Scheduling Heuristics for Live Video Transcoding on Cloud Edges

Efficient video delivery involves the transcoding of the original sequence into various resolutions, bitrates and standards, in order to match viewers’capabilities. Since video coding and transcoding are computationally demanding, performing a portion of these tasks at the network edges promises to decrease both the workload and network traffic towards the data centers of media providers. Motivated by the increasing popularity of live casting on social media platforms, in this paper we focus on the case of live video transcoding. Specifically, we investigate scheduling heuristics that decide on which jobs should be assigned to an edge minidatacenter and which to a backend datacenter. Through simulation experiments with different QoS requirements we conclude on the best alternative.