A real-time distributed computing mechanism for P300 speller BCI

Abstract

Among the diverse paradigms of Brain-Computer Interface (BCI), P300 Speller is underlined by its reliability and stability. However, the Information Transfer Rate (ITR) of P300 Speller is low. This paper proposes a real-time distributed computing mechanism based on Storm for P300 Speller. This mechanism can reduce the time of processing signals, building feature vectors and classifying them for P300 Speller, so that it could help improve the ITR of P300 Speller. This mechanism, built on Storm, includes electroencephalogram (EEG) data segmentation strategy, parallel feature extraction strategy, parallel classification strategy and classification synthesization strategy. The experiments showed that the algorithm for P300 Speller could be computed faster on this mechanism than it is done without this mechanism and ITR of P300 Speller could be improved significantly by this mechanism.