The Internet of Things (IoT) also called the Internet of Everything is a system of smart interconnected devices. The smart devices are uniquely identifiable over the network and perform autonomous data communication over the network with or without human-to-computer interaction. These devices have a high level of diversity, heterogeneity, and operates with various computational capabilities. It is highly necessary to develop a framework that allows to classify the devices into different categories from effective management, security, and privacy perspectives. Various solutions such as network traffic analysis, network protocols analysis, etc. have been developed to solve the problem of device classification. The signal of a device is an important feature that could be utilized to classify various network devices. We propose a framework to identify network devices based on their signal analysis. We have developed a training data set, by collecting signals from various Wi-Fi and Bluetooth devices in a specific geographic area. A machine learning-based model is proposed for the prediction of network device classification (e.g., a Wi-Fi or Bluetooth device) with 100% accuracy. Furthermore, clustering techniques are applied to the acquired signals to predict the total number of active Wi-Fi devices in a given region.
Classification of IoT Device Communication Through Machine Learning Techniques
Ahmad S.;Ullah Z.;Mostarda L.
;
2021-01-01
Abstract
The Internet of Things (IoT) also called the Internet of Everything is a system of smart interconnected devices. The smart devices are uniquely identifiable over the network and perform autonomous data communication over the network with or without human-to-computer interaction. These devices have a high level of diversity, heterogeneity, and operates with various computational capabilities. It is highly necessary to develop a framework that allows to classify the devices into different categories from effective management, security, and privacy perspectives. Various solutions such as network traffic analysis, network protocols analysis, etc. have been developed to solve the problem of device classification. The signal of a device is an important feature that could be utilized to classify various network devices. We propose a framework to identify network devices based on their signal analysis. We have developed a training data set, by collecting signals from various Wi-Fi and Bluetooth devices in a specific geographic area. A machine learning-based model is proposed for the prediction of network device classification (e.g., a Wi-Fi or Bluetooth device) with 100% accuracy. Furthermore, clustering techniques are applied to the acquired signals to predict the total number of active Wi-Fi devices in a given region.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.