Paper Title
Frequency Recognition in SSVEP-based BCI using a Hybrid of Canonical Correlation Analysis and Power Spectral Density Analysis

Abstract
Brain-computer interface (BCI) provides a direct communication channel between human brain and external devices. Steady-state visual evoked potential (SSVEP) appears over occipital area of a brain when a subject is looking at repetitive visual stimuli with a specific frequency. Canonical correlation analysis (CCA) estimates the stimulus frequency from canonical correlations of pairs of canonical variables. Though CCA has been successfully applied to SSVEP recognition for BCI application, it makes an incorrect decision in case of using a short time window length (TW) because canonical variable of recorded signal may not match that of reference signals composed of sine-cosine waves. However, power spectrum of canonical variables has the largest sum of power at stimulus frequency band even with a short TW, comparing that of other frequencies. From this phenomenon, we propose a hybrid of CCA and PSDA. The proposed method uses power spectrum of canonical variable instead of canonical correlation. Experimental results show that the performance of the proposed algorithm is better than that of CCA when they have the same TW. Keywords- Brain-computer interface (BCI), Electroencephalography (EEG), Steady-state visual evoked potential (SSVEP), Canonical correlation analysis (CCA), Power spectral density analysis (PSDA)