e-ISSN: 2582-5208 International Research Journal of Modernization in Engineering Technology and Science Volume:02/Issue:06/June-2020 www.irjmets.com www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science [778] IOT DEVICE IDENTIFICATION THROUGH NETWORK TRAFFIC ANALYSIS K Ravi Kumar *1 , Challa Hemanth *2 , Ch. Aravind Kumar *3 , K M Sahith *4 , G Andrews Prasanth *5 *1 Asst Professor, Department of Electronics and Communication Engineering, Vasireddy Venkatadri Institute of Technology, Guntur, AP, India. *2,3,4,5 Student, Department of Electronics and Communication Engineering, Vasireddy Venkatadri Institute of Technology, Guntur, AP, India. ABSTRACT In this digital era, everything is being automated. Here comes the necessity for internet connected things, where IOT comes into picture. A part from IOT, we also need autonomous systems to take decisions where ever possible. Tremendous developments are being made in the field of IOT and Machine Learning. In this work, we apply machine learning algorithms on network data for identification of IOT devices, connected to a network. Using supervised training, we can distinguish between the traffic generated by IOT and non-IOT devices. Through which we can identify the IOT devices. KEYWORDS: IOT, Security, Machine Learning, Network Traffic, Supervised Training. I. INTRODUCTION With the proliferation of Internet Technology, a lot of data is being generated everyday in various networks existing in this World Wide Web. Not just from mobile phones and computers, data is also generated from various kinds of IoT devices. A network may be comprised of many IoT and non IoT devices. Internet of Things opened up a whole new channel of possibilities for future in various domains. Along with the advantages, IoT devices also put forth a lot of challenges like security and governance to the organizations. There is also a threat of not knowing what devices are IoT devices and what are not. This situation threatens integrity and security of all devices and also risks the organization’s existence. Hence and identification of these devices is very important for the network administrators. The methods employed now are identification through IDs, MAC and IPs. However, these can be easily spoofed and misused. So, we have developed a machine learning model to effectively identify and manage the devices on our network, based on their network traffic traces. These models are capable of identifying the devices based on their network behaviors and identify the actual devices without the influence of third party or anonymous accesses. II. DATA SET In order to train the Machine Learning Models, the data set is derived from various sources. The data set comprises of network traffic traces of around 22 devices, connected on a WiFi Hub. The devices are kept under observation and the network traffic traces of all the devices are logged into capture files. Fig-1 Smart Home Network