Exploring the Effects of Different Windowing Methods on Asynchronous SSVEP-Based Brain-Computer Interface

In the field of asynchronous brain-computer interfaces, various algorithms and preprocessing methods are used to improve the performance of asynchronous control. However, few studies have explored the effects of different windowing methods. This study examines the impact of windowing methods on the performance of a 4-class asynchronous SSVEP brain-computer interface. We collected data from 7 subjects and conducted asynchronous SSVEP offline data analysis to verify the sequence and shape effects of windowing. The results show that while keeping the FPR almost unchanged, using weighted average windows before CCA provided a 20% faster reaction time; meanwhile, using exponential decay windows significantly shortened the reaction time by 6-8%. Therefore, compared to the worst window method, using the optimal window provides a significant performance improvement of 26%. This shows that choosing a suitable windowing method in an asynchronous brain-computer interface can effectively improve asynchronous performance in terms of reaction time.