Analysis of Extracting Distinct Functional Components of P300 using Wavelet Transform Mandeep Kaur 1 , P. Ahmed 2 , M. Qasim Rafiq 3 1 Assistant Professor, School of Computer Science, Lingaya‟s University, Faridabad, Haryana, India, mandeephanzra@gmail.com 2 Professor, School of Computer Science & Engg., Sharda University, Greater Noida, U.P., India, pervez.ahmed@rediffmail.com 3 Professor, Dept. Computer Sc. & Engg. AMU, Aligarh, mqrafiq@hotmail.com Abstract: - This paper investigates P300 features extracted through wavelet transform for BCI systems. Feature extraction is one of the key issues of signal processing for P300 based brain-computer interface systems (BCI). This paper examines and highlights the significance of using wavelets in P300 based BCI systems. We also mention various methods of feature extraction from P300 signals. The analysis suggests that wavelet transform is the best-suited tool for non-stationary signals like P300 signals. Key-Words: - P300 signal, Brain-Computer Interface, Fourier Transform, Wavelet Transform, Short Term Fourier Transform, Feature Extraction. 1 Introduction Research have proved the utility of EEG based Brain Computer Interface Systems (BCIs) for unblessed people [1] [2]. The electroencephalogram (EEG) signal when recorded for a particular stimulus, called as evoked related potential (ERP). The most popular evoked related potential (ERP) signal is P300, acquired from the central-parietal region of the brain [3]. These signals are non- stationary in nature and do not allow the accurate retrieval of the signal frequency information. These signals reflect only time-domain information i.e. Time-Amplitude representation. Therefore, to extract frequency information, signal transformations are required. The common techniques for this are Fourier Transform (FT), Short-Time-Fourier Transform (STFT), Wavelet Transform (WT); Continuous Wavelet Transform (CWT) and Discrete Wavelet Transform (DWT). The most common technique is Fourier Transform (FT), introduced by J.Fourier. Nevertheless, it provides only frequency domain information of the signals, minus time domain information. Therefore, it is merely used for the signals whose frequency content does not change in time called stationary signals [4] [5]. Fourier Transform transformation technique shows any periodic function as an infinite sum of periodic complex exponential function defined by the following two equations [5]: where x denotes signal, t denotes time, f denotes frequency, X(f) denotes signal in frequency domain i.e. FT of x(t), x(t) denotes signal in time domain i.e. inverse FT of X(f) and means integral correspond to all time instances and treated as window function of FT. The limitation of Fourier analysis is the loss of brief information in the frequency domain. Hence, it cannot use for non-stationary signals whose frequency content varies over the time. Electroencephalography (EEG), Electromyography (EMG), Electrocardiography (ECG) signals etc. are non-stationary in nature. The time-frequency Mathematical Models in Engineering and Computer Science ISBN: 978-1-61804-194-4 57