Adaptive coding with CPU energy conservation for mobile video calls

Abstract

Video calling incurs very high power consumption on mobile platforms usually powered by capacity-constrained batteries. To remedy this problem, we develop a cross-layer energy-aware optimization framework to reduce CPU energy consumption for mobile video calls. We explore the texture similarity between neighboring macro blocks during H.264 coding and control the quality-complexity tradeoff with a single self-adaptive parameter. Dynamic Voltage and Frequency Scaling (DVFS) is then applied to reduce energy consumption and guarantee the stringent one-way delay in real-time applications. Experimental results show that considerable CPU energy saving is achieved while a high Quality of Service (QoS) is preserved for user satisfaction. © 2012 IEEE.