MODELING NETWORK PROFILES

Large deviation(1) and fractal geometry(47) theories are used in this paper to build a new mathematical model based on fractals for network traffic profiles. This results in a sizing formula which gives the eff ective link or node capacity required on the network for accommodating a given traffic profile on the basis of available statistics. Statist ical analyses on the basis of spectral density theory confirm that SIT A's network traffic has a fractal quality. Traffic profiles are simul ated on the basis of fractional Brownian motion; this confirms the val idity of the sizing formula. Practical methods are provided for measur ing the quantities used in the study.