Efficient parameter selection for use of self-similarity in real time resource management

Investigations of high-resolution Ethernet local area network (LAN) traces and VBR video traffic has demonstrated that these traffic types exhibit self-similarity. In this paper, we study two commonly used methods in estimating traffic self-similarity level, and compare them in terms of their applicability in providing real-time feedback to resource management algorithms. These methods are variance-time plot and wavelet-analysis. The constraints for selection of a method, wavelet filter types and lengths, number of scales and length of traffic data to be stored and analyzed are given.