Power Saving Strategies for Two-Way, Real-Time Enabled Cellular Phones

Power Saving Strategies for Two-Way, Real-Time Video-Enabled Cellular Phones Jessica Julie Tran Co-Chairs of the Supervisory Committee: Professor Eve A. Riskin Electrical Engineering Professor Richard E. Ladner Computer Science and Engineering MobileASL provides low bandwidth and low complexity software video encoders to enable real-time video conversations on cell phones, therefore allowing people who are deaf to communicate in their native language, American Sign Language. This thesis presents two alternative power saving algorithms that utilize activity recognition to extend battery resources when using MobileASL. The first algorithm, called variable spatial resolution (VSR), adjusts the spatial resolution of transmitted videos. The second algorithm applies a technique called variable frame rate (VFR) and VSR to adjust both frame rate and spatial resolution of transmitted videos. My goals are to implement both power saving algorithms in the MobileASL software program; conduct a battery power study to determine the duration of the battery life while using each algorithm; and conduct a web-based user study to determine if participants could perceive changes in video quality. The battery power study revealed that running MobileASL without any power saving algorithms consumes 99.7% of the phone’s CPU and a full battery charge lasts on average 284 minutes. Implementing VFR or VSR algorithms separately extend the battery life to an average 307 and 306 minutes respectively and lowers the CPU usage to 26% and 32% respectively. Applying VFR and VSR algorithms together extend the battery life to 315 minutes and lower the CPU usage to 10%. The experimental design of the user study was a 2 x 2 within-subjects factorial design. Major findings include discovering a significant VFR*VSR interaction, (F1,15=5.3, p<.05), which led to determining that applying VSR reduces the extent to which participants perceive VFR to induce choppiness. Also, the application of VSR did cause participants to perceive blurry video quality,(F1,15=21.2, p<.003), and participants found the blurriness to be distracting, (F1,15=10.1, p<.01). The battery power study revealed that applying VFR, VSR, and both VFR and VSR all extend the battery life of a cell phone running MobileASL. Applying both VFR and VSR was found to extend the battery life the most. Therefore, the recommendation is for MobileASL to adopt the use of both VFR and VSR algorithms to extend the battery duration of the cell phone.