TY - GEN
T1 - Kesirli brown deviniminde hurst parametresi kestirim yöntemlerinin karşilaştirmasi
AU - Baykut, Süleymam
AU - Erol, Melike
AU - Özkurt, Tolga Esat
AU - Akgül, Tayfum
PY - 2005
Y1 - 2005
N2 - Most natural processes show self-similarity. Self-similarity or 1/f behavior has been a popular concept to describe the scale invariance of time-series. Fractional Brownian motion (fBm) is one of the most preferred models of 1/f processes since statistical behavior of fBm is determined by a single (Hurst) parameter, H. Consequently, estimation of H has been an important issue and several methods are developed to estimate H. It has been a complicated issue to determine which one of the estimators yields more robust estimates. In this work, we compare the performance of three methods, namely, Higuchi, Wavelet Based and recently proposed Principal Component Analysis methods. We apply these estimators to data sets with varying H values. We also analyze the effect of data length on the robustness of the estimators. Finally, we investigate the effects of periodicity interfering with fBm which is a common case encountered in practice.
AB - Most natural processes show self-similarity. Self-similarity or 1/f behavior has been a popular concept to describe the scale invariance of time-series. Fractional Brownian motion (fBm) is one of the most preferred models of 1/f processes since statistical behavior of fBm is determined by a single (Hurst) parameter, H. Consequently, estimation of H has been an important issue and several methods are developed to estimate H. It has been a complicated issue to determine which one of the estimators yields more robust estimates. In this work, we compare the performance of three methods, namely, Higuchi, Wavelet Based and recently proposed Principal Component Analysis methods. We apply these estimators to data sets with varying H values. We also analyze the effect of data length on the robustness of the estimators. Finally, we investigate the effects of periodicity interfering with fBm which is a common case encountered in practice.
UR - http://www.scopus.com/inward/record.url?scp=33846645453&partnerID=8YFLogxK
U2 - 10.1109/SIU.2005.1567774
DO - 10.1109/SIU.2005.1567774
M3 - Konferans katkısı
AN - SCOPUS:33846645453
SN - 0780392396
SN - 9780780392397
T3 - Proceedings of the IEEE 13th Signal Processing and Communications Applications Conference, SIU 2005
SP - 672
EP - 675
BT - Proceedings of the IEEE 13th Signal Processing and Communications Applications Conference, SIU 2005
T2 - IEEE 13th Signal Processing and Communications Applications Conference, SIU 2005
Y2 - 16 May 2005 through 18 May 2005
ER -