Order detection for dependent samples using entropy rate

Detecting the number of signals in a given number of observations, or order detection, is one of the key issues in many signal processing problems. Information theoretic criteria are widely used to estimate the order. In many applications, data does not follow the independently and identically distributed (i.i.d.) sampling assumption. Previous approaches address dependent samples by downsampling the dataset so that existing order detection methods can be used. By downsampling the data, the sample size is decreased so that the accuracy of the order estimation is degraded. In this paper, we introduce two linear mixture models with dependent samples. The likelihood for each model is developed based on the entire data set and used in an information theoretic framework to improve the order estimation performance for dependent samples. Experimental results show performance improvement using this new method.