The Effect of Aggregation on Self-similar Traffic

Traffic aggregation is a basic networking operation that merges separate traffic streams in network switching and cross-connect nodes. This paper discusses the experimental results on a generalized and protocol-independent self-similar traffic generator based on the RPG model of OPNET. The aggregation processes results from merging of multiple traffic traces with various Hurst parameters, where a large set of synthesized self-similar traffic traces have been used. We perform a wavelet-based analysis to investigate the behavior of these aggregated processes. The main conclusion is that the aggregated process still exhibits self-similarity, with Hurst parameter close to the largest one among the underlying separate source traffic.