Net Insight enables contribution over unmanaged networks

Swedish media infrastructure provider Net Insight has extended its existing remote production platform to support video contribution over unmanaged IP access networks.

This is designed to tap into the emerging field of video contribution from second-tier sporting or other events that may have a substantial global audience, but not enough people locally to justify deployment of dedicated infrastructure. 

According to Net Insight, the new Nimbra VA 210 will enable media service providers to address specialized or niche markets currently out of their reach because the costs of first-mile connectivity are too high. The product extends Net Insights’s existing Nimbra MSR launched in 2010 for fixed networks, which was used during the London Olympics to bring back high-quality camera feeds to Sweden and other countries, saving money for broadcasters.

Now, according to Net Insight’s business development director Per Lindgren, with OTT distribution extending reach for broadcasters, the content market is opening up for tier two or three live events with a big potential global audience. For these events, the cost of production has been prohibitive so far, especially for contribution, because of the high cost associated with live satellite or fiber links. Therefore, service providers are increasingly looking to deliver video of such events over the Internet via fixed or mobile broadband access links, but they want guaranteed Quality of Service.

The Nimbra VA 210 is designed to enable this QoS over unmanaged IP networks, bringing content back to a PoP (Point of Presence), where it can enter a media service network built with Nimbra MSR or other routers.

“This gives you high quality with low delay and so can change workflow for tier two and three event coverage,” said Lindgren. “You will see a lot more live events.”

The Nimbra VA 210 uses two techniques to deliver QoS. First, it uses content-aware Forward Error Correction (FEC) within MPEG streams to protect crucial I-frames that provide reference for the following predicted frames that only contain picture elements that have changed from the immediately preceding frame. It combines this with the second technique, selective retransmission over the UDP (User Datagram Protocol), which only transmits packets that have not already been received.

Net Insight has been fast off the draw here, as, until recently, content aware FEC was a research topic. The key point is that while over an end-to-end link, there is not usually enough time within a delay sensitive application like video streaming to retransmit packets that have been lost, it is possible over single hops between routers. The maximum acceptable one-way delay for interactive video is 150ms, while, on a router hop, a retransmission will usually only take a few milliseconds.

Therefore, it is possible to retransmit lost packets on a hop-by-hop basis. With content-aware FEC, this is done by storing packets in the cache of each router on an end-to-end link until its arrival has been confirmed by the next router. Each packet is sent once uncompressed across each link, plus one or more encoded versions merely containing a pointer to the cache on the previous router where that is stored. In the event of the full, uncompressed version not being received by a downstream router, the router opens an encoded version (if there is one) and then requests transmission of that packet uncompressed from the router immediately upstream.

This repeats until the number of encoded versions of the packet is exhausted, by which time it would be too late for retransmission. In this way, the degree of protection for a given packet can depend on its degree of priority, which would be high for an I-frame.

This may not be exactly how Net Insight has deployed content-aware FEC, but illustrates the underlying principle. It relies on the fact that packets are dropped, as a result of network congestion or otherwise, over one specific hop, so the error recovery can drill down to that hop and provide high resilience with low delay and less bandwidth cost.

It can either improve video quality, or reduce bandwidth, by at least 20 percent compared with conventional FEC operating at the raw data level.