Earthquake Early Warning Using Low-Cost MEMS Sensors

Abstract

Over the past several years, low-cost micro-electro-mechanical systems (MEMS) sensors have been actively used for detecting earthquakes in real time. To utilize the low-cost MEMS sensors for earthquake early warning (EEW), it is crucial to construct a high-density seismic network of the sensors along with a server system to process a large volume of acceleration data transmitted from the sensors. It is also of vital importance to design an effective and efficient detection algorithm for the networked system in order to ensure realtime earthquake detection with high accuracy and low false alarms. In this paper, we introduce our recent effort in South Korea to build an EEW system based on several thousands of low-cost MEMS sensors and a deep-learning-based detection algorithm, which is currently being developed and deployed nationwide. Its initial version of a networked system has been in operation for over a year and has been able to detect a few small-magnitude earthquakes. While the early results have been positive and promising, we envision to ultimately integrate the networked system of low-cost MEMS sensors with the existing network of traditional seismic stations so as to build a unified EEW system that has better coverage and detection performance.