Queue occupancy estimation technique for adaptive threshold based RED

Abstract

The proper estimation of queue occupancy in gateways is a key for effectiveness of actions taken based on its value. Adaptive threshold based random early detection (RED) uses queue occupancy status (e.g. low, medium and high) to make threshold adjustment decisions. Normally, current value of average queue size is used to determine occupancy. In this paper, we propose a method that takes into account instantaneous and average queue size to estimate queue occupancy. Both present and few past values are analyzed to estimate queue occupancy. The situation of under utilization of link and buffer overflow can be avoided with this method. Simulation results of throughput shows that our method keeps high throughput and low packet drop rate. It also tends to reduce peaks in queue and keeps minimum values higher.