In this paper, we show the validity of a quite simple resource allocation law, which is not restricted to one time scale. Besides the fact that this law describes the basic behaviour of resource allocation for sources with variable bit rates, it offers savings of time and memory in practice when systems are to be dimensioned and routing has to be performed, and particularly in dynamic resource allocation with real-time requirements. The law is given as a function of traffc demand and consists of a constant, a linear and a power term. Its coeffcients are in influenced by traffic and system characteristics as well as by QoS requirements and may be determined by a Genetic Algorithm. Many examples show the universal character of this law for sources with variable bit rates.

Ett resursallokeringsformel presenteras som tar hänsyn till olika tidsskalor. Genetiska algoritmer används för att ta fram koefficienterna. resursallokering i datanätverk, förbindelse, tjänstekvalitet, dimensionering, approximationsformel, genetiska algoritmer