Network Traffic Identification And QoS Control In Home Networks

With the continuous development of smart home, the introduction of the home gateway and the diversification of network applications, also the developing network terminal, bringing a variety of new problems and impact for the home environment. Intelligent home network equipment constantly introduced, such as smart TV, tablets, smart phones, greatly expanding the range of multimedia applications, video on demand, video conferencing, video telephony growing popularity, these networks tend to occupy a larger bandwidth, requiring more large network traffic, and the current average bandwidth of our country is still in the relatively backward position, network terminal and increasingly diverse network applications constrained by limited bandwidth, so that the user can not get a good web experience.Traditional network QoS (Quality of Service, QoS) technology is stuck in trouble, the means of traffic identification is too simple, and so the expansion of the network traffic identification method has become a research direction. The ability of QoS for network application identification is not enough. The application can be identified only by the characteristic left in the packet. In order to solve this, the paper presents real-time network traffic identification and QoS control system based on the home network.Mainly for the variety of applications in the home network and the characteristics of the applications, we do the research of network traffic identification methods. The work is as follows:Firstly, through the analysis of some common applications in the family, we focus on the identification of several types of applications which have a greater impact on network traffic. Traffic identification module uses port numbers, five quintuple information and BP neural network to successfully identify several types of applications. BP neural network uses Libpcap to capture the packets which flow through the routing system to extract the desired flow characteristics for off-line training. Data off-line training was to provide support for real-time identification online. In terms of system implementation, the traffic identification module is ported to be part of the home gateway OpenWrt, which can make the network terminal be controlled. Currently, traffic identification module success rate has reached more than 80% for application recognition. QoS can be achieved accurately controlled based on the recognition results.Secondly, for a variety of applications in family environment, the QoS policy settings preset, so as to solve the problem of limited bandwidth resources and competition. Paper proposes a method for controlling the management of QoS settings for different applications home network traffic. Establish a new queue scheduling on Linux systems, and take advantage of the traffic identification module identification classification results, propose a better solution to the traditional scheduling strategy for identifying cases of insufficient network applications.Thirdly, we built the QoS control system on the basis of traffic identification and QoS control. Router equipped with OpenWrt system as a home gateway, capturing traffic packet flowing through the routing system for analysis. As the center of the home network, almost home network terminals are all connected to the routing device, we will control the other network terminal equipment by controlling the routing system. Experiments show that QoS control system can effectively allocate network traffic in home environment. Use the tools under Linux TC configure, the user can control the system through a web interface, thereby enhancing the user’s web experience.