The analysis and modeling of queues forming and processing in the network devices

Authors: 

M.I. Kyryk, N.M. Pleskanka, D.V. Kozhurov

Lviv Polytechnic National University

The investigation has covered buffer resource and queue management as main and the most important network traffic parameter and access control and overload mechanism. Network devices interaction model can performed by three levels:
-    physical layer;
-    protocols layer;
-    queues consolidating two previous levels.
The basic mechanisms of processing queues have been examined. The rules, that packages in queue are processed according to, are called queues management algorithms. The requirements for processing queues mechanisms have been defined. The formation and processing network queues analysis and modeling mechanisms have been presented. After reading the main points of the mechanisms of formation and queuing, simulation model of queuing in multiservice network using data prioritization has been developed. The model has been implemented in a Matlab software environment.
The number of packages in the buffer has been calculated according to Norros formula. Every algorithm evaluation creterion is the ability to best quality service for each service providing the best parameters QoS. Simulation results comparative analysis has been carried out. For more detailed analysis of the service parameters quality, the results of different types algorithms for multiservice network are presented. QoS parameters, such as the probability of packet loss, delay and jitter, were determined for different flows with appropriate algorithm. It was determined that the FIFO algorithm is unacceptable for service data. PQ algorithm was designed to process high priority flows. The best simulations results showed WFQ algorithm, which can satisfy acceptable service quality according to all parameters.
The presented model makes it possible to choose the optimal queues algorithms for different services. This approach can be used on design stage, and also on increasing the network efficiency stage.