Real-Time Voice Quality Guarantee On The IP Based Networks

Abstract

In this paper, a new proposed VoIP (Voice over Internet Protocol) model has been implemented to guarantee the required voice quality based on the available network resources. Basically, the new proposed model depends on a simple Differentiated-Services (DS) blocks at the edge of the networks [1]. Although, it is so simple from the routers point of view, it is somehow complex from the host point of view due to the input characteristics analyzing process. This model has some determined values for both the required buffers through the path and the enough reserved bandwidth to guarantee the required quality. On the other hand, the QBone (Quality of service backbone) model [2] does not analyze the input traffic (simple at the host side) but it moves the complexity to the router side due to complex queuing disciplines implementation. In fact, the QBone model does not face any success till now from the practical implementation point of view because most of the users are not welling to pay much money to have voice conversations across the Internet. So, the new proposed model not only guarantees the voice session’s quality but also saves the network resources (saves up to 74% of the consumed bandwidth more than the QBone model). In this paper, the G.723.1 (5.3)/(6.4), G.711, and G.726 voice encoders’ characteristics representing the proposed model inputs have been analyzed by implementing some sessions capturing experiments. Then the model design and three simulation scenarios assumptions, descriptions, and analytical representation have been described. Also, the proposed model results, for two different configurations, have been obtained based on the node delay, routers buffers consumptions, jitter, and packet loss. Finally, a detailed comparison between both the new proposed model and the QBone model has been presented.