Drowsiness Detection has become one of the key factors of traffic accidents recently which can result in death, money loss or serious physical loss. Much research has been done to overcome this problem. Among the solution, by using physiological measures, especially EEG signal provides the most satisfied result in accuracy. In this study, we investigated various EEG signal augmentation techniques to enhance the training process of CNN drowsiness detection. We explore tons of data augmentation and evaluate through their performance through accuracy. Our investigation will handle data augmentation strategies that can significantly improve CNN performance, reducing overfitting and enhancing model adaptability to unseen data. In the end of the result, scaling data scores the best performance among the data augmentation technique. This research will help upcoming research to do baseline assessment and understanding about EEG works in drowsiness detection. By giving a comprehensive evaluation of augmentation methods, this study contributes to the development of more reliable and accurate drowsiness detection systems using EEG data when working with data augmentation.