Mobile Edge Decoding for Saving Energy and Improving Experience

Abstract

Video has become a fundamental network application, such popular and emerging application attracts a variety of attentions from both industry and academia. There exists a gap for mobile devices in wireless networks to obtain the same user experience as those in wired networks because constrained computational capability and limited power supply. Mobile edge computing may be a promising technique to offload computational tasks from low-end mobile devices to nearby highperformance servers. However, it is infeasible for video decoding under existing mobile edge computing architecture because both transmitting undecoded video and receiving decoded video will lead to unaffordable bandwidth consumption of current wireless networks. This paper analyzes popular real-time ultra-highresolution video decoding for massive low-end mobile devices in wireless networks, and solves two problems, video decoding architecture design and edge decoding ratio (EDR) determination. First, a novel mobile edge decoding (MED) architecture is introduced. Second, a feasibility condition of the mobile edge decoding system is proposed on the basis of edge decoding ratio. Third, an optimization problem with the core variable EDR is proposed to maximize user experience and minimize energy consumption. Forth, a low-complexity and suboptimal mobile edge decoding scheme (MEDS) with proved computational complexity is provided. The performance results show that the proposed MEDS can obtain better performance in terms of energy saving and user experience improving compared with that of existing schemes. This paper introduces a theoretical basis to be potentially used as a useful guidance to design and implement future emerging video decoding applications based on edge computing technique.