Reducing energy consumption of RNC based media streaming on smartphones via sampling

Abstract

There have been lots of research efforts on applying random network coding (RNC) technology in media streaming for increased reliability, throughput, etc. In an RNC based media streaming system, it is critical to optimize its RNC implementation; otherwise, the system cannot benefit from RNC due to its high computational cost. For example, an RNC decoder implemented in streaming applications on smartphones may exhibit excessive energy consumption draining batteries too quickly. In this paper, we deal with reducing the energy consumption of RNC based media streaming applications on smartphones especially in presence of other resource-competing applications. To reduce the energy consumption of RNC applications, we try to control the processor clock frequency via manipulating the frequency controllers in smartphone operating systems and the manipulation is accomplished through regulating the processor utilization for RNC applications. To estimate the processor utilization for RNC applications, we rely on a simple sampling approach, i.e., reading system files on a regular basis. Through experimental results, we show that our proposal reduces significantly the energy consumption of RNC applications on smartphones in presence of other intervening applications.