Simulation of an active queue management in a quality of service internet using random early detection (RED)

Abstract

This paper work presents simulation of an active queue management in a quality of service Internet using differentiated service random early detection (DiffServ RED) algorithm for congestion avoidance in packet switched network. RED algorithm is designed to accompany a transportlayer congestion control protocol such as TCP. We explored how the algorithm, earlier designed by [8], could be used to achieve differential packet dropping required by realtime traffic in the extended TCP/IP protocol for the Internet. The queue manager operates by setting thresholds (minimum and maximum) for the average queue size to detect incipient congestion. When the average queue size is below the minimum threshold, all arriving packets are properly enqueued for dispatch to their destinations. Any time the average queue size exceeds a preset minimum thresholds, the gateway drops packets at random or marks each arriving packet with a certain probability, where the exact probability is a function of the average queue size. A case when the average queue size exceeds the set maximum threshold, the gateway drops all other incoming packets. Differential packet dropping demands that sources violating their agreed sending profiles should have some of its packets discarded during incipient congestion to reduce global synchronization. This study simulated using ns 2, how RED logic marks or drops at random compliant and noncompliant packets means to highly increase the throughput and drastically reduce losses in order to step down congestion over the network. The RED gateway has no bias against bursty traffic and avoids global synchronization of many connections decreasing their windows at the same time.