No abstract

This well-structured and well-explained paper presents DataPlanner (DP), which provides a client-sided non-intrusive module that interacts with the ABR logic of HTTP Adaptive Streaming systems to take data usage into account data usage when selecting the quality of retrieved video segments. This module enables clients, for example, to take limited per-session data budgets into account in the selection of video segments' quality without affecting the video quality to the same extent as earlier proposals. Addressing the issue of a user's total data budget would be an interesting extension to avoid DP's need for users' per-session decision making.DP adapts its decision to the actual consumed bandwidth and the size or quality video segments. The paper studies 2 adaptation approaches and considers that adaptation can be performed either by capping the qualities that are offered by the server (cap-beforehand) or the qualities that the client requests (cap-afterwards). The two adaptation approaches are called DP-Q and DP-T, where DP-Q bases its decisions on a segment quality metric (VMAF in the paper) and DP-T based them on segment complexity that was found in the literature (reference [36] in the paper).

No abstract