A Holistic Approach to Understand HTTP Adaptive Streaming

Research paper on applying the holistic approach to monitoring HTTP adaptive streaming, published at ACM MMSys 2023 in Vancouver.

Published in the Proceedings of the 14th ACM Multimedia Systems Conference (MMSys ’23), Vancouver, Canada, pages 333-337.

Paper (ACM)

Summary

This doctoral study proposes a holistic, cross-layer view of HTTP adaptive streaming: data from every layer, combined into one representation, to make root cause analysis and cross-layer optimizations easier.

The holistic view

Layers hide information from each other. When something goes wrong, the cause often looks as if it lies in another layer, which the paper calls the finger-pointing problem. QUIC blurs the layers even more, as one protocol combines congestion control, encryption and stream multiplexing.

The client application and the remote application, each with layers from link to application logic, are connected by a network with the lower layers only; data from all of them feeds one holistic cross-layer view

A holistic view gathers data from the client application, the remote application and the network, from every layer, and analyses it together. The data has to meet three requirements: enough of it, in a format that can be compared, and with a description of what it means. The qlog format and the qvis tool are a notable example of this idea.

Cite

@inproceedings{vandersanden2023holistic,
  author    = {Vandersanden, Mike},
  title     = {A Holistic Approach to Understand {HTTP} Adaptive Streaming},
  booktitle = {Proceedings of the 14th ACM Multimedia Systems Conference (MMSys '23)},
  address   = {Vancouver, Canada},
  pages     = {333--337},
  month     = jun,
  year      = {2023},
  publisher = {ACM},
  doi       = {10.1145/3587819.3593026},
}
BibTeX