Research On Chirp-VEP-based Online Brain Computer Interface System

Brain computer interface (BCI) is a communication system that directly link brainand outward environment without through human’s normal neural pathway. AmongBCIs, steady state visual evoked potential (SSVEP) is widely researched and appliedbecause of its high signal-to-noise rate and fast response. However, its drawback ofsmall command set is also obvious due to the limited EEG frequency band and theharmonic effects of SSVEP.Considering the characters of visual evoked potential and limitations of SSVEP,this thesis introduced a novel visual evoked response paradigm based on Chirp stimuli,which could induce a new signal called chirp stimuli visual evoked potential(Chirp-VEP).To study BCI based on Chirp-VEP, a visual stimuli system based onFPGA was firstly developed with the functions of multiple parameters adjustment andonline use. Secondly, short-time Fourier transform, wavelet transform, Wigner-Villedistribution and Chirplet transform were adopted for time-frequency analysis ofChirp-VEP. Thirdly, the optimization of evoked experiments and verification ofoffline experiments were conducted to find the best flick parameter for inducingrobust Chirp-VEP and simulate the condition of online use for the establishment ofonline system. Finally, a complete online system was built up according to above-mentioned results and tested with relevant task experiments.The experiment results indicate that,(1) the FPGA-based visual stimuli systemdeveloped in this study could meet the research requirements of Chirp-VEP withmulti-commands and provided the hardware basis for its further development;(2)through the analysis of time-frequency characters and classification result, it’s foundthat the robust Chirp-VEP can be induced in13-19Hz frequency band of EEG withChirp rate in the range of0.1to1Hz/s;(3) The results of online test show that, withthe proposed system, the accuracy could achieve an average of97.8%to recognize4chirp rates of the same base frequency, and90%to recognize multiple chirp rates ofmultiple base frequencies (totally10stimulation commands). In summary,Chirp-VEP-based BCI with precise signal output, reliable performance and highaccuracy, provides a new solution for expanding command set and boosting information transport rate of BCI, which is helpful in the further development andwide use of practical BCI technique.