Determining long memory processes parameters based on multi-scale maximum likelihood estimation

In practical, there exists a problem in traditional maximum likelihood estimation (TMLE) that is the great computational burden. Based on the decorrelation property of discrete wavelet transform (DWT), we propose and evaluate multi-scale maximum likelihoods estimation (MMLE), and apply it to a kind of long memory processes with broad application background. Simulation results show that under some precision demands, MMLE reduced the computational complexity greatly and can be used as an alternative of parameter estimation method.