Thursday, May 30, 2013

DustDuino: A plan to crowdsource environmental reporting with low-cost dust sensors

A before and after photo of Shanghai inundated with smog, by Flickr's morgennebel.

There's some good news in the air. Or, well, about the air.

Last month, the American Lung Association released its 2013 report on the quality of the nation's air. The Lung Association's "State of the Air" report shows the country's air is continuing to improve.

The Lung Association attributes healthier air as a "direct result of emissions reductions from the transition to cleaner diesel fuels and engines and coal-fired power plants, especially in the eastern United States."

That much is good. But there's much room for improvement.

According to the report, "Of the 25 cities with the worst problem with spikes in particle pollution, fourteen had more days or worse problems in 2009-2011 than in the previous report." Six cities included in the report had their worst year ever (over the 14 years the report has been made) for short-term spikes in particulates.

Chicago, home to some of the dirtiest coal-fired power plants in the nation, continued to flunk both daily and annual levels of particulate pollution. This was where I did my own environmental investigation on how the Fisk coal-fired power plant threatened the health of the Pilsen neighborhood (see: "Battle in the Barrio").

The Fisk station dumped 755 tons of particulates into the air on an annual basis, according to a pollution report from the Environmental Protection Agency. The Clean Air Task Force found that the plant contributed to 15 deaths and 23 heart attacks annually, a figure based on EPA data.

The residents of Pilsen eventually won their battle, and the Fisk station was closed. EPA tests after the plant closure showed the particulate and radiation levels had returned to city-wide norms. Interestingly, one of the EPA air quality monitors was mounted in a baby stroller to measure levels around the perimeter of the plant.

Pilsen didn't have air quality monitors to begin with, though. Its residents had to lobby the Illinois EPA before the state placed an air monitor atop an elementary school in the neighborhood. The readings from the monitor led the IEPA to declare Pilsen a "nonattainment" zone for lead, a particulate which impairs the IQ, learning capabilities, and memory of children.

What's the extent of the problem?

Of course, Cook County isn't alone in its struggles. According to this nifty Lung Association info-graphic, 131 million Americans live in a place that flunks some kind of air pollution measure.

What's the cost of poor air quality? It's estimated that bad air contributes to 50,000 premature deaths a year, and costs the US economy $150 billion annually.

Outside of the United States, many developing countries are struggling to balance economic viability with ecological concerns. China, in particular, has gained notoriety for the precipitous decline of its air. According to the Washington Post, in January, the US Embassy in China recorded pollution at levels more than twice what the EPA considered hazardous to human health. (You can view the live air quality data from the embassy here)

Last week, China's government newspaper China Daily reported that "many first-tier cities, despite their outstanding competitiveness, are barely people-oriented and hardly satisfactory in ecological protection."

Awareness is key to tackling the air pollution problem. Citizens might be able to tell qualitatively whether the air is bad in their communities, but might lack the quantitative data that empowers them to identify problem areas. That's where low-cost sensors come in.

I should note that there are ongoing efforts to crowd-source air quality using similar technology. Notably, the Air Quality Egg is an Arduino-based device which measures a temperature, humidity, carbon monoxide, and carbon dioxide. The group behind the Egg's development have done a tremendous amount of groundwork in the world of community sensing projects.

AQE was voted one of the top Kickstarter projects in 2012, and it makes good work of the best low-cost chemical sensors available. Even so, several of the sensors lack of resolution at extremely low levels, and sometimes have issues with reproducibility. Cost might be a barrier to widespread adoption, as a fully-assembled Egg and base station costs $185.

The DustDuino is a shift away from a holistic measurement of air quality, to target one of the most serious challenges to clean air: particulate matter, commonly known as dust.

Focusing on just one sensor reduces the cost of the node, making possible to distribute larger networks of devices. And reducing complexity allows the communities to run workshops around the device, generating excitement and empowering citizens to develop even more sensing opportunities for their communities.

What are particulates, and how do you measure them?

Particulates, or Particulate Matter (PM) are a grab bag of pollutants. They're very tiny particles of solid or liquid matter, anything from cigarette smoke, coal ash, products of combustion, pollen, mold, to organic and inorganic compounds, and more.

What makes these compounds dangerous is their tiny size. A human hair measures between 50 and 70 microns across, but some of these particulates are under 10 microns. That's small enough to pass through your lungs and into your bloodstream, where they can create problems.

What kind of problems? Even short-term exposure to particulates can trigger asthma attacks in children, inflame lung tissue, increase the number of heart attacks among the elderly and other susceptible populations. They also can cause strokes and death from respiratory and cardiovascular issues.

The smaller these particles are, the more problems they can cause. The EPA tracks coarse particulates under 10 microns with its monitors, and has established a safe level for the number of those particulates in the air (PM10). But because of the specific dangers of ultra-fine particles, the EPA has a separate save level for particulates that measure 2.5 microns and under (PM2.5).

The American Lung Association relies on the EPA, and its network of 5,000 active air monitors, for its air quality data. PM monitors usually employ at least two kinds of filters: one to capture particulates that measure between 10 and 2.5 microns, and another filter to capture particulates under 2.5 microns. PM monitors measure the mass of the filters after they capture incoming particles from the air for a period of time. The more the filters weigh, the more particulates there are in the air.

It sounds like a simple process, but it's actually quite complicated. The volume and speed of the air sent through the filters has to be controlled constantly. Sometimes the air is precisely condensed. Then there's the matter of measuring the filter down to a microgram.

