Stabilizing Chaotic Behavior of RED

Abstract

The Internet is a so complex nonlinear network that many results show how the data flow exhibits chaotic attributes and the fractal nature of aggregate TCP/IP traffic. In this work, we study a new model of Random Early Detection (RED) using beta distribution configured by tuning decisions of dropping or accepting packets so that the queue occupancy level is kept at a given target level, thereby eliminating aggressive fluctuations of buffer underflow and overflow. Our proposed model programmed with Python incorporates new parameters (a, B) that make it possible to stabilize oscillations of averaged router queue length and to be close to the stationary state. We present study and numerical analysis from the same perspective of former studies for congestion control.