Dimensioning of dual leaky bucket parameters for regulating voice sources

An analytical approach to set the parameters of a dual leaky bucket for the purposes of regulating a voice over IP call at the network ingress is presented. The analytical model is used to determine the theoretical packet loss as a function of the dual leaky bucket parameters for traffic generated by a G.726 codec, then the theoretical packet loss is compared with that measured using a real voice conversation taking place over a Linux system and using an implementation of the G.726. The results are found to be in excellent agreement.