Summary
We leverage online machine learning and next-generation congestion control to maximize Quality of Experience (QoE) for adaptive video streaming and video conferencing. Compatible with any HTTP-based application, our technology will significantly disrupt the Over-the-top (OTT) video market.Today, the dominant congestion control protocols (TCP and increasingly BBR) are one-size fits all and consequently fail to deliver good QoE. Our solution is based on something new: Performance-oriented Congestion Control (PCC), developed by Compira co-founder Prof. Michael Schapira and Prof. Brighten Godfrey from the University of Illinois at Urbana-Champaign (a key advisor). Unlike TCP, PCC changes transmission rates in line with predicted effect on performance and specific application needs - at a granularity of 10s of milliseconds. Rate selection is optimized by machine learning, enabling us to swiftly overcome network congestion within the critical 'last mile' of data delivery.
Unfold all
/
Fold all
More information & hyperlinks
| Web resources: | https://cordis.europa.eu/project/id/190125980 |
| Start date: | 01-05-2022 |
| End date: | 30-09-2024 |
| Total budget - Public funding: | 3 572 425,00 Euro - 2 489 148,00 Euro |
Cordis data
Original description
We leverage online machine learning and next-generation congestion control to maximize Quality of Experience (QoE) for adaptive video streaming and video conferencing. Compatible with any HTTP-based application, our technology will significantly disrupt the Over-the-top (OTT) video market.Today, the dominant congestion control protocols (TCP and increasingly BBR) are one-size fits all and consequently fail to deliver good QoE. Our solution is based on something new: Performance-oriented Congestion Control (PCC), developed by Compira co-founder Prof. Michael Schapira and Prof. Brighten Godfrey from the University of Illinois at Urbana-Champaign (a key advisor). Unlike TCP, PCC changes transmission rates in line with predicted effect on performance and specific application needs - at a granularity of 10s of milliseconds. Rate selection is optimized by machine learning, enabling us to swiftly overcome network congestion within the critical 'last mile' of data delivery.Status
SIGNEDCall topic
HORIZON-EIC-2021-ACCELERATORCHALLENGES-01-01Update Date
09-02-2023
Images
No images available.
Geographical location(s)
Structured mapping
