An Improved Controlled Delay Algorithm For Active Queue Management

Active queue management is an important method to avoid network congestion and it can improve the quality of service(QoS). Available resources in the network are limited. With the rapid development of the network, the number of data traffic in network is much bigger than before. When the data traffic exceeds the upper control limit of the network, network congestion will appear. The network congestion will lead to the increment of packet lost rate and bigger transmission delay, seriously it can make the network paralysis. Frequent network congestion and difficult recovery are unable to provide good QoS guarantee. Active queue management algorithm is an effective method to avoid congestion control. It reduces the probability of network congestion and improves the QoS of network. It has great academic significance and practical application value.This thesis describes research significance and research status of active queue management. What's more, it studies the advantages and disadvantages of controlled delay(CoDel) algorithm. Then it makes several improvements on CoDel.This thesis improves the controlled delay(CoDel) algorithm for active queue management and proposes a probabilistic optimized controlled delay(PCoDel) algorithm. CoDel controls the queue delay strictly. When the bursty traffic comes, it will drop plenty of packets and decrease the network performance. On the basis of CoDel, PCoDel judges the network congestion by average packets delay in the queue. It adds the mechanism that dropping the packet by probability when packets enqueue.PCoDel adds cautious adaptive method to improve the stability of the algorithm,. It is the improvement of additive increase multiplicative decrease(AIMD) to control the average sojourn time of packets. When the average sojourn time becomes bigger, it correspondingly increases initial dropping probability. When the average sojourn time becomes smaller, it correspondingly reduces the initial dropping probability.Besides, in order to meet the requirements of users closely, algorithm should provide different quality of service(QoS) to different priorities. This thesis proposes weighted probabilistic optimized controlled delay(WPCoDel) algorithm. It can provide differentiated service. Finally, we do the experiments in different network simulations. The simulation experiments show the comparison with the traditional algorithms CoDel, sfqCoDel, PCoDel and WPCoDel algorithm under the situation of light load, overloading and mixed TCP and UDP flows. PCoDel and WPCoDel have the advantages in throughput, packet loss rate, delay, link utilization rate and queue length, especially in throughput and packet loss rate.