Haivision, based in Montreal, co-founded the SRT Alliance alongside Wowza.
The SRT Alliance has built further momentum in its bid to establish a common standard for low latency video streaming by attracting a flush of 39 new members to take the total above 200.
Among new recruits joining founders Haivision and Wowza, alongside some big hitters like Microsoft, Comcast and Harmonic, are Imagine Communications, Net Insight, Red Bee Media, Telestream. This represents a pooling of resources among companies some of which originally proposed their own proprietary technologies for low latency streaming.
It includes those like Red Bee Media that provide platforms or services and are under pressure from customers to cut latency, especially for live streaming of premium sports where delay can ruin the experience when events are shown at staggered times to viewers within close proximity. “We are increasingly distributing video for high value live events, often on a temporary basis for customers that demand high quality over secure and reliable networks,” said Steve Russell, Head of Media Management and OTT at Red Bee Media, which provides managed services for broadcasting and media. “SRT enables us to expand our customers’ reach while adhering to our preference for open standards.”
The SRT protocol itself started as a proprietary offering but went open source at NAB 2017 and will be demonstrated on the Alliance’s stand this year at NAB 2019. Haivision and Microsoft will also be hosting free-to-attend panel sessions featuring well known broadcasters and industry leaders discussing their SRT implementation and the impact of the protocol on their operations.
ESPN is one broadcaster that has deployed the SRT protocol to stream from remote events via low-cost internet connections, in place of using traditional satellite uplink services, saving an estimated $8 million so far. Microsoft has also used SRT for professional quality live video productions using low-cost internet connectivity at hotels and conference centers in place of expensive leased data lines or dedicated video circuits. Then Eurovision, operated by the European Broadcasting Union (EBU), has implemented SRT as part of its video contribution service, offering European broadcasters low-latency live video transport using internet connections.
The aim of any such protocol is to reduce streaming latency down to the level of broadcast, or even better where that is possible. Irrespective of whether a signal is broadcast or streamed, there is a fundamental latency associated with the speed of light, which itself imposes maximum distance limits for applications such as some forms of gaming that are highly delay sensitive.
For video, the objective is to reduce the additional sources of latency on top of those imposed by the laws of physics. The primary sources of streaming latency are packaging delay associated with converting video files or live input into chunks for distribution at varying bit rates, along with buffering in the player device to cater for variations in rate of IP packet transmission over unmanaged networks such as the Internet.
There are also delays associated with caching at various points in the distribution chain and switching within the network infrastructure. Then there are delays associated with retransmission of IP packets that either arrive with errors or have been dropped altogether. Finally, the encoding/decoding and encryption/decryption cycles impose delay where they apply.
The SRT Alliance has focused particularly on packet retransmission, which is an essential component of video streaming over unmanaged networks because quality playback can only be assured if packet loss is kept at low levels. There are two principal methods for coping with packet losses and errors, Forward Error Correction (FEC) and Automatic Repeat reQuest (ARQ).
Of these FEC has been most widely deployed because it is mature and well understood, involving addition of extra bits to packets to add some redundancy so that they can be recovered at the receiving end even if some of their data bytes have been lost or corrupted in transmission. One problem with FEC for streaming is that it makes packets bigger and therefore adds to delay as well as cost even when the network is behaving perfectly. The other related problem is that FEC provides a fixed level of redundancy irrespective of how badly a network is behaving, so that it breaks down when error rates exceed a certain threshold, unless there is some dynamic mechanism for stepping up redundancy levels. Even then there could be a negative feedback as packets get bigger and bigger while the network deteriorates.
Such factors led the SRT Alliance to select the alternative method of ARQ, which attempts to minimize the delay associated with packet retransmission. This has been a feature of the Internet from the outset since packet retransmission is integral to the TCP (Transport Control Protocol) widely employed in many applications including video distribution. But as the SRT Alliance noted, TCP requires that all bytes of a stream be delivered and in their original order. Even if just one or two bytes are lost TCP retransmits the whole packet. This is highly inefficient for video where the impact of individual byte loss within a packet is cumulative rather than corrupting the whole packet as when say transmitting a text or numerical values. It is best to skip over small numbers of errors, while retransmitting when these exceed a given threshold. Otherwise the effect of packet retransmission is to accumulate delays which lead to continual rebuffering, completely defeating the point of aiming to improve quality.
ARQ has added a clever component by combining acknowledgements from the receiving device with timeouts to ensure that packets are retransmitted only when necessary and then in a timely fashion without having to wait for retransmission requests. If the sender has not received an acknowledgement of a given packet by the timeout time it retransmits it anyway. This greatly reduces the impact of packet retransmission on buffering delay as well.
The SRT Alliance has also taken some other steps at an architectural level, beyond the protocol itself, by allowing connections to be made directly between signal sources and their destinations. This contrasts with many existing video transmission approaches that require a centralized server to collect signals from remote locations and redirect them to one or more destinations. Such centralized approaches have a single point of failure that can also become a bottleneck during periods of high traffic, the SRT Alliance contends. Avoiding a central hub also reduces end-to-end signal transit times and potentially halves bandwidth costs by requiring just one link to be provisioned instead of two. By using direct source-to-destination connections, the SRT protocol can reduce delay a bit further.
You might also like...
Optical disks rely totally on the ability of the pickup to follow and focus on the data track. It is taken for granted that these mechanisms are phenomenally accurate, work at high speed despite being made at low cost and…
The optical disk has some useful characteristics that have allowed it to survive alongside magnetic media. John Watkinson takes a look.
Immersive audio transforms the listening environment to deliver a mesmerizing and captivating experience for a wide range of audiences and expansive group of genres.
Without doubt, virtualization is a key technological evolution focus and it will empower many broadcast and media organizations to work differently, more efficiently and more profitably.
The hard disk drive rapidly converged on the concept of one head per surface with all of the heads moving together on a common positioner.