.png){kind=icon}
.png){kind=icon}
Return
Edge AI-Brain-Computer Interfaces System: A Survey
.png){kind=icon}
PRE
AI
DOI：10.1109/TNSRE.2025.3618688.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Edge artificial intelligence (Edge AI) has emerged as a transformative paradigm for enhancing the performance, portability, and autonomy of brain-computer interface (BCI) systems. By integrating advanced AI capabilities directly into electroencephalography (EEG)-based devices, Edge AI enables real-time signal processing, reduces dependence on external computational resources, and improves data privacy. However, deploying AI on resource-constrained hardware introduces challenges related to computational capacity, power consumption, and system latency. This survey provides a comprehensive examination of Edge AI-enabled BCI systems, covering the full pipeline from EEG hardware specifications and on-device data acquisition to signal preprocessing techniques and lightweight deep learning models optimized for embedded platforms. We review existing frameworks, specialized hardware accelerators, and energy-efficient AI approaches that facilitate real-time BCI processing at the edge. Furthermore, the paper reviews state-of-the-art solutions, examines key technical challenges, and outlines future research directions in hardware-software co-design and application development. This work aims to serve as a reference for researchers and practitioners seeking to design efficient, portable, and practical Edge AI-powered BCI systems.
Keywords:
Electroencephalography
Artificial intelligence
Signal resolution
Edge AI
Hardware
Real-time systems
Quantization (signal)
Brain-computer interfaces
Surveys
Spatial resolution
Brain-computer interfaces (BCIs)
electroencephalography (EEG)
edge artificial intelligence (Edge AI)
embedded systems
deep learning
real-time processing
on-device AI
neural signal processing
Journal
IF:
5.2
Papers:
449
Citations:
1.6W