Review paper on Analysis and Design of Proactive Active Queue Management Techniques with MATLAB Simulator on Various Performance Metrics

Abstract

Network congestion can seriously degrade the quality of services delivered to end users. Though a congestion control mechanism is conventionally implemented in end to-end systems to keep the core network flexible and simple, researchers are seeking other possibilities to effectively control congestion. The rapid development in hardware technology means that intermediate nodes, such as routers, gateways and switches are now powerful enough to handle extra tasks such as congestion control. Active Queue Management (AQM) controls congestion by handling incoming packets based on the status of the queue in an intermediate node. Random Early Detection (RED) is regarded as the first practical AQM scheme and has been recommended by IETF for deployment on the Internet. It can reduce end-to-end delay and prevent consecutive packet drops which usually results in oscillation in link utilization and TCP (Transport Control Protocol) global synchronization. Most AQM methods utilize Explicit Congestion Notification (ECN) to further improve the performance of control. In addition, Self Configuring Random (Adaptive) Early Detection (ARED), one emerging variants of RED. They all perform well in terms of queuing delay, packet drop rate and link utilization. These three performance indices are also known as network-centric (or router-based) metrics. However, a user is more interested in the end-to-end behavior of a network service such as end-to-end delay and packet loss ratio.