Research Of Active Queue Management In Network QoS Control

With the rapid development of the network and multimedia technology, both the network users and network applications are rapidly increasing, network users put forward higher request to QoS. The network QoS control has become one of the most important research fields. Queue management plays a very important part in the network transmission and is one of the key technologies of network QoS control and network congestion control.This thesis analyzes the current status and developing trend of research to queue management, analyzes and researches the classical active queue management algorithms, and then summarizes them. The analysis and simulation to RED reveals that the calculation of the average queue length in RED algorithm results in a poor response to the beginning and end of congestion. A modified algorithm called Adaptive-Weight RED (AWRED) is proposed in the thesis. The primary objective of AWRED is to optimize the calculation of the average size of the queue used for congestion avoidance and to consequently reduce the time required by RED to load changing and reduce the total loss of packets. Simulation results show that the proposed AWRED algorithm significantly improves the response time, throughput and delay of RED gateways and shows good robustness.Analysis of the RIO-C algorithm shows that correctly setting the parameters of RIO-C may prove difficult and error-prone, the performance of RIO-C is very sensitive to network conditions, the delay jitter is too big and it has the problem of bandwidth skew. An adaptive AQM algorithm called MRIO is proposed. MRIO bases on DiffServ and supports trustworthy routing. Various theoretical analysises are made to the algorithm and an extensive series of simulations was carried out using the NS-2 network simulator, using diverse network topologies and loads. Simulation results show that MRIO can keep average queue close to a target size while still providing protection for priority packets and fariness of bandwidth allocation and significantly alleviates the bandwidth skew problem, so the algorithm can better realize network QoS control.