Anomaly detection in smart city parking data: A case study

Abstract

Wireless Sensor Networks (WSNs) are a principal technology in many smart city projects to collect data from the streets and send it to the data centers of the municipalities. Nonetheless, WSN are easy to attack and, therefore, smart city administrators have to verify that the WSN data is faultless. In this article we present a case study on anomaly detection in smart city parking data using WSN information from a Barcelona smart city provider. In order to analyze this data, we used Oneclass Support Vector Machines (OC-SVM) in a dataset with application information from the parking sensor readings and also system status data. The results of the anomaly analysis were satisfactory, achieving a 97.13% detection rate and 13.13% false positive rate.