Finite-State VoIP Traffic Model for Variable Voice Payload Size

Telephone calls have an essential function as a communication tool during emergencies in putting people's minds at ease because they enable them to hear others' voices in real time. However, such events lead to increased telephone-call traffic and damage to communication equipment, resulting in limited available communication resources. Therefore, we need VoIP admission control for emergencies that effectively utilize these limited resources. In our VoIP control system, we assign the optimum sending parameters to each VoIP session, considering the available communication resources, the distance between users, and the users' characteristics. These parameters, which include voice payload size (VPS), affect the speech quality, sending bitrate, and delay. We maximize the number of VoIP sessions accommodated in networks by selecting the optimum parameters. In this paper, to implement such admission control, we propose a traffic model for a single VoIP session, which enables the VPS selection from multiples of 5 ms and unifies variable VPS traffic models. We also propose a method for deriving the proposed model from an existing traffic model. Furthermore, we show the VPS-selection range calculated by the proposed model in terms of delay and average bitrate.