Quality of Service Parameters Modeling of Voice over IP Traffic

Abstract

Voice over Internet Protocol (VoIP) is one of the most attractive and important service nowadays in telecommunication networks. However, when the voice traffic is transported over IP Networks, the packet based transmission may introduce impairments and it has influence on the Quality of Service (QoS) perceived by the end users. The voice quality of VoIP systems depends on many QoS parameters, such as: bandwidth, one way delay (OWD), jitter, packet loss rate, CODEC type, voice data length, and dejitter buffer size. Particularly, jitter and Packet Loss Rate (PLR) have an important impact on voice quality. These parameters are intricately related to each other and can be used to configure other parameters to optimum values in order to afford good levels of voice quality.To achieve the satisfactory voice quality, the VoIP applications must be designed by using representative traffic models. In order to implement representative traffic models it is necessary to monitor the QoS parameters, by means network measurements.In this lecture, it is shown that VoIP jitter can be modeled by selfsimilar processes with Short Range Dependence (SRD) or Long Range Dependence (LRD). On the other hand, it is described the VoIP packet loss based on microscopic and macroscopic behaviors, where these behaviors can be modeled by 2-state and 4-state Markov chains, respectively. Besides, the relationship between Hurst (H) and the PLR it is shown; where this relationship is based on voice traffic measurements and is modeled by means of a power-law function, characterized by three fitted parameters.On the other hand, the E-Model does not consider the impairments caused by jitter and packet loss burstiness (i.e., it assumes uniform packet loss). The packet loss burstiness level is interpreted as the concentration level of packets lost over a period of time (window time). Therefore, the power-law model that relates H and PLR, can be used to provide additional information of these impairments and improve the E-Model.