Identification of communication devices from analysis of traffic patterns

Abstract

Recently, variety of communication devices such as printers, IP telephones, network cameras are used widely, with the support of networking in consumer electronics. As a spread of IoT (Internet of Things), the number of embed devices are significantly increasing, however, such devices have lack of capability on security. It is therefore desirable that a network identifies these devices to take appropriate operations. In this paper, we propose an identification method of communication devices from monitoring patterns of traffic, here we use statistical metrics such as packet inter-arrival time or packet size, and we apply a machine learning for the identification. Through evaluations using real traffic, we show that our method can achieve over 90% of identification to 9 commiunication devices.