基于微云的移动视频预取系统

Cloudlet-based Mobile Video Prefetching System

现有的移动视频通话研究大多集中在视音频同步、码率调整和能量节约等方面,忽视了用户在公共AP间切换时视频通话服务的中断重连问题。针对这一问题,提出了基于微云的移动视频预取系统来尽可能地减少通话重连次数。该系统通过本地微云为移动的用户预取视频流,并设计贪心预取算法来提高预取命中率。同时,设计低概率多阶马尔科夫预取进程调度算法和起始AP选择算法来平衡贪心预取引起的微云资源消耗过多的问题。仿真实验表明,CMVP可以获得90%以上的预取命中率,使QoE显著提升。

Most researches on mobile video communication（MVC） {ocus on audio and video synchronization,bit rate ad- justment and energy saving, however, the change between APs which leads video communication service to interruption is not considered. We proposed a cloudlet-based mobile video prefetching system（CMVP） to this challenge. CMVP utili- zes cloudlets to prefetch video streams for mobile-phones. We designed a greedy prefetching algorithm to reduce the count of service interruption as far as possible. And to balance cloudlets＇ resource consumption, the low probability K to 1 markov prefetching progress schedule algorithm（LPMPPS） and first AP selection algorithm were designed. Our simu- lation results demonstrate that CMVP can access to more than 90% hit rate, and QoE can be improved remarkably.