Abstarct: In this thesis, the implementation of an acceptable method for detecting sleepiness in car drivers on the DSP processors has been investigated. For this purpose a databaxse consisting of 20 male volunteers with a mean age of 24 years has been created. For preparing the databaxse they were asked to drive with a car drive simulator in a terms of vigilance and sleepiness conditions. For evaluating the databaxse three method consists of wavelet transform, wavelet packet transform and matching pursuit have been used which according to our knowledge the last one is used for the first time for feature extraction from EEG signals. After that, SVM classifier has been used to classify the acquired features. The results show that when a person becomes sleepy, energy of wavelet coefficients corresponding to Delta and Teta frequency bands increased considerably, but amount of increment depends on each person and it is not equal for all of them. Moreover in order to implement the proposed method, at first a suitable hardware was designed and build to receive two EEG channels from brain signals which its salient features are portability and low power consumption (24 mW). Then, for processing the preprocessed and converted EEG signals, a hardware baxsed on TMS320C5509A processor, was designed and build too. This system is also portable and could work with batteries. The results show that estimating state of individual consciousness can be done by choosing an appropriate threshold for their coefficients energy in Delta and Teta frequency band of each person with acceptable recognition accuracy.