Monthly Archives: November 2013

Can mobile devices enable primitive sentience in the global computing network?

By: Jonah McLeod, Kilopass Technology Inc.

In a short video entitled The Nexus of Forces 2013, Special Report Update, “Chris Howard, Research VP at Gartner Group, made the following observation concerning the ongoing trend in computing. “The history of computing is about emancipating the computer from the box, from the mainframe and its architecture out to distributed computing and now into mobile computing, where we’re pushing more of the functionality out to the edge.” Today, the proliferating numbers of smart devices a large percentage of the world’s population carries represents the edge that Howard described. These computing devices are performing an increasing amount of decision-making independent of its owner.

What’s making this possible is the increasing numbers of sensors providing these devices with the inputs necessary to take on this initiative:  a tri-axis accelerometer, gyroscope, and magnetometer; compass; multiple cameras and microphones; as well as sensors that monitor ambient light, pressure, proximity, temperature, and humidity. Most recently Apple introduced a fingerprint sensor, which ties the phone to the owner and prevents anyone else from using the device.  Combined with software, these sensors are enabling smart devices to make determinations that augment the decision making of its owner or others mining the data from these devices in the cloud.

In the majority of today’s mobile devices, sensors interface directly to the host processor, which performs sensory data processing and event detection.  In newer devices such as the Apple iPhone 5s and Motorola X8, a dedicated sensor processor, the M7 and contextual processor, respectively, now take on this role to improve the power and processing efficiency of the smart device. Motorola claims their architecture eliminates the need for two additional batteries that would otherwise be required.  Putting all the sensors under control of a dedicated processor results in power savings. Software that manages which sensors provide information the user will likely want, given his/her current location, makes this possible.

Sensor Platforms explains that by offloading the management of a mobile device’s on-board sensors from the applications processor to a separate embedded CPU such as a Cortex M3, power consumption can be reduced over 70 percent. This savings can be further improved with a processor designed for power management, such as found in the newest Apple and Motorola devices.

To put mobile device sensors’ role back into the perspective of the edge that Chris Howard detailed earlier, consider the Nericell project carried out by Microsoft researchers in India. Nericell—a play on the word ‘Nerisal’, which means ‘congestion’ in Tamil—uses the accelerometer, microphone, GSM radio, and/or GPS receiver in smart phones to monitor road and traffic conditions:  detecting potholes, bumps, braking, and honking in Bangalore, India. This data is then uploaded to the cloud.  To save power in the mobile devices, Nericell uses relatively low energy resources—cellular radio or accelerometer—to trigger more expensive resources—GPS or microphone.

The Nericell project resembles Waze, a community-based traffic and navigation app in the U.S. that provides traffic congestion information to the cloud.  Like Waze, Nericell collects data from a large number of users’ mobile devices to perform sensing and processing and to transmit the data to a server for aggregation. The project estimates a daily drive time of four hours. During this time, brake and bump detection are active throughout while audio and GPS are triggered for specific events, for example to determine current location.  The project estimated that adding the capability to an HP iPAQ would result in an additional 10 percent energy burden on the mobile device.

A prototype of Nericell, running on Windows Mobile 5.0 Pocket PCs have yielded promising results offering a cost-effective alternative to deployment of dedicated sensors on vehicles and/or on the roadside. The ability to harness the computing power and the ubiquitous installed base of sensors on mobile devices is providing the global connected computing network a primitive form of sentience.