Skip to Main Content

Along a flowing creek nestled in the rolling hills of Pepperwood Preserve in Sonoma County, California, Carnegie Mellon University researchers are adapting internet of things technologies to alert authorities when flooding is imminent.

This research at a main watershed of the Russian River is supported by the United States Geological Survey (USGS), the agency within the Department of the Interior that studies the science underneath our nation’s lands and the hazards that threaten them. The USGS and CMU are exploring how sensor systems can be used to monitor major waterways to predict floods and other threats. However, the places that need sensing are often the worst places to install sensors. The earliest indications of potential problems are often found far from civilization—in remote areas that lack readily available power and wireless connectivity.

“I’m tempted to say that what we’ve got here is a failure to communicate. Not only is wireless coverage a problem, there are also power and programming problems,” begins Bob Iannucci, who is leading the research. Iannucci, the director of the CyLab Mobility Research Center at CMU Silicon Valley, explains that to create vast sensor systems, the sensors have to be inexpensive to install and maintain, and their energy source must last for years. While sensors appear to perform simple tasks, programming them to successfully run on ultra-low power presents more complications. Iannucci’s team is addressing low-power operation from the top down, including how to design the sensors, how to engineer the software, and how to internetwork them. The research is playing out in an outdoor setting that spans three thousand acres.

I’m tempted to say that what we’ve got here is a failure to communicate.

Bob Iannucci, Director, CyLab Mobility Research Center

“We’re looking at this as a full-system problem, rather than just a sensor or radio problem,” says Iannucci. He and his team of Carnegie Mellon graduate students are adapting low-power, wide-area wireless network (LP-WAN) technologies to monitor rural areas. LP-WANs make it possible to gather sensor data over large areas at low cost.

What we are building and why

At the heart of the Pepperwood Preserve installation is a solar-powered LP-WAN gateway that relays information between sensor devices at the streams being monitored in Carnegie Mellon laboratories in the Silicon Valley. The sensors, using ultrasonic ranging, measure the height of the streams and send data back through the gateway to CMU servers. The data will be analyzed and the results published on a webpage. Earth science researchers have the ability to remotely alter the schedule the sensors follow so that precious battery energy is only spent when the conditions warrant it.

“We are building a system that serves the USGS’ real needs, but it also provides a vehicle for us to do research on low-power wide-area networks as they scale up,” says Iannucci. In addition to a warning system, the USGS wants a system that preserves historical sensor readings to enable longitudinal studies of how waterway conditions change over time.

We are building a toolkit that can be used across a variety of sensing environments.

Bob Iannucci, Director, CyLab Mobility Research Center

“Sensing and telemetry [the process of collecting and transmitting data] are expensive. But LP-WANs and less expensive sensors have the potential to raise existing awareness of water hazards and to enable us to rethink how we manage our nation’s waterways. We are exploring if this approach is right for broader investment, and Pepperwood Preserve is a test bed,” says Jonathan Stock, director of the USGS National Innovation Center.

“Our Sentinel Site project is collecting detailed climate-hydrology-ecosystem data for the long term, to understand the impact of climate variability on our region’s watersheds and biodiversity. [Pepperwood] having burned over in the 2017 Northern California fires, our data sets are now an absolutely unique resource for understanding the role of fire in watershed processes in California’s Coast Ranges. By partnering with Pepperwood, USGS and CMU have the opportunity to contribute directly to short-term recovery and long-term resilience strategies for California as a whole,” says Lisa Micheli, president & CEO of the Pepperwood Foundation.

What lies ahead

The work Iannucci’s team is doing in the preserve is part of a larger CMU initiative called SMILE: the Synchronized Multi-sensor Integrated Learning Environment. “The concept is to have an array of dissimilar sensors that can be integrated, possibly on the fly, into a single homogenous system,” says Iannucci. For example, different kinds of sensors can be mounted on drones or road surfaces, and a unified, low-power, wide-area network can help bring all their data together. The environment can be anything: smart buildings, car-lined city streets, or grassy hills in California.

“We are building a toolkit that can be used across a variety of sensing environments,” states Iannucci. The kit’s components include a family of flexible sensors, a unifying LP-WAN, and programming and reasoning tools for achieving low-power operation and making sense of collected data through a combination of traditional analysis and machine learning techniques.

“We would be pleased if the work we’re doing with the USGS was generalizable enough to be applied in other environments,” says Iannucci. He explains that deep sensor science—more sensors in more places—can provide hyper-local insights such as how changes in soil moisture benefit plant life or foretell mudslides. The wider a sensor network is, the better and earlier we can detect forest fires, air quality issues, erosion, and other conservation challenges. Today, the costs of installing and operating dense sensor networks are high; however, Iannucci suggests that this will change.

“Solving core problems related to low-power environmental sensing has the potential to substantially lower deployment and operational costs,” he says. “This can enable more sensors for the same investment and, with it, the ability to draw better-informed conclusions about the environment.”

For media inquiries, please contact Dan Carroll at Dan Carroll.