Quality of Service measurement using artificial neural networks
Recently, the areas of Quality of Service (QoS) have witnessed vast progress mainly due to the appearance of new multimedia networking and computing. Parts of these advances are the QoS measurement, monitoring and analysis which have long been of interest to the networking community. Every multimedia application has its own parameters which have an effect on its overall QoS. In this paper, a new neural network QoS assessment approach has been developed in order to evaluate the network performance and QoS of these multimedia applications. The proposed approach takes into account the QoS parameters and requirements of each application. This allows objectively measuring some parameters like (delay, jitter and packet loss) and comparing them with required QoS parameters values. Then, these measurements are used as input to a fuzzy logic system to produce an output that represents the instantaneous QoS. The measured inputs and the assessed output will then be used to train a neural network to be utilized as a QoS monitoring system. This approach showed that the QoS can be assessed without the necessity for complicated analytical models. Based on the devised system, the measured QoS can be used to optimize the received quality of the multimedia services along with the changing network conditions and to manage the utilization of the network available resources. Overall findings of this study contribute to a method for drawing a realistic picture of the multimedia networks QoS and provide a firm basis and useful insights on how to effectively design future QoS solutions for routing, call admission control, and so on.