That's just one option, though. The Dekati ELPI, for example, electrically charges the particulate matter in the airflow, which then cascades down a series of charged metal plates. Sensitive electrometers measure the minute differences in the plates' charges as they're struck by charged particles. This is called a gravimetric impactor.

Illustration by

All that tech costs money. These monitors typically run thousands of dollars. That makes it difficult to deploy these units to the masses, and to crowdsource PM monitoring data.

How do you make a cheap, accurate dust sensor?

Sharp dust sensor, comparison to a quarter.

There is another option to measure particulates, besides expensive gravimetric impactors.

There is a class of small, inexpensive sensors that use simple infrared (IR) LEDs and IR receivers to measure the quantity of dust in the air. Ambient dust falls through a hole in the sensor, where an IR LED flashes at a certain frequency. This dust diffuses or absorbs the IR light, meaning the receiver will pick up less light as dust increases.

Teardown of the Sharp dust sensor. Note the orientation of the IR LED emitter and the IR receiver at 45 degree angles, at separate corners of the sensor.

A circuit within this sensor amplifies the minute difference between the LED output, and the receiver's input. The greater the voltage coming out of the dust sensor, the more dust that's floating in the air.

Circuit board of the dust sensor. The integrated circuit on the left side of the board amplifies the signal. Wires were soldered to bypass the sensor's connector.

The Sharp GP2Y1010AU0F uses this method to measure the total mass of all PM in the air. It costs about $10. Fortunately, it doesn't have as many drawbacks as other sensors, especially in terms of power consumption, reproducibility, and accuracy.

Of course, there's more to the dust monitor than the sensor alone. Something needs to operate the sensor, and turn the sensor's output into useful data. Preferably, this device should also send that data across the internet to a server, so data can be viewed and used by anyone.

Even better, if a community established a network of these dust sensors, data from these dust sensors could be aggregated to identify and tackle a specific problem in that community.

Because of its low cost, simplicity, flexibility, and easy learning curve, the Arduino microcontroller serves as the heart of the DustDuino. The prototype relies on the Arduino Uno to pulse the sensor's LED, convert the voltage of the sensor's output into a digit (analog to digital conversion), and to send that digit onto a device which can post it to a server.

For internet connectivity, it uses the Wireless SD Shield, and the RN-XV wifi module (see "Building a Wifi Temperature Node for Journalism" and "Nodes for journalists: a primer on bringing sensor data to the reporter" for details on these items).

All of these prototyping components are handy because they allow a developer to troubleshoot or fine-tune a system easily. Sadly, they also drive up the cost. The combined total of the Uno, Wireless Shield, and RN-XV is about $75 retail.

The DustDuino in an early prototyping phase, sans wireless shield, wired to a breadboard. Arduino Uno is connected to a 3D-printed mount.

The good news is this entire system only needs a handful of components, and the schematic is quite simple. Actually, it only needs three capacitors, a resistor, an oscillating crystal, a couple of small voltage regulators, the dust sensor, the ATmega 328P microcontroller IC, the RN-XV, and a couple connectors.

These components can fit on a single 3in x 3in printed circuit board (PCB). Purchasing the components individually and soldering it together would bring the cost down to about $55 dollars, and that's without any economy of scale.

How do you bring them to the community?

A prototype circuit board being etched at the CU Community Fab Lab,
One sensor can empower a person to learn about their local environment. But a network of sensors nodes can empower a community to wrangle a local ecological threat.

My community is fortunate enough to have a Fab Lab, a high-tech, digital fabrication laboratory that's opened to the public. It's enabled me to make the housing for my sensor nodes, using a laser engraver/cutter (see "Making a home for a sensor - with a laser"). It also is equipped with milling machines that can carve out a prototype printed circuit board from blank copper boards.

Community fab labs are great for bringing the latest prototyping technology to the public, and enabling digital literacy. I think they have the potential not just to create great things, but to create great things that can allow citizens to learn more about their local environment (see "Making mental munition: from bits to atoms to understanding").

The design needs just a little more work, and a few rounds of DustDuino printed circuit boards need to be prototyped, but a place like the local CU Community Fab Lab would be an excellent place to bring the community together around an environmental informatics workshop.

Participants would not simply solder together a PCB mindlessly. Rather, the attraction is an opportunity to learn about circuits, electronics, programming, and rapid prototyping. No prerequisites.

A person could walk in with a kit, and walk out with a fully-function dust sensor, along with new knowledge that can be applied to future artistic and scientific endeavors.

Where is DustDuino development at right now?

DustDuino prototype cube.
At the moment, the DustDuino exists as a working prototype. It's an assemblage of Arduino shields, wires, and connectors. I've also designed and made a laser-cut acrylic housing suitable for indoor dust monitoring, and a birch wood housing for possible outdoor deployment (complete with power source and outdoor extension cable).

DustDuino prototype outdoor cube being glued together. An outdoor power cord was integrated into the design.

At this prototype phase, it does transmit data to Xively (formerly Cosm, formerly Patchube). That is, when it's plugged in. Below is a Cosm readout from one morning when I was frying bacon. I believe the short-lived spike corresponds to the bacon-cooking. I could be wrong, however.

Cosm DustDuino feed.

DustDuino on Xively, the new version of Cosm.

As always, updates to come. I hope to talk more about this project at the workshop on sensor journalism at Columbia University's Tow Center, which takes place this weekend.