MEC-Based Energy-Aware Distributed Feature Extraction for mHealth Applications with Strict Latency Requirements

Abstract

Mobile health (mHealth) applications are expected to proliferate due to the recent advances in IoT sensing devices and wireless technologies. Monitoring brain signals using mobile electroencephalography (EEG) headsets provides opportunities for epileptic seizure detection and prediction using machine learning algorithms. To notify patients on time for taking preventive measures, it is vital to develop low latency solutions. Due to the limited computing and energy resources of mobile EEG headsets, we propose a distributed feature extraction method that relies on the user equipment (UE) and mobile edge computing (MEC) servers. We formulate an optimization problem for distributed feature extraction with a joint latency and energy objective function, and present an effective solution approach that captures performance trade-offs. Simulation results demonstrate the effectiveness of the proposed method as a function of different system and design parameters for an epileptic seizure prediction mHealth application.