Traditionally Poisson assumption and queuing analysis have been used for modelling communication traffic in packet switch networks. But recent research has revealed that teletraffic, appears as self-similarly rather than Poisson. A characteristic of self-similar traffic is its long-range dependence described by its autocorrelation function. A key step for modelling such traffic relies on a parameter called the Hurst parameter (H) in the autocorrelation function. There exist many well-known approaches for estimation of H but the various approaches may yield different values of H that model the same traffic behaviour in different ways. A lack of accurate estimation of H may result in inaccuracy for traffic management and communication delivery.

The research aims at determining the optimal estimation of H and examining a new method to prove the properties of existence and uniqueness of such optimal estimations of H, and searching for efficient algorithms to best approximate teletraffic with the optimal H.