Washington Scientists Forge Ahead Amid Uncertainty
The world is turned on its head, but there are still salmon to monitor as they navigate dams, crops to improve before climate change sets in, and energy grids to protect from cyber attackers.
On Thursday evening, as protestors marched through downtown and helicopters hovered in the sky, scientists from Washington’s premiere public research institutions showed off their latest works to the Seattle innovation community. In the still-shocking aftermath of the election, the SciTech Northwest event—hosted by the Technology Alliance and participating institutions—provided a welcome distraction in the form of substance, human ingenuity, and forward-looking progress.
Here were a handful of outputs from more than $2 billion in annual public research investment—the majority of it from federal sources—at University of Washington, Washington State University, and Pacific Northwest National Laboratory. These institutions, which recently agreed to work together more closely on research and education, are major underpinnings of the Pacific Northwest’s thriving innovation economy.
But hanging over the elegant room on the top floor of the Edgewater Hotel was the same fog of uncertainty settling in to nearly every aspect of life in America. What will become of the nation’s scientific enterprise in the time of President Donald J. Trump?
For the most part, the scientists were more interested in talking about their work than this uncertain future.
The conventional wisdom is that Trump and a Republican-controlled Congress will seek to shrink the federal government, possibly including federal research investment. But who really knows? In the online Presidential Science Debate 2016—one of the scant venues to address these issues—Trump expressed a measure of support: “Though there are increasing demands to curtail spending and to balance the federal budget, we must make the commitment to invest in science, engineering, healthcare and other areas that will make the lives of Americans better, safer and more prosperous.”
Here’s a closer look at three projects happening in Washington state now:
Smart Sensing in Phenomics. Farmers in Washington and around the world select new varieties to plant based on a huge range of factors from the juiciness of an apple to the drought tolerance of wheat. The process of developing new varieties—phenotyping—has long relied on the laborious and subjective visual observations of breeders, who may test thousands of new varieties and select just one or two for a certain desired trait.
That’s starting to change. WSU assistant professor Sindhuja Sankaran is part of a team of biological systems engineers bringing modern sensing technologies and data analysis to the field of phenomics. They are experimenting with hyper-spectral sensors and thermal cameras that can see what the human eye cannot. The temperature of an experimental plant’s leaves, for example, can reveal transpiration rates, a clue to how efficiently it uses water.
By mounting these sensors on drones, conventional small aircraft, and field platforms, researchers can gather data much more efficiently than a human walking through a greenhouse or a field.
The challenge is finding the right combination of sensors and platforms to gather the data for each crop, and then analyzing it in a way to provide useful information to breeders, Sankaran says.
It’s a new field, she says. Researchers in Europe and Australia have been at it for five or six years, but skepticism remains. Sankaran and her team recently received federal funding from the National Institute of Food and Agriculture to study the benefit of sensor use in breeding apples, camelina, quinoa, dried peas, lentils, and winter and spring wheat.
The Ants Go Marching. Glenn Fink, a senior cyber security researcher at Pacific Northwest National Laboratory, in search of inspiration for a decentralized, adaptable way to spot constantly evolving attacks on broad IT infrastructure running systems like the power grid, found some in Proverbs 6:6: “Go to the ant, thou sluggard; learn her ways, and be wise,” he recites. “I just took it literally.”
Ants are collectively quite smart and successful, outweighing the rest of Earth’s species. “Ants solve some very difficult, not just hard problems, but adversarial problems,” Fink says. Their response to attack, methods of communication, and distributed decision making provided useful models for devising a new approach to IT security that is fast, decentralized, resilient, and adaptable to changing methods of attack.
The system, Ant-Based Cyber Defense, essentially unleashes a colony of digital ants on an IT system. These tiny, automated, limited-function programs crawl from machine to machine, diagnosing things like CPU, memory, and network usage.
When a CPU Digital Ant detects elevated activity at a machine it visits, it reports that to another lightweight program installed on each machine called a sentinel. The sentinel compares the elevated CPU report to other information it has received from other visiting ants. The sentinel may have an explanation. “Oh yeah, I’m a compute server, so I’m running high-end computational software,” Fink says.
The ant program, in that case, moves on to inspect the next machine on the network. But if the sentinel has no explanation for the anomalous usage the ant detects, it “feeds” the ant—rewarding it for finding something. The ant moves on, and now at each subsequent machine it visits, it leaves a message for other ants crawling the system—akin to the pheromones actual ants use to communicate with each other—to the effect of “Back that way, I got fed,” Fink explains. “Other ants come across that digital pheromone field and say, ‘Oh, OK, I might get fed that way.’”
So the Zombie Processes Ant, the Page Faults Ant, the Unauthorized Access Ant, and all the others follow the pheromone toward the anomalous machine to perform their own inspection. The anomalous machines—perhaps only a handful in a network of thousands or millions—are highlighted for human security to investigate further. “The idea here is we want to find artifacts of unusual behavior that may be malicious, that people can look into later,” Fink says, calling it a “general-purpose signature discovery framework.”
The underlying idea is that even as attack vectors evolve, they will always reveal themselves to at least some extent through anomalous utilization of computing resources that are under attack. By automating the discovery of those anomalies—with the help of the Digital Ants—they can be contained more quickly.
The Interdisciplinary Triumph of Sensor Fish. Tracking endangered fish as they traverse hydroelectric dams used to involve performing minor surgery to implant short-lived battery powered devices. Zhiqun Deng, chief scientist in PNNL’s Energy and Environment Directorate, demonstrated several new sensor and power technologies that promise less-invasive, longer-lasting fish-tracking devices, which are providing dam operators and fisheries managers with better information to improve survival rates.
The latest iteration fish tracker is smaller than a finger nail, with a lithium ion battery that can last 100 days, sending a location signal every three seconds, Deng explains. It can be quickly injected into fish, reducing mortality associated with earlier tracking technologies. Next spring, researchers will use a new sensor with a strip of flexible electronic material that can capture kinetic energy created by the fish swimming, powering the device for years.
“We are lucky because of the major advances in micro-electronics,” he says. “Even if we had the idea, we couldn’t do it a few years ago.”
The improved location data these sensors capture can be compared to other data gathered by another novel device. Sensor Fish, a smolt-sized tube packed with sensors, boldly traverses dam spillways, powerhouses, and fish ladders to determine what exactly is killing the real fish. As it travels from a dam’s fore-bay, through the turbines, and out the tail race, Sensor Fish measures pressure, acceleration, and rotational velocity.
“Once you understand that, then you can design your turbines accordingly,” he says, or adjust dam operations during times of fish migration to improve survival rates. “Some operations are more friendly than others.”
The data are particularly important now as old dams are repowered. “Now is a good opportunity to replace them with a better design,” Deng says.
The combination of disciplines brought to bear on this problem is impressive: low-power sensors, energy storage, micro-electronics, fisheries biology, dam operations, and the IT systems and data analytics work that bring it all together.
“It’s a really big team effort from so many different disciplines that actually makes this possible,” he says.
For a deeper look at inter-disciplinary innovation, attend Xconomy Intersect on Dec. 8 in Seattle, where entrepreneurs, researchers, and investors will explore the crossroads from which the next big idea may emerge.