Community Seismic Network
Millions of Internet-enabled devices like phones, laptops, and game consoles now have accelerometers that can be used to detect and measure earthquakes. Harnessing the data from these sensors could allow us to quickly detect large earthquakes, and accurately estimate where damage has occurred and where emergency responses are needed shortly after a quake. However, processing the data from a city-wide sensor network is a challenging Internetscale problem involving real-time analysis of many noisy sensor measurements. The Caltech Community Seismic Network (CSN) Project is a collaboration among geophysicists, civil engineers and computer scientists to develop the sensor technologies, scalable infrastructure, and algorithmic tools needed to reliably perform large-scale seismic sensing.
Sensors in the Community
In contrast to traditional seismic networks that contain a small number of highly accurate sensors, the CSN project focuses on large numbers of inexpensive, community-held sensors, such as those in personally owned devices like smart phones. CSN-Droid, a free and opensource Android App, allows volunteers to join the network and contribute accelerometer measurements. The CSN project is also actively distributing USB accelerometers to volunteers in the Pasadena area. CSN has been working with local schools to recruit volunteers and teach about the science of earthquakes. Several hundred 16-bit accelerometers made by Phidgets, Inc. have been distributed to the community towards the goal of measuring data from 1500 sensors in the Pasadena area.
Figure 1. A CSN community USB accelerometer.
Mobile Sensing with CSN-Droid
Smartphones contain a powerful collection of sensors like GPS, accelerometers, and gyroscopes that make phones an ideal platform for collecting data about how a community experiences an earthquake. Because the motions of each person’s phone are unique, CSNDroid uses online learning and anomaly detection to gather the most informative measurements from the accelerometer in a volunteer’s Android phone.
Figure 2. CSN-Droid collects sensor data and provides users with information about earthquakes worldwide.
Figure 3. CSN-Droid sensors worldwide.
Large-Scale Processing in the Cloud
The CSN project envisions city-wide networks of community-owned devices, and is developing scalable techniques to process the enormous amounts of data that such networks would produce. For example, if only 1% of smartphones contributed sensor data, the network would need to process hundreds of thousands of continuous data streams. To meet this challenge, CSN distributes the data processing across the individual sensors and performs central processing using Google’s App Engine cloud computing platform. Initial processing at each sensor is used to filter normal sensor data, and only use bandwidth for transmitting potential seismic measurements. These selected measurements are processed on App Engine, which allows processing power to be rented on an as-needed basis. This allows large network loads such as during an earthquake to be handled easily, without requiring maximum capacity on a day-to-day basis.
CISN ShakeAlert and CSN
In addition to collecting data about earthquakes, community-owned Internet-enabled devices are also an ideal platform for notifying the community about important seismic events. In a collaboration with CISN ShakeAlert, the CSN project is developing the functionality to provide ShakeAlert warnings directly to volunteers’ Android phones. A prototype interface is being developed to display ShakeAlert warnings through CSN-Droid.
Andreas Krause, K. Mani Chandy, Julian J. Bunn, Matthew Faulkner, Annie Liu, Michael Olson