Difference-Comparison-based Malicious Meter Inspection in Neighborhood Area Networks in Smart Grid

Abstract

As the smart meters are vulnerable to physical attacks as well as cyber attacks, electricity theft in smart grids is much easier to commit and more difficult to detect than that in traditional power grids. In this paper, to facilitate the inspection of the malicious meters, a full and complete binary inspection tree whose leaves stand for smart meters is employed as a logical structure. We can logically configure an inspector (a meter for detection) at any node on the tree. By calculating the difference between the inspector's reading and the summation of the readings reported from the smart meters on the subtree of one node, as well as the difference between the total amount of stolen electricity on the subtrees of an internal node and its left child, we propose a difference-comparison-based inspection algorithm which allows the inspector to skip a large number of nodes on the tree and hence accelerates the detection speed of the malicious meters remarkably. Furthermore, for quickly identifying a complete set of malicious meters, we propose an adaptive reporting mechanism which adopts much shorter reporting periods during the inspection process. Analysis with proofs about the performance bounds of the proposed algorithm in terms of the number of inspection steps is provided. Simulations not only validate the theoretical analysis, but also show the superiority of the proposed algorithm over the existing works in terms of inspection steps, regardless of the ratio and the permutation of malicious meters.

SAI: A Suspicion Assessment-Based Inspection Algorithm to Detect Malicious Users in Smart Grid

“…In Fig. 10, we compare the SAI algorithm with the ATI algorithm [16] as well as the DCI algorithm [17] in static cases. 5 We assume that there are a total number of 250 users Fig.…”

“…However, the BCGI algorithm can only deal with the case where there is one malicious user. In papers [15]- [17], an "inspector box" is installed at the distribution room in each neighborhood area network (NAN). Malicious users are identified by comparing the inspectors' own readings with the users' reported readings.…”

“…Malicious users are identified by comparing the inspectors' own readings with the users' reported readings. The authors in papers [16], [17] adopt a binary tree as a logical structure to facilitate the inspection process. They propose a series of inspection algorithms, by which inspectors dynamically traverse on the binary tree to detect malicious users.…”

“…It is a heuristic approach by which inspectors skip some internal nodes on the binary tree to directly inspect nodes at lower levels. Difference-Comparison-based Inspection (DCI) algorithm [17] can skip a large amount of nodes on the binary tree, and hence can locate malicious users faster. However, it assumes that inspectors can check malicious users' logs to know their true electricity consumptions.…”

“…We here do not use the metric of the number of inspection steps as in papers[16],[17]. This is because the malicious users in this paper are generated based on a Bernoulli distribution.…”