We are proud to announce a new feature in Sleep As Android – Pair tracking. It helps you to get a cleaner record of your sleep phases by combining the measurements from your and your partner’s phone, using any combination of sensors.

The very core function of Sleep as Android is monitoring sleep phases – light and deep. There is a bunch of features triggered by a sleep phase. Many of you enjoy a pleasant wake-up in the light phase with our smart alarm. True sleep enthusiasts experiment with lullabies or lucid dreaming, which also need to be triggered in the right sleep phase.

We estimate the current phase by monitoring body movement throughout the night, and identifying distinct periods with relatively high/low amount of body movement. We support a wide range of different sensors to monitor the activity: in-device accelerometer, sonar, various smart wearables, or Sleep Phaser. Anyone can choose what best fits his taste and conditions.

Does my partner break my sleep record?

What happens when there are two people sleeping in the bed? Each of them moves independently and their sleep phases are often not synchronized. Does my partner’s movement affect my sleep phases measurement? Does his tossing and turning show as light sleep in my actigraph? Could they, for example, trigger my smart alarm at a wrong time?

Well, strictly speaking, they could. The reach of the sensors is 1-2 meters and indeed, some of the partner’s activity gets always recorded by the other device. The question is how significant its impact is.

We performed a series of measurements with various sensors to asses the effect. The answer is that it’s all right in most cases. The sensitivity of a sensor decreases quadratically with the distance of a moving object. If the two people sleep mostly on their respective halves of the bed, and if the sensors are placed at the opposite sides of the bed, as is shown in the picture above, the influence of the partner’s movement is negligible, and the sleep phase measurement is mostly correct.

Fair enough, but…

This simple answer is not good enough for some of our users though. If one partner is tossing too heavily in his bed, it actually may break the other’s record. Also, when two lovebirds cuddle too closely in the bed, or when the sensors ale placed too close to each other, the activity of the two people gets mixed, and the resulting record may be meaningless. So far, we have been able to record only the compound activity of the two lovebirds. It has been impossible to extract their individual sleep phases.

Pair tracking is the ultimate answer

No mixed records any more! We proudly announce a new feature in Sleep As Android, called Pair tracking, which aims to overcome this drawback, so that the both partners can enjoy the advantages of sleep phase detection in its full.

It works like this: The both partners measure the sleep with a phone (or Sleep Phaser) on their respective sides of the bed. Each sensor receives a mixed signal, as explained above. However, each of the two sensors measures the situation from a different perspective – there is always a stronger signal of the closer sleeper mixed with a weaker signal from the remote one. If we have the both signals, we are able, to a great extent, infer the separate activity of each of the two sleepers, clear each one’s record from the influence of the other partner, and hence make the sleep phase detection work again. It works with any combination of our supported devices – Sleep Phaser, smart wearables, sonar, or in-phone accelerometer.

Technically, when Pair tracking is activated, the two mobile phones connect to each other, and they keep exchanging the current activity data throughout the whole night. Each phone knows the activity measurement from the both perspectives, and can effectively subtract the partner’s activity from it’s own measurements.

Let’s see it in action

Let’s demonstrate it on a simple example, based on the real data. Here is an actigraph representing one night’s activity of Alice, the heroine of our story.

This is her ideal pure activity as if measured by a sensor attached to her body. It is a pretty typical sleep record, we can see there several phases of low and high activity.

However, she shares her bed with Bob, who had a heavy dinner and sleeps badly tonight. He keeps tossing and turning on his bed heavily all the night. If Alice uses a contactless sensor for her sleep tracking, and does not have Pair tracking activated, her actigraph will be strongly influenced by Bob’s movement and may look something like this:

Her deep sleep phases are mostly erased by Bob’s movement. It will affect her sleep statistics, as well as the functionality of smart alarm and other features.

However, if she turns our new Pair tracking feature on, her phone connects to Bob’s. It will continuously receive updates of Bob’s activity, and can almost completely discount it off her chart. The resulting record will look something like this. Distinct true sleep phases again. The dashed gray line represents the original ideal record for comparison.

Furthermore, it does not matter what sensor is Alice or Bob using. Our algorithm normalizes data from different sensors to the same amplitude and variance, so they can be combined freely. Either side can use any of our supported sensors/wearables and pair tracking will still work.

Looks too good to be true? Well, try it yourself, and see. Just update your Sleep As Android, check how to turn the sleep tracking on in our documentation, and enjoy your sleep phases undisturbed, no matter what you and your partner are doing throughout the night.