The Future of Mobile Computing – Intel
July 11, 2011, IMEC Smart Phone Forum, San Francisco—Justin Ratner, CTO at Intel, talked about the increasing demand for computing. He noted that last year manufacturers shipped almost 10 billion transistors for every person on earth or a total of 74.5 quintillion transistors. The compound annual growth rate for transistors in this century has been averaging 68 percent.
New and useful apps are driving an increase in smart devices, ranging from phones tablets and notebooks and now even to TV. There is much more to come. A lot of the data for these devices now resides in the cloud, and data centers are handling larger and larger data sets. For example, YouTube has over 2 billion videos and runs about 25 petabytes per day. Facebook users access 8.6 billion pages accounting for 1.7 petabytes per day.
The aggregate for all data is over 152 TB per day, and half of all data in the world are on the net. All of these data are creating new challenges for the data centers. All data centers are adding more memory try to keep up with the data demands, and many are using more electrical resources that are locally available. At issue here is the need to store and use high volumes of data with much more user friendliness.
The personal computer has created an innovation model but that is now changing. In the past, the industry saw constant spiral up in processor and software capabilities. Over time, user inputs drove new features and capabilities at ever higher performance levels. The result was a horizontal division of technologies. Now, however, performance is measured by the user’s experience, and people are looking for intuitive and comfortable computing experiences. This changes the definition of design requirements.
The changes in requirements also changes the needed skill sets to address customer needs. Companies now need futurists and ethnographers to observe consumers and get to a second level of end-user experiences. These people are creating a formal user experience design practice, aggravating design engineers because the data they are collecting is all subjective.
Now this practice of observing users is embedded in the total design flow. The technical people complete the design and other people check on consumers validating the products. To date, all interconnected devices result from their work on smart TVs. The smart TV is not just a wall of screens nor is it just the user interface. The user interface is only part of the total experience. The change in user experience and in design methods has resulted in a range of results. Now, there are multiple form factors, features, and functions that can be produced by variations in the silicon.
We have moved to context-aware mobile computing. Location, proximity to others, activities, and associative interactivity can be coupled with multiple sensors. The hard sensors can determine orientation, physiological, and biological parameters. Adding audio to the measurements can enable environmental analysis. Combine the sounds with location data and you can tell what a person is doing and where they are.
Another new sensor will be video for object recognition and geo-location. The video can help the computer infer location through landmarks. Sensor fusion will mix the inputs from hard and soft sensors and activity algorithms to provide contextual awareness.
An example of this fusion is a trial with Fodor’s to create a personal vacation assistant. Sensor abstractions and data stores allowed the handheld devices to learn its user’s preferences and interests. The device then suggests alternative activities based on the combination of location and preferences. The result transformed the travel experiences for the trial users. Obviously, privacy and security are important issues that have to be addressed at the initial stages of the design for a product that seems to know what you are thinking.
The implications of a handheld personal assistant are daunting. The R&D groups have to develop low-power integrated sensors and figure out how to power always-on sensor subsystems. The inference engines need algorithm acceleration to operate in real-time environments. Data and workload sharing between the client and the cloud needs work and the new apps and services to operate outside of a tourist environment don’t exist yet.
The combinations of 24/7 sensors and context awareness has applications in other areas. Health and safety issues are some of the first areas for investigation. By measuring and tracking physiological and biological data, a medical professional (computer) could determine physical and psychological well being. Short-and long-term changes indicate a possible medical issue and the system could schedule you for a check up for specific conditions indicated by the changes. In extreme cases like a fall or other injury, the system could be your personal “On-Star” and automatically call emergency services.
