Stability Analysis of Active Queue Management Techniques in NS2

Due to rapid growth of Internet and demands for multiple access the request for services have also increased significantly. As a result there is congestion in the network where buffer management plays an important role. Traditionally, Droptail was used as buffer management techniques along with TCP. Due to large queuing delay and TCP synchronization problem of Drop tail, Active Queue Management (AQM) technique has been used to overcome the problem. AQM algorithm has been proposed as one of the router based mechanism for early detection of congestion inside the network and helps end devices to control the congestion. However, to get high throughput and better link utilization the queue length needs to be more stable. In this paper we analyzed the stability and oscillatory behavior of various AQM techniques like Random Exponential Marking (REM), Adaptive Virtual Queue (AVQ), Random Early Detection (RED), Gibbs Kelly (GK), and Proportional Integrator (PI) with Drop Tail (DT) including RED variants such as RED, Gentle RED (GRED) and Nonlinear RED (NLRED), Adaptive RED (ARED), Refined ARED (Re-ARED). Our result indicates that RED can stabilize the queue around the target value and produce more throughput by minimizing the oscillation than others using NS2. However among RED and its variants NLRED performs better in terms of aggregate throughput and stabilizes the queue length than others.