Kasymov R. Improvements of QoS routing algorithms in IP/MPLS networks based on traffic forecast

The object is Telecommunication Networks with IP/MPLS; the aims is an improvements of routing algorithms in IP/MPLS networks by using of traffic forecast to estimate of unused network resources; the methods are theory of wavelet analysis; theory of artificial neural nerworks; theory of fuzzy logic; methods of statistical analysis; methods of multicriterion optimization, theory of the graphs; results and novelty - the method of short term forecasting based on wavelet decomposition and cascade-correlation neural networks usage is introduced; new approach for usage of adaptive neural-fuzzy interference system for early forecast quality estimation has been introduced; improved heuristic algorithm for maximum flow routing has been developed; Proposed usage of piecewise-linear penalty functions to engage the available bandwidth as a component of the cost function, which improved the efficiency of a resource utilization of network bandwidth; the field is technical sciences, telecommunications.