A compressive eletroencephalography (EEG) sensor design

This paper presents a design of compressive eletroencephalography (EEG) sensors that can generate high fidelity EEG measurement through a compressive sensing approach. The conventional sampling structure of EEG sensor consists of uniform electrodes placed on the scalp. Each electrode generates one channel of signals to be further processed. The proposed method contributes with a new EEG sensor design. It includes (1) non-uniform multiplex sampling structures and (2) incoherent compressive sensing protocols. In the new design, each electrode consists of a number of elements in pseudo-random shapes, yielding EEG signals with different spatial resolutions. The multi-channel sampling rates are controlled by a pseudo-random vector generator. The final measurement is the sum of the multi-channel signals within a time window after each sample is encoded with a Bernoulli number. This method allows to achieve compressive measurements of EEG signals with 16 sensing elements at a random sampling rate, resulting in a high compression ratio with little information loss.