SESSION 15 Neural Interfaces and Edge Intelligence.pdf

1、ISSCC 2025SESSION 15Neural Interfaces and Edge Intelligence for Medical Devices15.1:A 3.9mW 200words/min Neural Signal Processor in Speech Decoding for Brain-Machine Interface 2025 IEEE International Solid-State Circuits Conference1 of 37A 3.9mW 200words/min Neural Signal Processor in Speech Decodin

2、g for Brain-Machine InterfaceTun-Yu Chang,Jeng-Bang Wang,Yu-Hsuan Tsai,Chia-Hsiang YangNational Taiwan University15.1:A 3.9mW 200words/min Neural Signal Processor in Speech Decoding for Brain-Machine Interface 2025 IEEE International Solid-State Circuits Conference2 of 37Outline Introduction Speech-

3、Based Brain-Machine Interface Algorithm Algorithm-Architecture Co-Optimization System Architecture Experimental Verification Summary15.1:A 3.9mW 200words/min Neural Signal Processor in Speech Decoding for Brain-Machine Interface 2025 IEEE International Solid-State Circuits Conference3 of 37Introduct

4、ion Brain-machine interface(BMI)establishes a direct communication pathway between brain and machine Applications of BMIVR/AR user interface,neural prosthesis,machine control,etc.VR/AR User InterfaceNeural ProsthesisMachine Control15.1:A 3.9mW 200words/min Neural Signal Processor in Speech Decoding

5、for Brain-Machine Interface 2025 IEEE International Solid-State Circuits Conference4 of 37Visual-Stimulation-Based BMI Algorithm 1:user observes flickering targetsDecodes through correlation between neural signal&reference frequency Visual-stimulation-based processor 4Insufficient communication rate

6、(4.7 words/min),external stimulus required 4 W.Byun,Symp.VLSI Circuits,20211 M.Nakanishi,IEEE TBME,2018Reference FrequencyFlickeringCharacterhelof1f2f3fnNeural Signal.with maximum correlationOutputCharacter hCommunication Rate 020406080100120140160180200Visual-Stimulation-Based15.1:A 3.9mW 200words/