Novel Method to Discriminate Awaking and Sleep Status in Light of the Power Spectral Density

Sleep stages are mainly classified via electroencephalogram (EEG) which involves some prominent characters not only in amplitude but also in frequency. What is more, researchers are using computer assisted analysis to acquire the panoramic view of long duration sleep EEG. However, unlike the empirical judgment, it is fairly difficult to decide the specific values of sleep rules for computer based analysis due to the fact that those values vary with individuals and different distance from reference to signal source on scalps. This paper will introduce a novel method using power spectral density to discriminate awaking, light sleep and deep sleep EEG just according to the features extracted from EEG.