lacking a holistic and systematic approach. Furthermore, when considering malware attacks, to the best of the authors’ knowledge, the resilience of CPPS to against such attacks is still rarely investigated in the existing literature.

More importantly, in order to reduce greenhouse gas emissions, more and more Renewable Energy Sources (RESs) such as wind energy and photovoltaics are now integrated in the power system, turning the CPPS further into renewable CPPS. However, the stochastic nature of the RESs drives the renewable CPPS works closer to its limits and introduces plenty of uncertainty factors, which poses new threat to the stability of the system [23], [24], [25]. Therefore, renewable CPPS is more vulnerable to malware attacks and requires a better resilience to resist such attacks. In this field, although several works [26], [27] have studied the resilience of renewable energy penetrated power systems, research on renewable CPPS is still relatively lacking.

To bridge the gap, in this paper, we develop resilience enhancement methods for renewable CPPS to defend against malware attacks. Our main contributions are summarized as follows.

• Compared with previous studies, we first establish a novel simulation framework to analyze the resilience of renewable CPPS, taking into account both the malware attack process and the uncertainty of RESs.

• We consider multiple stages in the resilience process of renewable CPPS. According to the characteristics of different stages, we put forward corresponding prevention and restoration methods to enhance the resilience of the system, providing a holistic and systematic methodology.

• Based on the proposed simulation framework, we evaluate the effectiveness of the resilience enhancement methods and show how external factors influence it. Our findings provide practical insights for building a resilient renewable CPPS.

The rest of the paper is organized as follows. Section 2 presents the modeling methodology for renewable CPPS, and Section 3 describes the malware attack process in the system. In Section 4, we introduce the resilience process of the system and develop resilience enhancement methods based on it. Finally, we conduct case studies in Section 5 and draw conclusions in Section 6.