DeLoad: Demand-Driven Short-Video Preloading with Scalable Watch-Time Estimation

TL;DR

DeLoad is a novel preloading framework for short videos that dynamically adjusts download tasks and uses deep reinforcement learning to optimize user experience and bandwidth efficiency, validated by extensive real-world data.

Contribution

It introduces dynamic task sizing and a practical watch time estimation method, enhancing preloading strategies for short video streaming.

Findings

Significant QoE improvements (34.4% to 87.4%)

Increased user watch time by 0.09% after deployment

Reduced rebuffering events and bandwidth consumption

Abstract

Short video streaming has become a dominant paradigm in digital media, characterized by rapid swiping interactions and diverse media content. A key technical challenge is designing an effective preloading strategy that dynamically selects and prioritizes download tasks from an evolving playlist, balancing Quality of Experience (QoE) and bandwidth efficiency under practical commercial constraints. However, real world analysis reveals critical limitations of existing approaches: (1) insufficient adaptation of download task sizes to dynamic conditions, and (2) watch time prediction models that are difficult to deploy reliably at scale. In this paper, we propose DeLoad, a novel preloading framework that addresses these issues by introducing dynamic task sizing and a practical, multi dimensional watch time estimation method. Additionally, a Deep Reinforcement Learning (DRL) enhanced agent is trained to optimize the download range decisions adaptively. Extensive evaluations conducted on an offline testing platform, leveraging massive real world network data, demonstrate that DeLoad achieves significant improvements in QoE metrics (34.4% to 87.4% gain). Furthermore, after deployment on a large scale commercial short video platform, DeLoad has increased overall user watch time by 0.09% while simultaneously reducing rebuffering events and 3.76% bandwidth consumption.