Power Aware Laplacian Multi-Set Canonical Correlation For Reducing The Power Consumption In Wireless EEG Sensor Network

Abstract

Now a days, miniaturized EEG modules suffers from poor spatial coverage in monitoring the EEG signals. The utilization of wireless EEG sensor network (WESN) helps to improve the spatial coverage that can communicate better for shorter distances. However, it leads to high consumption of energy since it is associated with collection and transmission of data through WESN. One way to reduce the power consumption is to remove the eye blink artifacts, which creates additional noises in EEG channels. In this paper, we propose Laplacian Multi-Set Canonical Correlation (LMCC) to remove the eye blink artifacts. It considers the correlation of the EEG channels to a maximum extent even with constrained bandwidth resources. The LMCC considers local EEG channel within-view and local EEG channel between-view correlation using neighbor graph. Further, it discovers the non-linear correlation among the multi view EEG data and eliminates it using local linear problem formulation. The performance of LMCC method is tested over real and synthetic EEG signals. The results shows that the proposed method removes the blink artifacts from the EEG signal in a better way with reduced power consumption in wireless networks.