Spatial Encoding Techniques in Time-Multiplexed Neural Recording Front-Ends

This paper reviews two spatial encoding techniques for time-multiplexed recording front-ends. These techniques use bipolar and monopolar recordings, respectively, and are aimed to compress neural data taking advantage of the large spatial correlation of LFPs/ECoGs measured from nearby electrodes. The pros and cons of both architectures are analyzed and demonstrated with experimental results from a 32-channel time-multiplexed spatially-encoded neural front-end designed in 0.18 μm technology.