Capacity allocation for voice over IP networks using maximum waiting time models

Abstract

As voice services impose stringent quality of service (QoS) guarantees to perform well over IP networks, large network resources should be allocated to their traffic class. It gets unaffordable when hard guarantees are required as in deterministic-based mechanisms such as the guaranteed services model of the integrated services (IntServ) architecture. However, the amount of network resources could be drastically decreased if only a small number of all voice connections are allowed to be negatively affected. In this work, a new capacity allocation method based on the maximum waiting time model is explored. It is established from the following concept: by providing statistical quality guarantees to those packets that experience the maximum waiting time among all packets of the active voice connections, all other packets are implicitly protected from excess delay and, thus, from service degradation. This method is investigated and mathematically analyzed for the voice service class in converged IP networks.