Summary
IEICE Information and Communication Technology Forum
2016
Session Number:B1
Session:
Number:B1-02
Application of non full-availability models in the analysis of multi-service network systems
Maciej Stasiak,
pp.-
Publication Date:2016-10-01
Online ISSN:2188-5079
Summary:
The analysis of modern networks, in particular of those with the capacity to support multi-service traffic, requires a development of appropriate analytical models. One of the most important problems faced by engineers is the need to develop simple and effective models for systems which have non full availability character. Hitherto, no satisfactory solutions have been proposed that would make the evaluation of the performance parameters possible for different classes of calls with differentiated QoS requirements. Until now only results for borderline full availability cases have been obtained. These results, however, are usually of low accuracy and cannot provide useful basis for engineering methods and algorithms. In traffic theory, the problem is formulated as “non full availability models", and it is addressed adequately by researchers and engineers alike. In the Chair of Communications and Computer Networks at Poznan University of Technology, models of multi-service non full availability systems that enable the researcher to determine the values of the performance parameters for individual classes of calls have been developed. These models are based on the model of multi-service Erlang Ideal Grading (EIG), which can approximate complex non full availability network systems, such as cells in 4G and 5G networks, multi service overflow systems, multi service switching networks, multi service queuing access systems, etc. To the best knowledge of the authors, the proposed models are, in fact, the first solution to the problem of multi-service non full availability systems modelling. These models can also constitute the basis of a network systems analysis that would take into account different scenarios for the service of multi rate traffic streams.