High Throughput Ultrasonic Multi-implant Readout Using a Machine-Learning Assisted CDMA Receiver.

Untethered, wireless peripheral nerve recording for prosthetic control requires multi-implant communications at high data rates. This work presents a multiple-access ultrasonic uplink data communication channel comprised of 4 free-floating implants and a single-element external transducer. Using code-division multiple access (CDMA), overall channel data rates of up to 784 kbps were measured, and a machine-learning assisted decoder improved BER by >100x. Compared with prior art, this work incorporates the largest number of implants at the highest data rate and spectral efficiency reported.