A Parametric Study For Congestion Control In Queuing Networks

Congestion avoidance mechanisms are used to prevent the saturation of the bottleneck under mean-field domain where large numbers of TCP connections interact with one another. It is clear that selections of parameters play an important role on the performance of networks. In the queuing networks, where numbers of TCP connection are too many, there is very high level of incoming traffic at the bottleneck. In order to avoid the congestion at the bottleneck, Random Early Detection (RED) algorithm is implemented and proven of good performance. To improve the existing performance, reported in previous works, this study demonstrates the impact of constant parameters on the performance of queuing networks and justifies with simulation results.