Improved Radial Movement Optimization With Fuzzy Neural Network Enabled Anomaly Detection for IoT Assisted Smart Cities

Abstract

Recently, an extensive implementation of the recent Internet of Things (IoT) model has resulted in the development of smart cities. The network traffic of smart cities using loT systems has developed rapidly and established novel cybersecurity problems later these loT devices are linked to sensors that are directly linked to huge cloud servers. Unfortunately, IoT systems and networks can be identified as extremely exposed to security attacks that aim at service accessibility and data integrity. Additionally, the heterogeneity of data gathered in distinct IoT devices, composed of the disturbances acquired in the IoT systems, renders the recognition of anomalous performance and threatened nodes very difficult related to typical Information Technology (IT) networks. Accordingly, there is a critical requirement for reliable and effectual anomaly detection (AD) for identifying malicious data to promise that it could not be utilized in IoT lead to decision support systems (DSS). This manuscript offers an Improved Radial Movement Optimization with Fuzzy Neural Network Enabled Anomaly Detection (IRMOFNN-AD) technique for IoT Assisted Smart Cities. The main purpose of the IRMOFNN-AD algorithm lies in the accurate and automated detection of the anomalies that exist in the IoT environment. For the feature selection process, the IRMOFNN-AD technique uses the IRMO system to elect an optimum set of features. Additionally, the IRMOFNN-AD algorithm applies the FNN model for the detection and classification of anomalies. Besides, the sine cosine algorithm (SCA) has been employed for the parameter tuning of the FNN algorithm. The simulation value of the IRMOFNN-AD system has been tested on benchmark IDS datasets. The extensive results illustrate the better detection outcomes of the IRMOFNN-AD system interms of different measures.