Detection of Potentially Compromised Computer Nodes and Clusters Connected on a Smart Grid, Using Power Consumption Data

Abstract

Monitoring what application or type of applications running on a computer or a cluster without violating the privacy of the users can be challenging, especially when we may not have operator access to these devices, or specialized software. Smart grids and Internet of things (IoT) devices can provide power consumption data of connected individual devices or groups. This research will attempt to provide insides on what applications are running based on the power consumption of the machines and clusters. It is therefore assumed that there is a correlation between electric power and what software application is running. Additionally, it is believed that it is possible to create power consumption profiles for various software applications and even normal and abnormal behavior (e.g., a virus). In order to achieve this, an experiment was organized for the purpose of collecting 48 h of continuous real power consumption data from two PCs that were part of a university computer lab. That included collecting data with a one-second sample period, during class as well as idle time from each machine and their cluster. During the second half of the recording period, one of the machines was infected with a custom-made virus, allowing comparison between power consumption data before and after infection. The data were analyzed using different approaches: descriptive analysis, F-Test of two samples of variance, two-way analysis of variance (ANOVA) and autoregressive integrated moving average (ARIMA). The results show that it is possible to detect what type of application is running and if an individual machine or its cluster are infected. Additionally, we can conclude if the lab is used or not, making this research an ideal management tool for administrators.

A Comprehensive Review on Smart Grids: Challenges and Opportunities

“…Neighborhood Area Networks (NAN) [22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37] Software Defined Networks (SDN) [20,[38][39][40][41][42][43] Interdependent networks (IN) [44][45][46][47][48][50][51][52][53] Field Area Networks (FAN) [21,49,[54][55][56][57][58][59][60][61][62][63]69] Wireless Sensor Networks (WSN) [64][65][66][67][68][69]...…”

“…Note that 17% can find both in NANs and WSNs, respectively, pointing out a great opportunity to develop smart grids with softwaredefined topologies. Internet of Things [19,23,24,27,28,30,31,36,37,58,64,66,73,74,79,105,107] Machine learning [20,29,32,51,55,56,60,62,69,78,80,81,85,98,102,106] Data mining [47,107] Machine learning and neural training [21,54,69,81,96] Short term memory network [41,61] Power Line Communication Technology [53,63,90] Power electronics [35,50,99,101,<...>…”

“…Table 3 shows that most of the Encryption issues do not clearly define the algorithms they use, although there are some articles that use Multi-dimensional Data aggregation and even Cognitive Risk Control. Multi-dimensional Data aggregation [42,[55][56][57]60,62,64] Cognitive Risk Control [37,82] Not Defined [19,20,[22][23][24]26,28,29,31,35,36,38,39,43,[45][46][47][48]50,[52][53][54]58,61,63,73,[76][77][78][79]81,[98][99][100][101][102][103][104][105]…”

“…Almshari et al [23], propose network security as an important point nowadays because of the many requirements to be match by the customer such as the privacy of the most user's sensitive data of an electrical network. The whole of sensors or smart devices that make up an intelligent network consume energy, as well as share their information among the members of the network, is considered as an IoT network.…”

A Novel Monte-Carlo Simulation-Based Model for Malware Detection (eRBCM)

“…Different malware detection approaches in the literature have adopted different machine-learning techniques, such as random forest (RF) [5][6][7], neural network [9][10][11], decision tree [12,13], naïve Bayes [14,15], KNN and SVM [15], ARIMA [16], and reinforcement learning [17,18].…”