Stochastically bounded burstiness for communication networks

A network calculus is developed for processes whose burstiness is stochasti cally hounded by general decreasing functions, This calculus is useful for a large class of input processes, including important processes exhibiting "subexponentially bounded burstiness" such as fractional Brownian motion. M oreover, it allows judicious capture of the salient features of real-time t raffic, such as the "cell" and "burst" characteristics of multiplexed traff ic. This accurate characterization is achieved by setting the bounding func tion as a sum of exponentials.