Received Signal Strength and Optimized Support Vector Machine based Sybil Attack Detection Scheme in Smart Grid

Abstract

As smart grid enables two-way flow of data and electricity with Advanced Metering Infrastructure, it is prone to security vulnerabilities. Sybil attack, one such vulnerability exhibits multiple identities of same node. As a consequence, the compromised or malicious nodes present in smart grid inject false information that would cause a serious impact in a critical infrastructure i.e. smart grid. Hence, the proposal aims to detect this attack based on node's Received Signal Strength, address, energy consumption and distance using machine learning algorithm. Support vector machine outperforms other machine learning algorithms like logistic regression, K-Nearest Neighborhood, Naive Baye's, and K-Nearest Neighborhood in terms of accuracy, training time, misclassification cost, prediction speed, sensitivity or recall, specificity, F1 score, precision, and Area Under the Curve (AUC) and Receiver Operating Characteristic Curve (ROC). Further, the performance of the model is optimized using hyper parameter tuning. The proposal is implemented in MATLAB. The results exhibit 96.5% accuracy that clearly demonstrates the efficacy of the model.