Neural Signal Processing using Discrete Wavelet Transform for Neural Interfaces

This paper presents a neural signal processing ASIC based on Discrete Wavelet Transform (DWT). The recorded neural signals from 256 channels are analyzed by fast DWT algorithm with special ALUs, then, compressed by run-length encoders (RLE). The processed data are delivered through RF links and reconstructed in a host receiver. This design operates at 200
MHz
clock with 2.5V and was implemented with TSMC 0.25
μ
m technology. Recorded neural data test shows 1:89:3 (1.12%) compression rate and perfect in-band noise rejection.