Predictive Abuse Detection for a PLC Smart Lighting Network Based on Automatically Created Models of Exponential Smoothing

Abstract

One of the basic elements of a Smart City is the urban infrastructure management system, in particular, systems of intelligent street lighting control. However, for their reliable operation, they require special care for the safety of their critical communication infrastructure. This article presents solutions for the detection of different kinds of abuses in network traffic of Smart Lighting infrastructure, realized by Power Line Communication technology. Both the structure of the examined Smart Lighting network and its elements are described. The article discusses the key security problems which have a direct impact on the correct performance of the Smart Lighting critical infrastructure. In order to detect an anomaly/attack, we proposed the usage of a statistical model to obtain forecasting intervals. Then, we calculated the value of the differences between the forecast in the estimated traffic model and its real variability so as to detect abnormal behavior (which may be symptomatic of an abuse attempt). Due to the possibility of appearance of significant fluctuations in the real network traffic, we proposed a procedure of statistical models update which is based on the criterion of interquartile spacing. The results obtained during the experiments confirmed the effectiveness of the presented misuse detection method.

State‐of‐the‐art in smart streetlight systems: a review

“…A predictive abuse detection system for PLC-based smart streetlights based on exponential smoothing methods is proposed in [77]. In order to detect an anomaly/attack in the network traffic of a smart streetlight system, this study utilises a statistical model to achieve forecast intervals.…”