An Intelligent System to Detect the Type of Devices Sending and Receiving Data in the Network

Abstract

Nowadays mobile and fixed devices are used interchangeably for surfing the web due to the huge improvements performed in mobile devices in the recent years. Both mobile and fixed devices with Internet connectivity are supplied with different types of connection, thus users can select the best one at any time depending on their environment. In general, the mobile devices allow users access to Internet using the 3G network or a common WiFi connection, and the fixed ones generally use a wireless or wired connection. Selecting one or another type of connection implies different features of the network environment, so Internet service providers need to adapt their infrastructure to guarantee acceptable levels of Quality of Service in every type of connection. In this paper we study the behavior of the devices according to their nature, that is, if it is a mobile or fixed device. First, we have classified the most significant network parameters and software application values in order to know the nature of the device. Our proposal is based on an intelligent system based on neural networks and finite state machines that lets the Internet service provider know the type of device belongs to the traffic going to its network. The system analyzes the transport and application layers from TCP packets to discriminate the percentage of Internet traffic generated by mobile and fixed devices. Test results show the success of the developed system.