[From the last episode: A magnetometer can detect the earth’s magnetic field – or any other one.]

OK, so we’ve looked at three different sensorsA device that can measure something about its environment. Examples are movement, light, color, moisture, pressure, and many more. that tell us something about our motion – which means that they can also tell us something about our position. There’s a little bitThe smallest unit of information. It is a shortened form of "binary digit." Since it's binary, it can have only two values -- typically 0 and 1. of overlap between the sensors, however. Could we make use of that?

Second opinions are always good — especially if they come from different sources or backgrounds. An internal-medicine doctor and an oncologist would probably work better together to figure out what’s up with your lungs than two internists or two oncologists.

We can do that with sensors too, not only to have them cross-check each other, but to do totally new things that the individual sensors couldn’t do on their own. There’s a sci-fi-sounding name for this: sensor fusionThe process of combining information from multiple sensors (real or virtual) in order either to give more confidence in a result or to come up with something completely new..

Fusing Sensors

With a little bit of care, we can combine data from any sensors to create something new. Now, you sort of want to have in mind what you’re doing – lots of combinations would be meaningless. But multiple ways of looking at motion can give you more confidence in your position. A magnetometer is not the same as a compassIn its electronic version, a magnetometer that has its reading compensated by a tilt meter (done by an accelerometer)., but with the help of another sensor, you can make a compass.

What may sound even odder is that you can even think of information sources as sensors. Let’s say you want to get to know someone’s food preferences. Well, you might take their Facebook feed, where there are pictures of their happiest meal memories (not to be confused with their memories of a Happy Meal), and combine it with their Yelp reviews, and you can get a better picture of what they like to eat.

Note that, in this example, neither of the “sensors” directly provides favorite-food information. They both have bits and pieces that, when extracted and combined, can act like a “favorite-food sensor.” In fact, some of your cell phones (which carry many sensors) recognize, in their internal structure, the notion of a virtual sensorA source of data that, strictly speaking, isn’t a sensor, but that can be used as if it were a sensor.. That’s something that isn’t a real, dedicated sensor, but that turns out information just as if it were real. As a user (or as other softwareIn this context, "software" refers to functions in an IoT device that are implemented by running instructions through some kind of processor. It's distinct from "hardware," where functions are built into a silicon chip or some other component. in the phone), you don’t really care where the “sensor data” came from – just as long as you can trust it.

Sensors Everywhere

With that mindset, you can treat anything like a sensor. A dictionary? Sensor. A (digital) recipe book? Sensor. Some company’s product catalog? Sensor.

The trick, then, is finding out which of these “sensors” is useful and how you can combine it with other sensors. Now we’re veering directly into the world of the dreaded algorithm. Because algorithmsA way of calculating something, typically as a step-by-step recipe. A familiar one to all of us is how we do long division on paper: that's an algorithm. Algorithms abound in the IoT. are how sensor data gets fused. For instance, maybe you start by identifying what foods are in Facebook posts and then rank them based on Yelp reviews? Ah, but a bad review might not mean you don’t like the food – it might just mean that you didn’t like that particular restaurant. So… maybe rank them by the number of reviews per type of food?

This is the way that designers create algorithms. Frankly, Google’s rating system for deciding which posts come up first in the search results acts this way – some complex way of combining a bunch of information from different sources to give a result. And they tweak the algorithms constantly.

Which makes an important point: there’s no one right way to fuse sensors. In some cases – like making a compass out of a magnetometer – you may have the benefit of some fundamental mathematical relationship that everyone would agree on. But most designers take things further, weighting different sources by trustworthiness or accuracy, for instance. This is a way that companies can compete: whoever gets the best results from the same data – or from innovative sources of data – wins. At least in the sensor competition, anyway.

So, in the next couple of posts, we’ll look at so-called inertial measurement unitsA combination sensor with an accelerometer and a gyroscope, and, often, with a magnetometer. The first two rely on inertia to tell you where you are and where you’re going., or IMUs, and we’ll see how we can combine accelerometersA sensor that measures acceleration and deceleration., gyroscopesA sensor that detects when it changes direction., and magnetometersA sensor that detects magnetic fields -- from the earth or from anywhere else. – and other things – to create better location information.