Neural-Driven Smart IoT Systems: Enhancing Brain-Computer Interface Control Using Deep Learning and Edge Computing

Abstract

The rapid development of brain-computer interface technology, alongside the growth of the IoT, has favoured new domains of human-machine interactions in health, rehabilitation, and assistive technologies. However, many of the classical BCI systems face issues while dealing with performance, latency, and scalability issues. To deal with the aforementioned, the paper proposes to bring together deep learning and edge computing to support BCI control in IoT smart systems. The paper aims to reduce latency and improve performance and usability in BCIs. Deep learning algorithms for efficient signal classification include convolutional neural networks without edge computing, and sensor-to-sensor processing happens as close to the object being controlled as possible in real-time, which means less dependence on remote cloud servers. Combining these two technologies into one approach optimizes BCI-controlled interactions with IoT devices such as prosthetics and exoskeletons. The results revealed that the proposed model demonstrated an improvement over traditional methods based on several key metrics: accuracy (96%), latency (90 ms), throughput (90 signals/sec), and power consumption (50 W). With this model, the latency reduced with this integrated edge computing will improve energy efficiency, making it suitable for real-time efficient applications. Therefore, the integration of deep learning with edge computing in BCIs shows high value for upgrading health care, rehabilitation, and assistive technologies. Finding novel ways to be more efficient and scalable concerning actual IoT applications completes the picture of an integrated system for more personalized, responsive, and precise control over BCI systems.