Exploring the Feasibility of Single-Frequency Multi-Target SSVEP-Based BCI for Online Control

Brain-computer interface (BCI) systems based on steady-state visual evoked potential (SSVEP) have gained widespread adoption due to their efficiency and accuracy. However, the traditional SSVEP method suffers from limitations such as visual fatigue and interference between different stimuli. To address these issues, this paper proposes a novel paradigm and classification algorithm for single-frequency multi-target SSVEP-based BCIs. The proposed approach allows for enhanced instruction encoding using fewer flicker blocks. In this study, electroencephalograph (EEG) signals were recorded during steady-state visual stimulation at 16 locations within the human visual field through carefully designed EEG experiments. Feature extraction was performed using typical correlation analysis, revealing a decrease in the evoked effect of SSVEP with increasing eccentricity of the stimulus block relative to the center of the visual field. After that, data with varying eccentricities were classified using a support vector machine (SVM) based on the Riemann kernel. The classification accuracy exhibited a trend of initial increase followed by decrease as eccentricity increased, enabling identification of the optimal target location for online BCI. Finally, the optimal time window length for online BCI was determined by evaluating the information transmission rate (ITR).