Analytical Description of a Parameter-based Optimization of the Quality of Service for VoIP Communications

In this work, the perceived quality of VoIP communications is studied. The distributions of the number of consecutive received and lost packets, respectively named gap and burst, of a VoIP communication are modeled with discrete two-state and four-state Markov chains. Algorithms for estimating the transition probabilities between states and from these, the packet loss rate and the respective gap and burst length distributions, are described. Through a study of monitored VoIP calls, it is shown that these models can adequately represent the geometric-type decay of these distributions and that although two-state model performs well for homogeneous losses, for non-homogeneous losses the four-state model fits better. An analysis of the performance of a packet-level FEC scheme, based on -packet redundancy, is presented. The perceived packet loss rate that results of applying this correction scheme is quantified. For the studied measurements, 1-packet redundancy is sufficient to decrease the perceived loss rate below 1%. Also, the impairments of the perceived quality of voice after the FEC technique and a de-jitter buffer is quantified. The resulting equations can be used to optimize the adjust parameters of the VoIP call, e.g., level of redundancy, type of codec used and de-jitter buffer size. The proposed methodology can be extended if other types of improvements are included. Key-Words: VoIP, Packet loss distribution, Packet reception distribution, -packet FEC, QoS.