Smartphone sensor generates crowdsourced pollution maps
By Richard Moss
July 13, 2014
Fine dust pollution triggers all manner of health problems, but accurately tracking its concentration across cities and regions takes considerable manpower. That could get a whole lot easier with a sensor that attaches to a smartphone and measures particulate matter (fine dust) levels in the air, which is under development at the Karlsruhe Institute of Technology.
The idea is to use the aggregated data from these smartphone-attached sensors to generate pollution maps. Just how many people with these sensors would be required to make a reliable and worthwhile map is yet to be determined, lead researcher Matthias Budde tells Gizmag, largely because the sensor is still being developed and it hasn't been properly field tested. "The more stable and accurate the employed sensor is by itself, the more meaningful individual measurements become and the less important is the density of the measurements," he says.
This is not the first research into the topic. The University of California's CitiSense project already explored the idea of using a smartphone to track local pollution levels, but Budde and company's work differs in that it reads fine dust concentrations rather than just the coarser pollutants emitted by cars and other combustion vehicles.
Smartphones could be easily adapted to the task of reading particulate matter levels, the researchers believe, because they are already capable of emitting light through the built-in camera's LED flash. Indeed, an app that estimates pollution by analyzing photos was developed back in 2010.
The KIT researchers have gone a step further by attaching a light trap to the phone that can allow for higher light intensity – thereby allowing a more accurate reading. For the purposes of the research, Budde and his colleagues adapted a low-cost off-the-shelf Sharp GP2Y1010 dust sensor.
Light directed into this light trap gets scattered by dust and smoke in the air, with the brightness of pixels in a subsequent photo converted into dust concentration. The current sensor's accuracy is in the region of one microgram per cubic meter, enough for smoke and coarser dust, but not for fine dust.
The researchers suspect this could be significantly improved by moving to a newer sensor and by improving the coupling between the flash and the optical fiber that reroutes its light to the sensor, as well as by tuning the evaluation algorithms. A future version could also store the photos uncompressed, further increasing accuracy.
Get enough sensors out into the field and there will be an extra boost in accuracy, as nearby readings could be calibrated against each other – perhaps even in real time, the researchers suggest, for crowdsourced live pollution maps.
A production-ready version would likely clip onto the back of a phone with a magnet, with a form factor better suited to smartphones, and we could see this happen in the next year or two.
The problem remains, though, that once the sensor is ready for public use, there needs to be enough of an incentive to get people to use it. For some people, it'll be enough to simply know that participating will be a benefit to themselves and others.
For everyone else, Budde suggests some kind of game system might be feasible. At its most basic, this would involve rewarding participants for taking measurements, with bonuses at low-coverage locations.
Whatever the case for incentives and execution, crowdsourced, smartphone-based sensing – or something similar – could prove vital in developing an accurate and high-resolution picture of fine dust pollution in urban, rural, and suburban areas around the world.
It may be obvious that pollution levels are harmful in cities like Beijing, where smog is often visible throughout the day, but elsewhere the size and intensity of hot-spots and temporary peaks may not be so clear – exposure levels can vary widely even in different streets of the same city block.
And with existing measurement grids proving woefully inadequate for detailed data – Beijing has 18 measurement stations for its 16,000 sq km, while New York City has 13 across its 1,200 sq km – participatory sensing solutions such as this could be crucial in exposing rule-breaking emitters and protecting public health.
You can access research publications on the topic by Budde and his colleagues at Budde's TECO page.