1. Realizing the potential of mobile devices as experimental devices: Human computer interface and performance considerations
Chiung Ching Ho & C. Eswaran
Faculty of Computing and Informatics
23/09/13

2. Presentation outline Introduction Research question Related works
Methodology and description of experiments
Experimental work done
Analysis
Conclusion

3. 1.0 Introduction 6 billion devices and rising as of 2012
6 billion mobile devices and rising, in use in all spheres of life

4. Presentation outline Introduction Research questions Related works
Methodology and description of experiments
Analysis
Conclusion

5. 2.0 Research questions
To what extent can mobile devices be used as experimental devices?
How does screen size, processing power and battery life impact the usage of mobile devices as experimental devices? 1 2

6. Presentation outline Introduction Research questions Related Works
Methodology and description of experiments
Analysis
Conclusion

7. 3.0 Related Works “Don’t key in data while moving”
Problems When Moving
> Errors
> Time
Solution
Small form factor, pen , robustness
Non visual feedback
Gesture and speech controls
Adaptive widgets
“Don’t key in data while moving”

8. Increased screen size = More space for data and screen controls vs higher power consumption
Smaller screen can be mitigated using one handed navigation, horizontal display, tabbed and pivot controls, zooming, new ways of navigation
Trade-off between large screens and small screens and methods for optimizing small screen space

9. Increase of processor speed and memory size from 1996 to 2012 on log scale ( from 24 MHz and 16 Mb RAM to 1.5 GHz and 2 Gb Ram )

10. Presentation outline Introduction Research questions Related Works
Methodology and description of experiments
Analysis
Conclusion

11. Experiment No.1 : HTK based smart device user verification
Processing power and memory
206 MHz and 32Mb RAM
Limited processing power prevented the use of computationally expensive algorithm
Screen size
240 by 320 pixels
Limited space for widgets and data
Battery life
Limited
Needed to be connected to a power source
Lack of suitable mobile machine learning library
Computation was done on the server machine

12. Processing power and memory
1 GHz and 768 Mb RAM
Increased processing power allowed the use of computationally expensive algorithm
Screen size
480 by 800 pixels, double the screen size as compared to the device used in Experiment 1
Limited space for widgets and data
Battery life
Limited
Needed to be connected to a power source frequently owing to the use of accelerometers, and larger screen size. This slowed down the data capturing process
Lack of suitable mobile machine learning library
Computation was done on the server machine, as process involved signal processing and classification
Graphing was done on the mobile device
Experiment no. 2: Accelerometer based gait smart device user verification

13. Presentation outline Introduction Research questions Related Works
Methodology and description of experiments
Analysis
Conclusion

14. 5.0 Analysis
Increased drain on battery due to large screen size and connectivity needs
Increased computational resources and useable screen area for controls and visualization

15. Dedicated mobile machine learning libraries
Adaptive connectivity that is enabled on demand
Longer lasting battery

17. 6.0 Conclusion
With the easy availability of mobile devices, and the array of sensors and connectivity options , mobile devices have a promising future as experimental devices
To realize this opportunity, advancement in the following areas needs to be made
Battery life
Screen independent controls (voice or gesture)
Dedicated mobile libraries for machine learning, signal processing
Interoperability with other ‘plug and play’ sensors

18. Thank you!
Q and A