Statistical multiplexing of homogeneous fractional Brownian streams (Q702042)

Fractional Brownian motion has been introduced as a mathematically tractable Gaussian traffic model with parameter \( 0.5< H < 1 \) [\textit{I. Norros}, Queueing Syst. 16, 387--396 (1994; Zbl 0811.68059)]. In this paper, the supremum functional of a fractional Brownian motion with a negative linear drift is used to model the queue length of a single channel FIFO fluid queue with long range dependent input and a constant service rate. Statistical multiplexing gains are studied under various scenarios for combining the number of independent fractional Brownian motion traffic streams with the same Hurst value. The analysis reduces retransmission in ATM networks carrying long-range dependent data traffic thereby increasing information throughput.