查看文章

With the help of big data analytics, predictive resource allocation (PRA) techniques for video on demand (VoD) have been recognized as promising methods to save time-frequency resources, for a number of VoD packets can be transmitted in good channels in advance to avoid the predicted transmissions in bad channel conditions. With the increasing demands on a fantastic user quality of experience, a smooth playback and a low start-up delay are of equal importance to the emerging VoD with high fidelity, which inevitably leads to a critical delay requirement of VoD packets. However, the issue of resource estimation with quality of service (QoS) requirements is still an unsolved puzzle in PRA. In this paper, we propose a martingales-based physical resource block (PRB) abstraction method, where the random characteristics of the service process are embedded in the minimum PRB consumption. Based on the …