Robust, Reconfigurable, and Power-Efficient Electrophysiological Recording Systems
Nov 19, 2013
from 10:00 AM to 12:00 PM
|Where||Engr. IV Bldg., Tesla Room 53-125|
|Contact Name||Vaibhav Karkare|
|Add event to calendar||
Advisor: Prof. Dejan Marković
Recording of electrophysiological signals in a remote, uncontrolled environment requires the signal chain to be robust to saturation by large, non-stationary interferes. Conventional recording front-ends incorporate a high voltage-to-voltage gain and saturate for 5-mVp-p interferes. In this work, the voltage-to-phase conversion inherent to the operation of VCO-based ADCs is leveraged to directly digitize the input, providing saturation tolerance for 200-mVp-p interferes. The front-end is reconfigurable and can support (potentially simultaneous) recording of various biosignals. A prototype 8-channel recording chip, implemented in 65-nm CMOS, and a commercial RF transceiver are assembled onto a nickel-sized wireless biosignal-recording mote.
Wireless transmission of raw waveforms becomes infeasible for multi-channel neural action potential recordings due to the high data rates required for transmission. On-chip processing of the recorded data not only enables real-time processing but also provides output data-rate reduction. In this work, we use algorithm-level modifications and careful architecture-level choices to implement power-efficient ASICs for spike sorting—a common processing step for action potential waveforms. A prototype 16-channel spike-sorting chip, implemented in 65-nm CMOS, consumes 75 μW of power to provide more than 200-times reduction in the output data rate.
Vaibhav Karkare received a M.Sc. in Physics and B.E. in E.E. from B.I.T.S., Pilani in 2006 and a M.S. in E.E. from UCLA in 2009. His research interests include the design of recording and processing circuits for wireless sensor interfaces.