Leveraging 5G and cloud computing for outlier detection in IoT environments: A KNN approach

Abstract

Internet of Things (IoT) becomes a prominent sensing paradigm between the devices. Its evolution in the global digital increases extensively in various domains. For IoT application's sensors are the primary source for generating data. These collected data are subject to the identification and detection of outliers/anomalies. The massive volume of data generation makes anomaly detection a complex and challenging task. The anomalies affect the data accuracy and data quality. In this paper, the k‐NN classifier is proposed for enhancing classification accuracy. K‐NN follows a non‐parametric strategy and is one of the known classification algorithms. In the proposed system, k‐NN is utilized to perform classification or regression with estimations of their k nearest neighbors. The proposed system consists of three major processes such as data preprocessing, classification, visualization. This study explores the utilization of 5G connectivity and cloud computing infrastructure for outlier detection in IoT data streams. Leveraging the K‐Nearest Neighbors (KNN) classifier, our methodology focuses on efficiently identifying anomalies in IoT data. By integrating 5G connectivity for real‐time data transmission and cloud‐based machine learning for scalable analysis, we demonstrate a robust framework for outlier detection in IoT environments. The Experimental work with the proposed method is carried out using training and observation is tabulated with respective classes. As a result, on the three metrics, the proposed k‐NN proves its efficiency is far better than the others, with an average of 98.4% of accuracy.