Research On Queue Management Mechanism Of QoS Based On PTN Network

With the transformation and growth of the communications network, packet transport network has become the new direction. Quality of service as one of the most important technology for packet transport network, it can guarantee what the users required, so that packet loss probability, time delay, jitter and bandwidth can meet certain requirements. Packet classification and queue management are the main technology for quality of service mechanisms, their purpose is to distinguish the packets with different priority, and therefore the packets with higher priority can be sent preferential.Based on the problems about queue management mechanisms found in the network management project, that was the much more packet loss probability and larger jitter of the network caused by queue management mechanisms, though it can avoid the network congestion through monitoring the length of network queue, this thesis made a further analysis about the queue management mechanisms of quality of service in packet transport network. With the purpose to improve the quality of the network communication, the thesis proposed an improved algorithm about queue management mechanisms on the basis of the WRED(Weighted Random Early Detection) algorithm.The thesis was divided into three parts. Firstly, the quality of service and queue management mechanisms was analyzed based on the problems found in network management project, especially the WRED algorithm. Aimed at the random packet loss strategy, we introduced a threshold on the premise of acceptable network-delay, which can reduce the parameter sensitivity and avoid sudden loss before the network congestion, so that the network performance will not be reduced. The thesis also introduced a nonlinear loss method to reduce the packet loss probability and decrease the possibility of the network congestion. Secondly, we adjust the range of packet loss based on the different priorities in algorithm of differential service about queue management mechanism to meet the high network demands of new business, such as VoIP and IPTV. The thesis increased the dropped lower limit of high-priority service to decrease its packet loss probability, so the high-priority service got priority services, which can meet the service of different levels. Finally, the thesis integrate the improved random packet drop strategy and the algorithm of differential service to build up a new queue management mechanism. The new mechanism acquired packet priority from the node, and then determined the dropped lower limit of different services. At the same time, the oversized threshold was introduced based on the improved random packet drop strategy to decrease the parameter sensitivity. It was feedback by monitoring the average queue length of network. The nonlinear loss method was used to reduce the loss probability. Thus the mechanism reduced the probability of network congestion and improved the quality of communication network. To verify the feasibility and advantages of the improved queue management mechanism algorithm, the thesis simulated the improved algorithm through using NS2 software and compared its results with WRED_gente algorithm, and analyzed the feasibility and practicability of the improved algorithm from several aspects, such as the end to end packet loss probability, the link throughput and the jitter.