We’ve all heard the stories about how much data there is out there (lots and lots) and how much new data is being created (lots and lots and lots). Various accounts highlight how we are creating more information each year or so than has been previously been created in the entirety of human civilisation.
What’s driving this? Well it started with the Internet but now the world of the Internet is leaking out of our desktops and laptops and into the physical world in which we live.
Gordon Moore’s eponymous law, which either predicted or drove (depending on who you read) a doubling of computing power every 18 months or so, has created a world of ubiquitous and almost infinite computation. And the rise of the smartphone has created the ultimate sensor platform by turning our precious mobile devices into always-on, always connected, mobile sensor nodes, capable of sensing everything from our tastes and preferences to our location, activities, and even our moods. Cheap computation and small sensors are combining to create the big data world of the sensor web.
While the modern smartphone comes with an impressive array of basic sensors (location, accelerometers, cameras, microphone, direction, light, etc) these sensor modalities represent just the tip of the iceberg in terms of what can be sensed. In fact the only real limitation seems to be the genius of app developers as they turn their attention to innovative ways of combining standard sensors or creating “proxy sensors” by looking for readily available data patterns that are correlated with what needs to be sensed.
The early innovations were obvious, using the accelerometers and location sensors to detect activity, create maps, and log traffic patterns. But who would have thought that a smartphone camera could be used to detect a person heart rate by detecting tiny variations in skin tone due to subcutaneous blood flow? Or what about accurately evaluating lung function by asking someone to blow into their smartphone microphone, using sound as a proxy for evaluating air flow volume? What about predicting a user’s mood and mental state from changes in their communication patterns; a sharp increase in “phone fidgeting” as an indicator for anxiety or a sudden decrease in reliable communication patterns as a signal of a depressed mental state.
Things start to get really exciting however when we get a glimpse of what mighty be possible by taking advantage of the connected nature of smartphones and their sensors, to take advantage of communities of users as participatory sensor networks. Each user is essentially a mobile sensor, detecting signals as they move about the world. And patterns in these signals can be used to monitor events and make predictions.
One of my favourite examples, is the application of a simple sensor designed to be used with a standard asthma inhaler; it’s produced by a company called Propeller Health. The sensor registers when the inhaler is used. That’s really all it does. It’s cheap and reliable and easy to install. It does do one more thing, however: it talks to a nap on the user’s smartphone to log each usage. That in itself does not sound particularly innovative — it might be useful to keep track of when the user needs to refill or may help a clinician to evaluate medication efficacy — but the magic starts to happen when we consider the location sensing capabilities of smartphones.
The thing about inhaler use is that it tends to happen at or near the point at which the user begins to feel discomfort. Now imagine a city of thousands, tens of thousands, or even hundreds of thousands of inhaler users, across a range of respiratory conditions. Each time a user feels discomfort they use their inhaler and the app registers the location. This information can be sent back to a central server which quickly builds a real-time map of inhaler usage. And, as if by magic, an air quality sensor network is born. Smog problems trigger a spike in inhaler usage downtown. A change in wind direction, which brings in smoke from nearby forest fires, triggers a usage spike in the suburbs.
These spikes can be mapped in real-time and appropriate actions taken. Registered sufferers can be alerted to nearby problems or re-routed on their way to work. Nearby clinics can be forewarned of a probable increase in patient drop-ins. Local pharmacies can be alerted to likely changes in medication demand. And neighbours and carers can be encouraged to check in on their loved ones.
Imagine the cost and complexity of building this type of air quality sensor infrastructure using traditional methods? It would require thousands of complex chemical units to be installed and maintained. They would also need wireless communications and dedicated power sources. They would clutter up our pavements and roadways and costs millions to install and maintain. The Propeller Health alternative delivers the same or better functionality as a side-effect of a natural proxy indicator for air quality, namely inhaler usage. Matters of maintenance, power, and connectivity are no longer relevant.
This type of thinking is transforming the world in which we live. It has already changed how we engage with online services: the likes of Google, Amazon, and Apple have long ago figured our how to turn our online activity into sophisticated sensor networks capable of detecting our intentions, preferences, and interests. But the next phase of innovation will see these ideas leaking into the physical world with the potential to fundamentally disrupt everything from healthcare, transport, government, policing, the environment, and energy usage. This is the age of small sensors and big data.