Mixed-signal energy feature extractor of EEG frequency bands

Abstract

This paper proposes a SAR-based circuit suitable to obtain the amount of signal energy contained in EEG frequency bands. It uses a reconfigurable topology which, in a first stage, acts as a conventional data converter for the incoming neural signal and, in a second stage, performs the squaring operation needed for energy extraction. A simple digital circuit keeps track of the most recent outputs from the squarer and provides the accumulated value of the input signal energy. The system has been simulated in an XFAB 0.18μm technology showing correct measurement of the energy.