Modeling and Performance Analysis of Dynamic Random Early Detection (DRED) Gateway for Congestion Avoidance

One of the most prominent congestion avoidance schemes in the Internet architecture is the Random Early Detection (RED) algorithm. Several modifications and enhancements have been made to the original RED so as to make it more responsive to congestion avoidance at the gateways. In this paper, we introduced the Dynamic Random Early Detection (DRED) model, which uses a newly introduced parameter i.e. warning line. A robust and efficacious technique to measure the burstiness of incoming traffic has been developed and tested. This involves the estimation of the average queue size, avg, which is dynamically adjusted hence the name of our scheme. The empirical results obtained from the simulations show that our DRED scheme responds early enough to the increased number of packets at the gateway. Also, the maximum drop probability of packets show improved performance over the original RED. It was concluded that our scheme demonstrated superiority by avoiding global synchronization and there is great reduction in the fluctuations of the actual queue size. Also, its early response avoids buffer overflow at the gateways when the queue is near full.