Context Detection is the next MEMS Wave

Internet of Things Leave a Comment

Summer is almost over, but I must still be in beach mode: Thinking about the evolution of sensing and the IoT in mobile applications, smart watches, fitness bands, and other wearable devices has me seeing waves rolling onto the sand.

The first wave hit the beach and brought with it a subtle irony: MEMS accelerometers, manufactured in silicon—processed sand—delivering the ability to sense motion. At the time, this new capability encouraged the construction of simple pedometers that counted steps and estimated distance traveled. Coupled with GPS and mapping software, often over the Internet, these little silicon machines could monitor both activity and location.

The next wave came ashore, and with it came better, more accurate accelerometers and gyroscopes that could more precisely monitor activity and better recognize change of direction. In fact, this increased accuracy could even compensate for periods of satellite blindness – in tunnels, parking garages, and urban canyons, for example.

Subsequent waves have washed pressure sensors, magnetometers, and other environmental sensors into our lives, further improving accuracy – by detecting altitude, absolute direction, and compensating for temperature and humidity effects. These incremental waves have been more than small ripples; they have created applications with increasing accuracy and value to both consumers and industry.

And the wave coming now to crash on the beach is a big one: it ties together multiple cost-effective MEMS sensors, such as a 3-axis accelerometer, a pressure sensor, and a microphone, with algorithms that can recognize and track user-activity modes (these could be walking, standing still, jogging, and climbing up/down stairs for fitness tracking, detecting gestures for a more natural user interface, or identifying the state of a device to switch between power modes, device carry positions and so on).

These context-detection algorithms can be designed to use any available sensor data independent of the underlying technology in the devices. In fact, the algorithms are more reliant on performance parameters such as the probability of false detection of context and latency in detecting transition between various user modes and

The beauty of this coming wave is that it will enable more natural interactions with our devices as well as new types of applications, improved productivity, and improved reliability – while consuming much less power. And this wave may be accompanied, or followed, by a wave delivering contextual audio, that puts information into appropriate contexts.

These context-detection algorithms are something you’re going to want to learn more about, if you are now—or planning in the future—to design applications for the Internet-of-Things movement.


There will be a session on Contextual Awareness presented by Dr. Mahesh Chowdhury at the free STMicroelectronics Developers Conference, in Santa Clara on Oct 4. The conference will also host a  presentation by Dr. Paul Beckmann of DSP Concepts that will talk about his vision of how sound will be incorporated into the technology of the future.