On Improvement of Independent Wavelet Models to Heterogeneous Network Traffic

In our previous work, we showed empirically that independent (Haar) wavelet models were parsimonious, computationally efficient and accurate in modeling heterogeneous network traffic measured by both auto-covariance functions and buffer loss rate. We also proved analytically that such models were capable of capturing any decay rate of auto-covariance functions at large lags. But the simplicity of independent Haar wavelet models also results in deviations from the actual auto-covariance functions in small lags which should be improved. In this work, we focus on improving independent (Haar) wavelet models using two approaches. One still uses independent wavelet coefficients but more complex wavelet basis with higher vanishing moments. The other still uses Haar wavelet basis but incorporates major dependence among wavelet coefficients using (low-order) Markov models across time scales. Both approaches are measurement-based with parameters that can be readily estimated using traces of heterogeneous traffic. The performance in approximating auto-covariance functions at small lags is improved by both approaches at a moderate computational cost, which is O(N ) for a trace of length N .