← All reviews

A Web-Based User Survey for Evaluating Power Saving Strategies for Deaf Users of MobileASL

Jessica J. Tran, Tressa W. Johnson, Joy Kim, Rafael Rodriguez, Sheri Yin, Eve A. Riskin, Richard E. Ladner, Jacob O. Wobbrock · 2010 · Proceedings of the 12th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2010) · doi:10.1145/1878803.1878825

Summary

This paper presents research on power-saving video compression algorithms for MobileASL, a project enabling real-time, two-way video communication on mobile phones so Deaf people can communicate in American Sign Language. Running video calls drains phone batteries rapidly — only 284 minutes without optimisation. The authors developed two new algorithms: variable spatial resolution (VSR), which downsamples video frames to 1/4 size during non-signing periods, and a combination of VSR with the previously developed variable frame rate (VFR), which reduces frame rate from 10-12 fps to 1 fps when not signing. Both algorithms use activity recognition — baseline differencing of pixel luminance between consecutive frames — to detect when a person is actively signing versus passively listening. VFR alone extended battery life to 307 minutes (CPU usage reduced to 26%), VSR alone to 306 minutes (CPU to 32%), and both combined to 315 minutes (CPU to 10%). To evaluate perceived video quality, the researchers designed a linguistically accessible online survey for Deaf participants, presenting instructions in both English text and ASL video, using a vertical 5-point Likert scale (reflecting ASL's spatial grammar), and distributing through Sorenson Communications' Deaf customer network. The survey was completed by 148 ASL-fluent participants aged 18-75.

Key findings

Using aligned rank transform (ART) nonparametric factorial analysis on 596 data points, the study found that VFR produced perceived video choppiness (F(1,591)=80.94, p<0.001) and VSR produced perceived video blurriness (F(1,591)=131.57, p<0.001), as expected. However, a surprising and practically significant finding was that when both VFR and VSR were applied together, they ameliorated each other's negative effects. The VFR*VSR interaction was significant for both choppiness (F(1,591)=23.48, p<0.001) and blurriness (F(1,440)=9.38, p<0.05): VSR's blurriness "smoothed out" VFR's choppiness, while VFR reduced the distracting nature of VSR's blurriness. This is a fortunate result because the combined algorithm saves the most battery. The survey methodology itself yielded important lessons: the researchers consulted Sorenson Communications' marketing director for Deaf-accessible survey design, adopted a vertical Likert scale layout based on linguistic research suggesting ASL speakers may not assign the same spatial meaning to horizontal scales as English speakers, and provided ASL video interpretation of all questions alongside English text.

Relevance

This research addresses a practical barrier to communication equality: if mobile video calls drain batteries too quickly, Deaf people lose access to their primary language while away from power sources. The finding that combining two individually degrading compression techniques actually produces better perceived quality than either alone has broader implications for video accessibility — it suggests that trade-offs in video quality should be evaluated holistically rather than in isolation. For accessibility practitioners, the survey methodology is equally valuable: the careful attention to making research instruments linguistically accessible for Deaf participants — ASL video instructions, vertical Likert scales, culturally appropriate distribution channels — provides a model for conducting inclusive user research. The paper also demonstrates respect for Deaf culture by capitalising "Deaf" to denote cultural identity and framing the work around communication access rather than hearing loss.

Tags: deaf and hard of hearing · sign language · American Sign Language · mobile technology · video compression · video communication · battery life · accessible survey design · deaf culture