The definition of context is: The interrelated conditions in which something exists or occurs.

These conditions are very dynamic for mobile devices, because we take them with us everywhere we go. With billions of mobile devices being used worldwide, there is a huge connected network of sensors with an enormous amount of unprecedented possibilities yet to explore.

These possibilities are personal, public and social. Especially using the context socially will help users make faster and better informed decisions in both personal and professional life.

“Context from sensors”

The contextual information for mobile devices is largely inferred from sensor data. These devices are getting packed more and more with sensors:

• Microphone - sound level, recording, recognition
• Camera - colors, barcode scanning, image recognition
• GPS - location, direction, speed
• Accelerometer - movement, fall, vibrations
• Ambient light sensor - light intensity
• Proximity sensor - object proximity
• Magnetometer (digital compass) - direction relative to earth's north

Some examples of contextual information that can be inferred from these sensors are:

• Location - where are you?
• Activity - what are you doing?
• Presence - are you available for communication?
• View direction - what are you looking at?
• Weather - what is the weather?
• Health - what is your health?
• Traffic - what is your traffic situation?

Combining this contextual information together with existing (map data, calendar information) and/or historical information (old positions, previous activity) brings great opportunities for new mobile applications and services. Applications will be able to deliver the information you need, will be able to adjust the user interface to the current context and offer new ways of interaction.

“Enabling augmented reality”

Precise location with view direction is becoming a very important piece of contextual information, because it links the user and the surrounding objects. This is for example needed in augmented reality - the technology to combine real-world and computer-generated data. (video impression on augmented reality in the future)

To make augmented reality usable, high position accuracy is required, especially in urban areas and inside buildings. This accuracy might be lower when natural feature tracking has sufficiently evolved. Using a 3-axis accelerometer and an angular rate sensor (for the yaw) one could in theory determine an individual’s location. However, even the smallest errors in acceleration get exponentially larger when calculating the velocity and position, excluding this as a valid option for now.

“Dead reckoning for position”

Most of the research on improving mobile positioning is now focused on using the magnetometer for heading and using accelerometers to determine the traveled distance. Using both heading and distance to calculate the current position is called dead reckoning.

Using traditional pedometers to calculate the distance is not very accurate, because these often depend on a fixed step size. Recent research focuses on using accelerometers to detect steps and determine the size of each step.

In the near future algorithms in mobile software will use the information quality of multiple sources to calculate the best mix to determine the current position of the user:

• Using GPS outdoor to calibrate other sensors that will be used indoor;
• Blending GPS and dead reckoning information to improve outdoor positioning in urban areas with large multipath effects;
• Using triangulation on radio communication for indoor positioning.

“6-axis sensor for view direction”

The HTC Dream G1 (Google Android) and probably also the next iPhone contain a 6-axis sensor. This is a 3-axis magnetometer and a 3-axis accelerometer combined. This makes it possible to correctly detect the acceleration motion, gravitational direction and magnetic direction at every device orientation (portrait, landscape, flat, upright). One can calculate a 3D view direction by using both the gravitational and magnetic direction. The 6-axis sensor used by HTC is the AK8976A from Asahi Kasei.

Nokia is also shipping its high-end models, like the Nokia N97, with both a magnetometer and accelerometer. Mobile device suppliers not yet embedding these sensors are probably not seeing the great potential value for their buyers.