The location of optimum set-point using a fuzzy controller

Abstract

This paper presents an intelligent algorithm using fuzzy set and fuzzy logic theory to select an adaptive initial set-point of the bottle-neck buffer to complement a recently proposed delay-based end-to-end congestion avoidance scheme. The set-point is self-tuned from an initial value to an optimum value according to the requirement of the control system. If the initial set-point is far from the optimum set-point, especially when the size of the bottleneck buffer equals to the value of the bandwidth-delay product, the loss ratio and the delay and delay jitter must be increased. Aimed at improving all parameters of QoS regardless of the properties of the network such as the buffer size and the link speed, this paper derives a fuzzy controller to select the initial set-point based on the measured maximum queuing delay to make the initial setpoint close to the optimum value. As a result, parameters of QoS are not degraded when bottleneck links have different buffer sizes and link speeds. The proposed algorithm is tested through the use of extensive simulations to prove its capability to improve network performance in comparison with existing algorithms. © Springer-Verlag Berlin Heidelberg 2005.