A concept of sub-bands event related potentials to increase classes of brain computer interface system

Event related potential (ERP) based brain computer interfacing (BCI) achieves significant stability. Changes in electroencephalogram (EEG) signals related to various tasks have been significantly observed. However, each ERP related to particular task can be only exploited as one-to-one relation with specific command or operation. This limits the variability of BCI system and increases the amount of work to identify task related accurate pattern changes in EEG. In this paper, sub-band analysis of detected ERP is proposed in order to factorise one-to-one relation into one-to-many for increasing the variability of BCI system. First, the hypothesis based on analysis of event related spectral perturbation (ERSP) is stated, and then the hypothetical concept is generalised with sub-bands decomposition of ERP, followed by culminative power estimation. Results show that the proposed technique can be easily implemented as a method of combined factorised feature extraction (CFFE) to execute multiple commands from single ERP.