PERFORMANCE ANALYSIS OF ACTIVE QUEUE MANAGEMENT ALGORITHMS

Santhi V, Natarajan A.M
2009 International Journal on Information Sciences and Computing  
I.INTRODUCTION Congestion in a network or internet creates obvious problems for the end system: reduced availability and throughput and lengthened response times. When a packet is dropped before it reaches its destination, all of the resources it has consumed in transit are wasted. Improving the congestion control and queue management algorithms in the Internet has been one of the most active areas of research. The Internet has mainly relied on the cooperative nature of TCP congestion control
more » ... order to limit packet loss and fairly share network resources. However new applications are being deployed which do not use TCP congestion control and are not responsive to the congestion control and are not responsive to the congestion signals given by the network. Such applications are potentially dangerous because they drive up the packet loss rates in the network and can eventually cause congestion collapse. The aim of this paper to compare the performance analysis of active queue management techniques called Blue, Stochastic Fair Blue, and Core-stateless Fair Queuing with RED. Blue and Stochastic Fair Blue is used to reduce packet loss rates, queuing delay, high link utilization with minimal amount of buffer space. Core-stateless fair queuing enforces fairness among large number of connection with small amount of state information. BLUE uses the packet loss and the link utilization history to manage congestion. Also, only a single marking probability is maintained, when the queue is continually dropping packets due to buffer overflow, this probability is incremented, and when the queue is idle or empty, it is decremented. Abstract Active Queue Management (AQM) can potentially reduce packet loss rate in the Internet. This is used by routers for control congestion, where packets are dropped before queues become full. A number of active queue management algorithms for TCP/IP networks such as random early detection (RED), Fair RED (FRED), BLUE, Stochastic Fair Blue (SFB), and Corestateless Fair Queuing (CSFQ) have been proposed in the past few years. This article presents a comparative study of these algorithms using ns-2 simulations. The performance metrics used in the study are queue size, packet drop marking probability, packet loss rate and bandwidth utilization. The study shows that BLUE is better than RED to avoid global synchronization for maintaining single marking probability. And also, the results shows that, among the five algorithms, SFB and CSFQ are more effective at stabilizing the queue size and controlling the packet loss rate among non-responsive flows while maintaining high link utilization. The performance of SFB and CSFQ are obviously better than that of RED, FRED and BLUE.
doi:10.18000/ijisac.50036 fatcat:f72xswzzo5bufneoo6xxio5zli