Chinese Version 中文版：您昨晚睡得好吗？
Every night before going to bed, I make sure I am accompanied by one thing; my sleep tracker. Every toss and every turn, the moments I am dreaming, the length of my sleep, and the time I am disturbed by the cat jumping on my bed; the device is capturing all my sleep patterns.
Regularly enjoying a good night’s sleep is widely understood as one of the most important factors in a healthy living and in improving performance in our day to day tasks. Over the past 50 years, the average self-reported sleep duration in the United States has fallen by 1.5-2 hours per night, with 38% of adults admitting unintentionally fall asleep during the day at least once per month, and weight gain being directly attributed to poor sleep patterns. The effects of sleep deprivation on the economy are enormous. A report in 2008 revealed that 1 in 4 Britons have had time off due to exhaustion from a lack of sleep.
Technology has always been an important factor in sleep tracking. Medical professionals have carried out sleep studies for a number of years, historically using cameras and wired electrodes to monitor those suffering from insomnia and similar sleeping disorders. This can help to construct a clearer image of how someone sleeps; if we can measure it, we can improve it.
Now this technology is becoming used more widely, with consumer devices giving troubled sleepers the means to self analyse their sleep patterns. There has been huge progress in the accuracy and availability of the tools used to measure sleep quality over the past years, with sensor-based devices measuring small disturbances during sleep and visualizing them in graphs. This has largely developed with the uptake of sensor hubs.
A sensor hub is a processor connected to multiple bio-metric and environmental sensors, running a data fusion algorithm while consuming a very small amount of power. Sensor hubs are now found in almost every smartphone, and their accuracy and low-power profile have also enabled them to be built into standalone wearable devices.
Wristbands have been one of the main driving forces of the wearable revolution, with numerous devices utilizing an ARM Cortex-M processor alongside a sensor hub to track movement. The two main use cases for these devices are daytime fitness tracking and sleep monitoring. Having previously used a smartphone to track my sleep patterns, I have been using an ARM Cortex-M3 powered Misfit Shine for the last four months. This has improved the amount I track my sleep , as I need not remember to turn the app on, worry about my smartphone battery, or place it under my pillow. Instead the device works continuously and at low-power, constantly tracking my sleep and daytime activity from my wrist, while lasting around four months before swapping batteries requiring a new battery (the Misfit Shine does not need charging). I am now able to see at what times I sleep best, and how external factors affect my sleep pattern. After a rich dinner, for example, my sleep pattern is disturbed, while eating before 8.30pm normally ensures I quickly fall asleep quickly.
The basic principal of operation is straight forward. Data from multiple sources is collected and ‘fused’ by the ARM processor to understand the profile of a user’s sleep. Data from the sensors about small disturbances to the wristband’s position is used to interpret the quality of sleep. to determine context of the user. In the case of sleep tracking, it is data about position and movement. Sensors are always on and continuously sampling, with the Cortex-M processor remaining in deep sleep mode until there is a change in the physical state requiring attention. The whole design is event driven which drives power consumption to the minimum and allowing for the longest possible battery life. Every few months we see new sensors are being added into the activity trackers, but the same principal of operation remains unchanged.Cortex-M processor within continues to execute the same functions.
A wide variety of these bands are available, each equipped with different functions and apps. One of the newest tracking bands to enter the market is the Cortex-M4 powered Microsoft Band. The device features a heart rate monitor, UV sensor, ambient light sensor and numerous other sensors to provide 48 hours of usage, monitoring not only sleep patterns but activities, messages, the weather and much more. The device is also supported by Android, iOS and Windows, meaning a huge range of connectivity for the device. Also popular is the Cortex-M3 powered Fitbit Flex which provides a longer battery life of five days. This has a much simpler display, relying more on the connectivity to a smartphone or PC via Bluetooth LE.
The challenge for these devices is the form factor. The size dictates the number of sensors that can be used inside, and the space for the battery. The tracker needs to have a processor to handle the sensors, data fusion, and communication with an external device. The processor must thus execute these tasks using minute amounts of power, to ensure that the battery lasts for days or months between charges and replacement.
These are the conditions for which the ARM Cortex-M processor family was created for. A set of 32-bit processors with extreme energy efficiency to handle low-power scenarios, the Cortex-M series processors are used in devices requiring leading MCU performance, a rich feature set (including built in memory protection unit and powerful interrupt control) and a large ecosystem of supporting software and tools. The Cortex-M4 is particularly popular and will become even more so in 2015, with instructions for signal processing and floating-point arithmetic which greatly enhances the performance of sensor-driven designs. In terms of physical implementation, suppliers are racing to provide ARM-based designs in the smallest possible silicon. Freescale, for example, is selling a Cortex-M0+ based processor which is just 1.6mm x 2.0mm, and at the same time, Atmel have just released the SAM L21, an ARM Cortex-M0+ based MCUs with power consumption as low as 40 µA/MHz in active mode.
Technology has advanced the design from a helmet with hundreds of wires to a bracelet with just three components inside (processor, sensor hub and radio). People are now able to track and monitor their sleep and see how external factors can influence sleep patterns. Armed with this information, it is becoming increasingly easy for people to understand factors affecting their sleep. With time, this will hopefully improve the amount of good quality sleep we have, and reduce the amount of time spent at home with exhaustion or at the doctors being examined.
What does the future hold for sleep tracking?
One could see multiple scenarios; a pillow with a built in sleep tracker or a button on a user’s pyjama with a built-in sleep tracker. Given the reduction in size of MCUs and the abundance of sensors, there is no limit to where this technology will be deployed. Now that we have the tools to measure, it is up to us to use them to improve!
Sounds obvious now, but I hadn't thought about experimenting like that!
I will try to take notes about when/how/why I go to bed and see what kind of impact it has. I just need to try and only vary on parameter at a time
You can indeed detect sleep apnea with the noise if you are alone (breathing pauses), or by measuring a drop in blood oxygen concentration. For the later, you would need to wear something on the finger at night...
You could also probably use something around the bust which would measure your rib cage moving. I there is already something available for sport activities measuring heart rate and other things.
The CPAP, BiPAP or VPAP is then another machine to "wear" and carry around when traveling.
Honey on the milk is much more efficient !
wondering if this can be the good tool to track and alert on sleep apnea ...which is a main danger with some numerious side effects?
The way it helped me was being able to connect the lack of good sleep to stuff that happened during the day or before sleep.
you then experiment and see the result.
the clear one is a big meal before going to sleep , you see the impact right away.
For me, it was also the temperature setting. A bit too low and I will wake up many times during the night since my body is trying to tell me that it is cold.
a trip early the next day, you bet, I will wake up at 4, 5 and then 6am.
I fixed that one by setting four alarms in sequence.
the fact that you can experiment and see the outcome is very powerful.
When you travel sleep pattern change for the better and worse. you can then check for the differences with the home setting.
Most of all, it is very personal. it is your experiment, your bed, your sleep.
as opposed to some one just saying, drink milk before going to bed, it is a known remedy!
I was wondering how we could use the data to influence our sleep pattern.
Knowing is one thing, but I think we need to be able to change if we want improvements.
It could be fairly easy to change the number of hours spent in bed.
But when the tracker tells you you wake up 30 times during the night, I wonder how much influence when you are not really awake...