Asynchronous Multi-Sensor Change-Point Detection for Seismic Tremors

Abstract

THIS PAPER IS ELIGIBLE FOR THE STUDENT PAPER AWARD. We consider the sequential change-point detection for asynchronous multi-sensors, where each sensor observe a signal (due to change-point) at different times. We propose an asynchronous Subspace-CUSUM procedure based on jointly estimating the unknown signal waveform and the unknown relative delays between the sensors. Using the estimated delays, we can align signals and use the subspace to combine the multiple sensor observations. We derive the optimal drift parameter for the proposed procedure, and characterize the relationship between the expected detection delay, average run length (of false alarms), and the energy of the time-varying signal. We demonstrate the good performance of the proposed procedure using simulation and real data. We also demonstrate that the proposed procedure outperforms the well-known "one-shot procedure" in detecting weak and asynchronous signals.

High-Dimensional, Multiscale Online Changepoint Detection

“…This data set was also studied by Xie et al. (2019), and was obtained from http://service.ncedc.org/fdsnws/dataselect/1/. To begin, we removed the linear trend in each coordinate and applied a 2–16 Hz bandpass filter to the data using the GISMO toolbox 3 ; these are standard pre‐processing steps in the seismology literature (e.g.…”

“…Ground motion sensor measurements were recorded using geophones at a frequency of 250 Hz in three mutually perpendicular directions, at 13 stations near Parkfield, California for a total of 740 s from 2 AM on 23 December 2004. This data set was also studied by Xie et al (2019), and was obtained from http://servi ce.ncedc.org/fdsnw s/datas elect/ 1/. To begin, we removed the linear trend in each coordinate and applied a 2-16 Hz bandpass filter to the data using the GISMO toolbox 3 ; these are standard pre-processing steps in the seismology literature (e.g.…”

“…To begin, we removed the linear trend in each coordinate and applied a 2–16 Hz bandpass filter to the data using the GISMO toolbox 3 ; these are standard pre‐processing steps in the seismology literature (e.g. Caudron et al., 2018; Xie et al., 2019). In order to reduce the effects of temporal dependence, we computed a root mean square amplitude envelope, downsampled to 16 Hz, and then extracted the residuals from the fit of an autoregressive model of order 1.…”

Online Changepoint Detection on a Budget

“…Furthermore, we compare the online BCPD algorithm to state-of-the-art drift detection algorithms. The test datasets consist of Martingale [10], global historical temperature [1], Nile [7] and earthquake data from Northern California Earthquake Catalog [26]. Aside from real world data, we also generate datasets where changepoints are known.…”

Fast Online Changepoint Detection via Functional Pruning CUSUM statistics

“…However, as technology develops, the demand from several fields for online changepoint detection procedures has increased drastically over recent years. Examples include, but are not limited to, IT and cyber security (Jeske et al, 2018;Tartakovsky et al, 2012;Peng et al, 2004); detecting gamma ray bursts in astronomy (Fridman, 2010;Fuschino et al, 2019); detecting earthquake tremors (Popescu and Aiordǎchioaie, 2017;Xie et al, 2019); industrial processes monitoring (Pouliezos and Stavrakakis, 2013); detecting adverse health events (Clifford et al, 2015); and monitoring the structural integrity of aeroplanes (Alvarez-Montoya et al, 2020;Basseville et al, 2007).…”