Adaptive strategy for time window length in SSVEP-based brain-computer interface

Most of the existed methods for steady-state visual evoked potential (SSVEP) recognition require a specified time window length (TWL) to estimate the dominant frequency components in EEG signals. Typically, the TWL is manually predetermined and then fixed during running of the SSVEP-based brain-computer interface (BCI), which may not give the optimal information transfer rate (ITR). This study proposes an adaptive strategy and integrates it into multivariate synchronization index for SSVEP recognition with the automatically determined TWL. The optimal TWL is adaptively selected according to the estimated synchronization index, which provides the relatively higher ITR for the SSVEP recognition than a fixed TWL does. Experimental results evaluated on nine healthy subjects demonstrated effectiveness of the proposed adaptive strategy for the TWL in SSVEP-based BCI.