A Hardware/Software Prototype of EEG-based BCI System for Home Device Control

An Efficient Hardware Implementation for a Motor Imagery Brain Computer Interface System

“…Overall results of the algorithm (for complete number of classes) are listed in Table 1, showing better performance than the best reported results of BCI competitions [54,55] and also recent state-of-the-art papers [31,48] for both datasets (67.2% for BCI competition IV, dataset 2a; 73.54% for BCI competition III, dataset V). Moreover, the results show a high performance of 81.9% for captured signals, re ecting the excellent performance of the proposed algorithm.…”

“…However, the main shortcoming of these methods is that they provide low accuracies and are not suitable for hardware implementation as they have high computational complexity. However, the methods proposed in [31] and [48], mentioned previously, outperform the best methods reported in [54] and [55], and can be considered as two of the best performing methods on BCI competition IV-dataset 2a. Nonetheless, it should be noticed that these two methods only classify the rst two classes (movements of right and left hands).…”

“…Several recent state-of-the-art papers working on motor imagery (such as [31,48]) have considered merely two classes of right-and left-hand movements for experiments. To compare the proposed method with similar state-of-the-art methods, we evaluate this work with two mentioned classes, too.…”

“…This paper proposes a more e cient hardware implementation with less hardware resources compared to the design in [47]. A more complete work of [47] is reported in [48]. A novel approach is introduced in [48], de ning the tasks and working with devices by a state machine.…”

“…A more complete work of [47] is reported in [48]. A novel approach is introduced in [48], de ning the tasks and working with devices by a state machine. The user can do the task or switch to other tasks by thinking about right and left hand movements.…”