ROHM’s Smart City Sensor Module Detects Earthquakes—But the Real Star Is the Algorithm

July 11, 2019 by Gary Elinoff

Built with inexpensive hardware components, ROHM's algorithm-based module has built-in false detection function.

Built with inexpensive hardware components, the algorithm-based module has built-in false detection function.

At Sensors Expo 2019, ROHM Semiconductor introduced the BP3901, a sensor module specifically designed to detect earthquakes. The key feature is a ROHM-developed algorithm that can effectively detect earthquake vibrations, and not register false positives when presented with more benign vibrations, even if they are large in magnitude.

The registration of false positives is further deterred by the device’s angle correction function. Through this feature, unintentional tilts of up to ±15° will have no effect on overall detection accuracy. The lack of a similar feature has proven to be a considerable source of problems in previous earthquake detection devices.


The Smart City demo at ROHM's Sensors Expo 2019 booth. Image by Kate Smith.


The physical device is composed of little more than an MCU and a 3-axis accelerometer (from ROHM Group company Kionix) in a compact 11.8 mm x 8.6 mm x 2.5 mm module package.


The BP3901 from ROHM Semiconductor. Image source: ROHM Semiconductor

The BP3901 will stay in standby mode, drawing a scant 3.5uA until seismic activity is detected. This is a vital factor, as it allows the device to be installed in remote, hard to access locations with no need for battery replacement for years on end.

Taking Advantage of the Cloud—or Not

An interesting capability of the BP3901 is that the tiny device can, by itself, determine if an earthquake is happening. It does not need to send information to the cloud for analysis and a “yes or no” decision.

This complete elimination of the problem of latency can be a lifesaver. Critical processes, such as nuclear power plants or chemical plant operations, can be immediately shut down, potentially avoiding catastrophe. In addition, the device saves calculated seismic data from up to 16 seismic events in non-volatile memory.

On the other hand, ROHM expects to further development in the direction of integrating the device with Wi-SUN based networks that is optimized for smart meter outdoor IoT. In this manner, networks of BP3901 will be able to leverage the cloud to glean region-wide earthquake data. This will help both scientists and regulators to better understand the scale and distribution of earthquake-related data.

Seismic Intensity

Japan has developed a measurement of seismic intensity called the SI Index, which measures the effects of an earthquake on people and on man-made structures. This is which is what the BP3901’s algorithm is designed to measure. However, Satya Dixit, Senior Director, System Applications and Solutions Marketing at ROHM Semiconductor, informs us that the conclusions are valid anywhere, perhaps with minor modification for local regulation.

The illustration below presents waveforms produced by earthquakes, and other large waveforms generated by other sorts of powerful vibrations.


The BP3901 distinguishes between earthquakes and other powerful vibrational sources. Image modified from ROHM Semiconductor

Uses for the BP3901 and Its Algorithm

The ultimate purpose of the device is twofold. It can be installed in anything that has to react immediately in the event of an earthquake. These include:

  • Smart meters (power, gas)
  • Unmanned transport vehicles
  • Infrastructure such as roads and buildings
  • Household appliances (i.e. fan heaters, lighting)
  • Distribution boards (breakers), AC outlets
  • Bridges

It can also be used to record seismic events for later study, with placement virtually anywhere.

The Beginning of a Trend?

There were other examples of companies at Sensors Expo 2019 that did not provide hardware at all, but just provided offerings in data analytics and algorithms. This is likely because sensor networks for the IoT/smart cities generate A LOT of data—and companies are now trying to not only provide the sensors that gather data but also help engineers/customers/companies interpret that data into a useful format. 

Indeed, at the recent CHIPS Alliance conference, the concept of open source hardware, as opposed to open source software, was presented. So will companies start to get their hardware designs free of charge and begin paying only for the algorithms that run on them?