Traac model based on the fractional Brownian motion fBm contains three parameters: the mean rate m, variance parameter a and the Hurst parameter H. The estimation of these parameters by the maximum likelihood ML method is studied. Explicit expressions for the ML estimates ^ m and ^ a in terms of H are given, as well as the expression for the log-likelihood function from which the estimate ^ H is obtained as the minimizing argument. A geometric sequence of sampling points, t i = i , i s i n troduced in order to see the scaling behaviour of the traac with fewer samples. It is shown that by a proper`descaling' the traac process is stationary on this grid leading to a Toeplitz-type covariance matrix. Approximations for the inverted covariance matrix and its determinant are introduced. The accuracy of the estimation algorithm is studied by simulations. Comparisons with corresponding estimates obtained with linear grid show that the geometrical sampling indeed improves the accuracy of the estimate ^ H with a given number of samples.
[1]
Ugo Mocci,et al.
Broadband network traffic [i.e. teletraffic] : performance evaluation and design of broadband multiservice networks : final report of action COST 242
,
1996
.
[2]
B. Ninness.
Maximum likelihood estimation of the parameters of fractional Brownian motions
,
1995,
Proceedings of 1995 34th IEEE Conference on Decision and Control.
[3]
Ilkka Norros,et al.
A storage model with self-similar input
,
1994,
Queueing Syst. Theory Appl..
[4]
Jan Beran,et al.
Statistics for long-memory processes
,
1994
.
[5]
Mohamed A. Deriche,et al.
Maximum likelihood estimation of the parameters of discrete fractionally differenced Gaussian noise process
,
1993,
IEEE Trans. Signal Process..
[6]
Audra E. Kosh,et al.
Linear Algebra and its Applications
,
1992
.