young-vlogger-streaming-a-live-video-6D8

Video streaming

Over the Top (OTT) video streaming services are growing rapidly, estimated to reach 2 billion active subscriptions by 2025. 

OTT video streaming comprises a multitude of service types

live-streaming.png
video.png

Live

channels

Cloud digital video recorder (cDVR)

live.png

These are consumed at home and on-the-go across different networks

1.png

Cable

4.png

Fiber to home

3.png

Public WiFi

5.png

LTE

2.png

5G

Reliance on internet means uneven QoE

Unlike traditional TV services, OTT services rely on the Internet, where good Quality of Experience (QoE) is hard to achieve.

Bad QoE leads to user frustration and reduced engagement with content and, in turn, churn, high support costs, and loss of advertising revenues. See this blog post for more details. While providers invest heavily in better servers and in placing CDN/cache nodes close to the users, widespread poor QoE persists.

One-size-fits-none content delivery

Different streaming services have utterly different performance requirements (e.g., low latency for Live vs. high throughput for VoD), and drastically different network conditions affect how users experience streaming content. However, content delivery across the Internet remains “one-size-fits-all,” which fails to adapt to the service’s needs and the prevailing network conditions. This “one-size-fits-none” leads to poor QoE.

Compira Labs’ solution: Personalized, QoE-oriented content delivery

The Compira Labs solution provides online rate adaptation at the streaming service’s traffic sources, which is tailored to the service-specific performance requirements and the local network conditions at each individual traffic origin. By incorporating this solution into the server-side network stack at edge nodes, operators, service providers, and CDNs, can continuously optimize QoE for subscribers. The result: higher user satisfaction and engagement, and reduced churn and support overhead.

Solution Components

Compira Edge (CE)

The CE performs online-learning rate selection by employing Performance-oriented Congestion Control (PCC). The CE also continuously collects network statistics and supports  remote configuration of the PCC element. It is available as a Linux kernel module for TCP and also available for QUIC based delivery. The CE is installed on video caches (CDN edge nodes).

Compira Cloud  (CC)

A ML-powered, cloud-based analytics engine. The CC receives a continuous stream of network statistics from the CE agents and leverages it for

  • Customizing individual traffic sources to adapt to the service’s needs and their local network conditions

  • Providing visibility into network behavior via a multi-tenant SaaS dashboard.