Research And Application Of Network Traffic Control Technology

Real-time multimedia services and non real-time services are applied more and more widely on Internet. These applications require different QoS, including different bandwidth, delay and jitter. Due to the idea of best-effort servicing in TCP/IP, with the booming of network traffic, the packet loss rate increases remarkably, network performance becomes worse. So, it is a hot research topic to introduce the concept of QoS and support different kinds of QoS requirements through multiple methods in network.On the other hand, the scale of network is growing and the topology is becoming more complex than ever. It is a great challenge to network management and operation, which affects the network performance directly. It is an important problem to improve the quality of network operation through resource allocating and network traffic controlling appropriately.Traffic control technologies are employed to solve these problems above in this dissertation. How to control the in/out traffic, prevent the occurence of congestion, guarantee the quality of key services, meanwhile control the ordinary services to share the bandwidth fairly are discussed.The traffic control model that supports DiffServ is discussed in detail, especially the queue scheduling and traffic shaping modules. Queue scheduling is one of the key mechanisms that support the quality of real-time services in network nodes. Some algorithms based on DifferServ model are introduced. An algorithm ADWRR improved from DWRR is proposed, which supports the real-time multimedia services. Employing ADWRR combined with Token Bucket algorithm to shape the traffic can decrease the end-to-end delay and transmission jitter so better quality of services can be provided.At last, the traffic control modules are added into NS2 kernel, and the simulation experiments are designed to validate the performance of ADWRR and Token Bucket algorithms. It is shown that the traffic shaping module can decrease the end-to-end delay and transmission jitter of audio/video traffic flows, meanwhile provide fair bandwidth sharing for other flows.