Data-Driven Bandwidth Prediction Models and Automated Model Selection for Low Latency

Abdelhak Bentaleb, Ali C. Begen, Saad Harous, Roger Zimmermann

Research output: Contribution to journalArticlepeer-review

5 Citations (Scopus)

Abstract

Today's HTTP adaptive streaming solutions use a variety of algorithms to measure the available network bandwidth and predict its future values. Bandwidth prediction, which is already a difficult task, must be more accurate when lower latency is desired due to the shorter time available to react to bandwidth changes, and when mobile networks are involved due to their inherently more frequent and potentially larger bandwidth fluctuations. Any inaccuracy in bandwidth prediction results in flawed adaptation decisions, which will in turn translate into a diminished viewer experience. We propose an Automated Model for Prediction (AMP) that encompasses techniques for bandwidth prediction and model auto-selection specifically designed for low-latency live steaming with chunked transfer encoding. We first study statistical and computational intelligence techniques to implement a suite of bandwidth prediction models that can work accurately under a broad range of network conditions, and second, we introduce an automated prediction model selection method. We confirm the effectiveness of our solution through trace-driven live streaming experiments.

Original languageEnglish
Article number9154522
Pages (from-to)2588-2601
Number of pages14
JournalIEEE Transactions on Multimedia
Volume23
DOIs
Publication statusPublished - 2021

Keywords

  • ABR
  • CMAF
  • DASH
  • HTTP adaptive streaming
  • bandwidth prediction
  • chunked transfer encoding
  • low latency

ASJC Scopus subject areas

  • Signal Processing
  • Media Technology
  • Computer Science Applications
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Data-Driven Bandwidth Prediction Models and Automated Model Selection for Low Latency'. Together they form a unique fingerprint.

Cite this