A large dataset for VEP based brain-computer interfaces employing narrow band code modulation and frequency-phase modulation

Brain-computer interfaces (BCIs) realize the information transmission between the brain and the external world. Visual evoked potential (VEP) based BCIs have gained widespread attention in the field of multi-instruction interaction attributed to their high information transfer rate (ITR). However, improving the ITR and the practicability is challenging for existing VEP based BCIs due to factors such as the encoding efficiency and the calibration time. Therefore, to address this issue, this study proposed a new encoding method employing narrow band random sequences and provided a large dataset for VEP based brain-computer interfaces. Narrow band random sequences are random sequences with a specific frequency band. The dataset encompasses three paradigms that employ three kinds of encoding sequences: narrow band sequences with a frequency band of 15~25 Hz (NBRS-15), narrow band random sequences with a frequency band of 8~16 Hz (NBRS-8), and sequences utilizing joint frequency-phase modulation method with a frequency range of 8-15.8 Hz (JFPM-8). The dataset includes 59-channel electroencephalogram data for 100 subjects, and the quality of the dataset is validated through quantitative analyses on EEG characteristics and classification performance. The proposed dataset is recommended for the development of classification algorithms for VEP based BCIs, and for exploring the potential of code-modulated VEP based BCIs employing narrow band random sequences.