One of the most important services of today’s communication networks is voice service. Voice over Internet Protocol (VoIP) or internet telephony is the technology to traort voice packets over internet instead of using traditional PSTN (Packet Switching Transmission Network). VoIP traffic also contains connections that are machine-to-machine transactions such as faxes, point-of-sale credit card charge verifications, and computer connections to the Internet Transmitting voice over internet is the necessity of modern applications which combine voice and data. VoIP system has many options of configurations. Normally a VoIP system includes encoder such as G.711, G.728, G.729, G.723.1 and packetizer (RTP, RTCP) in the sender part, the internet, and depacketizer and decoder in the receiver part. There are a number of benefits of Voice over Internet Protocol which underlines the importance of VoIP: reduced communication cost, the use of the integrated IP infrastructure, participating in a multimedia application. VoIP can also utilize the advantages of packet-switched networks, as high network utilization while keeping the quality of circuit-switched networks. However, the best-effort nature of IP networks can not guarantee the requirements of delay sensitive voice traffic. VoIP network management is needed to support Quality of service (QoS) for VoIP applications. QoS requirements for voice are more stringent than for the other applications. Voice is a real-time application. Speech, to be intelligible, needs special treatment. So QoS standards for VoIP traffic are set for voice. The design and performance analysis of VoIP QoS techniques requires adequate traffic models which is the target of this thesis. Although good traffic models for unitary VoIP applications do exist, the simulation of thousands of connections is very slow making simulation studies very heavy. Aggregate traffic models are needed to achieve faster simulations without losing accuracy. In this thesis, we characterize the superposition of VoIP applications. There have been many attempts to model a single voice over IP stream and their aggregation and there are also well explored theories on circuit switched telephone networks. This thesis wants to expose a way, how to combine these approaches to have a proper model for aggregate voice over IP traffic on a network. Aggregate traffic models are essential in performance evaluation studies on large scale networks. However, simple aggregate traffic models are needed to enhance simulation efficiency and allow analytical QoS evaluation. This thesis, presents a characterization study of VoIP aggregated traffic. In this thesis, we propose a method to simulate the aggregated traffic of a number of On-Off traffic sources by a reduced equivalent set that will produce the same traffic characteristics. On-Off traffic source models are used extensively to model the source traffic in the Internet. However, when traffic from a large number of sources is to be simulated, it becomes computationally prohibitive to generate each source separately. In such a case, a reduced set of aggregated models would be needed, if it can generate the same traffic intensity and traffic characteristics. An approach based on matching the first and second order statistics of the original aggregate to reduce the number of traffic generators is presented in this thesis, with some simulation results. Keywords: VoIP Traffic, Aggregate Traffic Modeling, Traffic Modeling, On-Off Source
