On the self-similar nature of Ethernet traffic
SIGCOMM '93 Conference proceedings on Communications architectures, protocols and applications
On the self-similar nature of Ethernet traffic (extended version)
IEEE/ACM Transactions on Networking (TON)
Analysis, modeling and generation of self-similar VBR video traffic
SIGCOMM '94 Proceedings of the conference on Communications architectures, protocols and applications
Wide area traffic: the failure of Poisson modeling
IEEE/ACM Transactions on Networking (TON)
Experimental queueing analysis with long-range dependent packet traffic
IEEE/ACM Transactions on Networking (TON)
Self-similarity in World Wide Web traffic: evidence and possible causes
IEEE/ACM Transactions on Networking (TON)
On multimedia networks: self-similar traffic and network performance
IEEE Communications Magazine
On the use of fractional Brownian motion in the theory of connectionless networks
IEEE Journal on Selected Areas in Communications
Evaluation and estimation of second-order self-similar network traffic
Computer Communications
| Hi-index | 0.24 |
Recent measurement studies have revealed that multimedia network traffic has complex statistical characteristics. They have presented convincing evidence that the traffic exhibits self-similarity in nature. Generating self-similar traffic traces is increasingly important to research on characterization of self-similar processes and their impacts on network performance. However, self-similar traffic traces are difficult to be generated due to their long-range dependence. No matter fractional Gaussian noise model or fractional autoregressive integrated motion average model suffers from that computation efforts, they are proportional to the length of the traces that we want to generate. In this paper, we propose two new trace-generating schemes, whose computation efforts are independent of the length of the trace: autoregressive Gaussian processes with trends and autoregressive Gaussian processes with aggregations. The results show that it is effective to generate the long-range dependence with these two schemes. The autocorrelations and queuing performance of traces generated from these two approaches are close to those of the empirical trace.