LiHEA: Migrating EEG Analytics to Ultra-Edge IoT Devices With Logic-in-Headbands

Traditional cloud computing of raw Electroencephalogram (EEG) data, particularly for continuous monitoring use-cases, consumes precious network resources and contributes to delay. Motivated by the paradigm shift edge Internet Things (IoT) monitoring, we focus on this paper first step carry out EEG analytics at last frontier (i.e., ultra-edge) our considered cyber-physical system ensuring users’ convenience privacy. To overcome challenges due computational energy resource constraints IoT devices (e.g., headbands/headsets), in paper, envision a smart, lightweight model, referred as Logic-in-Headbands based Edge Analytics (LiHEA), which can be seamlessly incorporated with consumer-grade headsets reduce delay bandwidth consumption. By systematically investigating various traditional machine deep learning models, identify select best model envisioned LiHEA. We consider use-case detecting confusion, representing levels distraction, during online course delivery has become pervasive novel coronavirus (COVID-19) pandemic. apply unique feature selection technique find features are triggered confusion where delta waves, attention, theta waves were announced three most important features. Among customized random forest demonstrated highest accuracy 90%. Since dataset size might have impacted performance learning-based approaches, further convolutional generative adversarial (DCGAN) generate synthetic traces representative samples original thereby enhance variation data. While performances models significantly increase after data augmentation, they still cannot outperform model. Furthermore, complexity analysis is performed best-performing algorithms, emerges viable