Accurate resource estimation for generalized VoIP sources

Abstract

Current voice codecs like G.729, G.723.1 or AMR can generate short background descriptors (SID) frames during voice inactivity periods for Comfort Noise Generation (CNG). This feature alters the classical on-off traffic pattern typically used to model the traffic generated by codecs with a Silence Suppression scheme. Therefore the CNG feature leads to severe inaccuracies in the dimensioning analysis done through traditional models based on multiplexing on-off sources like MMPP or fluid model.In this paper, we focus on the VoIP dimensioning issue. First, we define the traffic pattern generated by those codecs that include CNG (generalized VoIP sources). Second, we extend the traditional MMPP and fluid analytical models to multiplex our generalized VoIP sources and propose a simple but efficient dimensioning algorithm. Results are validated by simulations fed by VoIP traces and demonstrate a significant improvement in accuracy with respect to current on-off based approaches.

Analysis and efficient provisioning of access networks with correlated and bursty arrivals

“…Shah-Heydari et al [1] used Markovmodulated Poisson processes (MMPPs) for characterizing multimedia traffic with short-term and long-term correlation and presented a model using five parameters to approximate a superposition of N homogeneous two-state MMPP to characterize bursty multimedia traffic. Estepa et al [2] extended the traditional MMPP and fluid analytical models to multiplex generalized voice over Internet protocol sources and proposed a simple dimensioning algorithm. Kamoun [3] considered a practical queuing system with a finite number of input links, and whose arrival process is correlated and presented performance analysis of queuing systems with correlated arrivals and service interrup-553 given by Oliver et al [9] and claimed that whether the departure process of an MMPP/M/1 can be a MAP (or MMPP) is an open problem (see also reference therein).…”