Adaptive video streaming requires efficient bitrate ladder construction to
meet heterogeneous network conditions and end-user demands. Per-title optimized
encoding typically traverses numerous encoding parameters to search the
Pareto-optimal operating points for each video. Recently, researchers have
attempted to predict the content-optimized bitrate ladder for pre-encoding
overhead reduction. However, existing methods commonly estimate the encoding
parameters on the Pareto front and still require subsequent pre-encodings. In
this paper, we propose to directly predict the optimal transcoding resolution
at each preset bitrate for efficient bitrate ladder construction. We adopt a
Temporal Attentive Gated Recurrent Network to capture spatial-temporal features
and predict transcoding resolutions as a multi-task classification problem. We
demonstrate that content-optimized bitrate ladders can thus be efficiently
determined without any pre-encoding. Our method well approximates the
ground-truth bitrate-resolution pairs with a slight Bj{\o}ntegaard Delta rate
loss of 1.21% and significantly outperforms the state-of-the-art fixed ladder.

直接预测每个预设比特率的最佳转码分辨率以实现高效比特率阶梯构建。采用时间注意力门控循环网络来捕捉时空特征，并将转码分辨率预测作为一个多任务分类问题。我们证明了通过此方法可以有效地确定内容优化的比特率阶梯，且不需要任何预编码。我们的方法近似于地面实况比特率 - 分辨率对，Bjøntegaard Delta 率损失为 1.21％，并且明显优于现有的固定阶梯。

高效的逐个标题比特率梯度预测的最佳转码分辨率

Optimal Transcoding Resolution Prediction for Efficient Per-Title  Bitrate Ladder Estimation

We consider the optimal design of a scheduling policy for adaptive video
streaming in a wireless network formed by several users and helpers. A feature
of such networks is that any user is typically in the range of multiple
helpers. Hence, in order to cope with user-helper association, load balancing
and inter-cell interference, an efficient streaming policy should allow the
users to dynamically select the helper node to download from, and determine
adaptively the video quality level of the download. In order to obtain a
tractable formulation, we follow a "divide and conquer" approach: i) Assuming
that each video packet (chunk) is delivered within its playback delay ("smooth
streaming regime"), the problem is formulated as a network utility maximization
(NUM), subject to queue stability, where the network utility function is a
concave and componentwise non-decreasing function of the users' video quality
measure. ii) We solve the NUM problem by using a Lyapunov Drift Plus Penalty
approach, obtaining a scheme that naturally decomposes into two sub-policies
referred to as "congestion control" (adaptive video quality and helper station
selection) and "transmission scheduling" (dynamic allocation of the helper-user
physical layer transmission rates).Our solution is provably optimal with
respect to the proposed NUM problem, in a strong per-sample path sense. iii)
Finally, we propose a method to adaptively estimate the maximum queuing delays,
such that each user can calculate its pre-buffering and re-buffering time in
order to cope with the fluctuations of the queuing delays. Through simulations,
we evaluate the performance of the proposed algorithm under realistic
assumptions of a network with densely deployed helper nodes, and demonstrate
the per-sample path optimality of the proposed solution by considering a
non-stationary non-ergodic scenario with user mobility, VBR video coding.

本文研究了无线网络中自适应视频流调度策略的优化设计问题，其中采用了 “分而治之” 的方法和 Lyapunov Drift Plus Penalty 方法，提出了两个子策略解决拥塞控制和传输调度问题，并提出了一种自适应估计最大排队延迟的方法。