Skip to Main Content
The performance of an individual flow that belongs to a DiffServ class aggregate is an actual issue when the flow refers to QoS-demanding applications, like Voice over IP. However, through a careful choice of a control flow, a performance inference about flow aggregate individuals can be achieved. Computer simulations with synthetic VoIP traffic allow us to establish a method that relates delay and delay jitter metrics of a reference flow to infer about individual performance of flows belonging to the same DiffServ class aggregate. The evaluation was carried out for homogeneous flow aggregates (flows generated by the same voice codec) as well as non-homogeneous (several codecs). Extensive simulations and statistical analysis allowed us to conclude about the control flow characterization, its accuracy to represent the performance of individual flows of the aggregate and the impact of the VoIP traffic models over the achieved results.