Improved algorithm for eeg – based bci application

Brain Computer Interface enables individuals to communicate with devices through Electroencephalography (EEG) signals in many applications that use brainwave controlled units. New algorithms using EEG waves for controlling the movements of a drone by eye-blinking and attention level signals are presented. Optimization of the signal recognition obtained is carried out by classifying the eye-blinking with a Support Vector Machine algorithm and converting it into 4-bit codes via an artificial neural network. Linear Regression Method is used to categorize the attention level to low or high level with a dynamic threshold, yielding a 1-bit code. The control of the motions in the algorithm is structured with two control layers. The first layer provides control with eye-blink signals, the second layer with both eye-blink and sensed attention levels. EEG signals are extracted and processed using a single channel NeuroSky mindwave 2 device. The proposed algorithms have been validated by experimental testing of five individuals of different ages. The results show its high performance compared to existing algorithms with accuracies of 91.85% and 90.37% for 9 control commands. With a capability of up to 16 commands and its high accuracy, the algorithms can be suitable for many applications.