An efficient aggregation scheme resisting on malicious data mining attacks for smart grid

Abstract

In the smart grid, efficient power supplies require near-real-time users’ electricity usage metering data, but these data might leak users’ private information, e.g., living habits. To address this problem, a number of privacy-preserving data aggregation schemes have been proposed in the literature. In this paper, we present a new type of attack, called malicious data mining attack, by which the adversary can infer a target user’s electricity usage data. When considering this attack, the majority of existing data aggregation schemes have one of the following two shortcomings. In one aspect, the schemes based on homomorphic encryption can output an accurate aggregation result, but most of them are vulnerable to this attack. In another aspect, the schemes based on differential privacy able to withstand this attack, but the random noises introduced prevent accurate aggregation results from being computed. In this paper, we propose a novel data aggregation scheme that is not only secure against the malicious data mining attack, but also capable of outputting an accurate aggregation result. Detailed security analyses indicate that the proposed scheme satisfies the desirable properties for privacy-preserving data aggregation in the smart grid, and the simulated results demonstrate that our proposed scheme enjoys low computation and communication overhead.

Introduction

The smart grid is a new type of power grid [7], [8] and is proposed to solve all challenges of future electricity supply [12]. It combines traditional power technology with modern information technologies and contributes to more effective infrastructure for energy management [1], [25]. However, the requirement for information exchange and the dependence on networking will undoubtedly expose the smart grid to potential weaknesses associated with communications and network system [1]. To enhance the effectiveness of energy usage, the smart grid needs to collect users’ near-real-time electricity consumption data in order to derive some useful information, such as the electricity consumption peaks period and customer demands. But near-real-time electricity consumption data may contain users’ sensitive information [23], [31]. For example, the low power consumption may be a sign that residents are sleeping or out of the house; the high power consumption may indicate that householders are in the house [18], [32]. Therefore, the realization of protecting users’ privacy is crucial before the smart grid being widely deployed in real-world applications.

In the smart grid, users’ metering data is sent to the control center through a gateway. Assuming that there are n users and the gateway forwards the n encrypted metering data to the control center at fixed intervals (e.g., 5 minutes)[11], [14]. The control center can decrypt the n encrypted data and then analyzes them. But this method is inefficient. And if the external adversaries invade the database of the control center, the adversaries can obtain the metering data of any user [16], [28]. Secure data aggregation is one kind of method to address this issue [28], [34]. By using secure data aggregation techniques, the gateway first aggregates the n encrypted metering data into one encrypted value, and then send it to the control center. After decrypting, the control center can only obtain the aggregation results (e.g., sum, average) but not the individual metering data. Therefore, the external adversaries cannot know anything about the metering data of each user. However, we consider an even more severe scenario: an external adversary can block the communication between a target user and the gateway in some interval, and then the adversary invades the database of the control center. In this way, the adversary can obtain the aggregated metering data of n users in the previous interval as well as the aggregated metering data of $n-1$ users in the current interval. Therefore, the adversary will have the ability to gain some information about the metering data of the target user through the difference of the aggregation result in these two intervals. By launching this attack intermittently, the adversary can infer sensitive information about the target user by analyzing these differences using data mining methods. In real-world applications, the adversaries with this kind of attack ability are common. Hence, the data aggregation scheme in which privacy can be preserved in this scenario may attract more considerable attention in real-world applications. However, when this attack is taking into consideration, the majority of the existing data aggregation schemes are subject to some limitations. Many schemes relying on homomorphic encryption [9], [28], [33], [35], [36], [37] lack the ability to resist this data mining attack, while schemes relying on differential privacy [13], [19], [24], [30] cannot output the accurate aggregation result although they are immune to this attack. This paper pays close attention to realize an efficient data aggregation scheme that is secure against the malicious data mining attack but also capable of obtaining accurate aggregated results.

In this paper, we propose an efficient data aggregation scheme (named AMDM) with accurate aggregated results while resisting malicious data mining attack. This paper contributes to the following three aspects:

• We introduce the malicious data mining attack: a malicious external adversary can derive some information about the metering data of the target user through intermittently blocking the communication between this user and the gateway and intruding the database of the control center.

• We present an efficient data aggregation scheme for the smart grid. The proposed scheme not only resists the malicious data mining attack but also outputs accurate aggregation results without revealing the individual user’s metering data.

• We prove that the security properties, such as data confidentiality, authentication and data integrity, and data privacy, are satisfied in our proposed scheme. Moreover, we conduct performance evaluations of our proposed scheme, demonstrating that it enjoys low computation costs and communication overhead.

The rest of this paper is organized as follows: In Section 2, we formalize the system model, as well as outline the threat model, describe a malicious data mining attack, and identify out security requirements, and provide the main idea in our scheme design. Then, we describe the preliminaries required to understand our scheme AMDM in Section 3. After that, we present our scheme AMDM in Section 4, followed by the security analysis and performance evaluation in Sections 5 and 6, respectively. Section 7 discusses the related works. Finally, Section 8 concludes the paper.