Date Awarded

2013

Document Type

Dissertation

Degree Name

Doctor of Philosophy (Ph.D.)

Department

Computer Science

Advisor

Gang Zhou

Abstract

In today's world of ubiquitous Smartphone use, extending the battery life has become an important issue. A significant contributor to battery drain is wireless networking. Common usage patterns expect Smartphones to maintain a constant Internet connection which exacerbates the problem.;Our research entitled A Network Traffic Approach to Smartphone Energy Savings focuses on extending Smartphone battery life by investigating how network traffic impacts power management of wireless devices. We explore 1) Real-time VoIP application energy savings by exploiting silence periods in conversation. WiFi is opportunistically placed into low power mode during Silence periods. 2.) The priority of Smartphone Application network traffic is used to modifiy WiFi radio power management using machine learning assisted prioritization. High priority network traffic is optimized for performance, consuming more energy while low priority network traffic is optimized for energy conservation. 3.) A hybrid multiple PHY, MAC layer approach to saving energy is also utilized. The Bluetooth assisted WiFi approach saves energy by combining high power, high throughput WiFi with low power, lower throughput Bluetooth. The switch between Bluetooth and WiFi is done opportunistically based upon the current data rate and health of the Bluetooth connection.;Our results show that application specific methods for wireless energy savings are very effective. We have demonstrated energy savings exceeding 50% in generic cases. With real-time VoIP applications we have shown upwards of 40% energy savings while maintaining good call quality. The hybrid multiple PHY approach saves more than 25% energy over existing solutions while attaining the capability of quickly adapting to changes in network traffic.

DOI

https://dx.doi.org/doi:10.21220/s2-wy0s-7w61

Rights

© The Author

Download

Share Feedback

Share

COinS