Correct and efficient estimation of the Hurst parameter of long-range dependent (LRD) video traces is important in traffic analysis. The low computational cost and the wavelets' scale invariance make the wavelet transform suitable for analysis of LRD processes. In this paper, we apply wavelet-based estimation of the Hurst parameter to MPEG-1 and MPEG-4 encoded video sequences. Frequency-domain estimators (periodogram and wavelet-based) produce different Hurst parameters compared to time-domain estimators (R/S and variance-time plot). Wavelet-based estimators often produce Hurst parameters that are dose to or greater than one. Our analysis indicates that possible causes for the unreliable performance of the wavelet-based estimators are the non-stationarity of the scaling exponent and the existence of both short-range and long-range dependent components in the video traces.
展开▼
