Adaptivity and Future RIST Capability (by VideoFlow)
By Adi Rozenberg, Co-Founder and CTO, VideoFlow
For a long time, we have been hearing from broadcast industry experts that streaming live video over an IP network is challenging. A live streaming solution over IP needs to cope with packet loss, and packet jitter to meet the requirements for either content contribution or distribution.
Packet loss and packet jitter are always the immediate suspects; however, there is a third and more menacing problem – bandwidth availability. Ensuring sufficient and stable bandwidth is often overlooked because of the use of high SLA connections and networks, including private links, leased lines, or managed networks like MPLS.
High SLA connections and networks can provide a minimum bit rate per connection to secure minimum bandwidth, but it comes with a high price tag.
Even high SLA connections and networks cannot guarantee the minimum bit rate for reasons including oversubscription, gear malfunction, and even unexpected users’ behavior. There are incidents where the bit rate is not available, and if this last for too long, there is no video.
Simple ARQ methods can not handle bitrate fluctuations well. Moreover, ARQ intensifies the problems caused by a drop in bitrate. Think of pushing a 10 Mb/s stream through an 8 Mb/s connection. In this scenario, 20% of the packets will drop triggering ARQ in the receiver to send retransmission requests to the sender for each lost packet. The sender will generate 20% more recovery packets in an attempt to recover the additionally lost packets. Sending the recovery packets will drop the effective bitrate to 6Mb/s because both the video packets and the retransmission packets use the same connection; therefore, the receiver will send more requests for retransmission. It is easy to see that the link will soon be saturated with retransmitted packets leaving no bit rate for the video packets.
Backed by more than 20 technology vendors and a fast-growing number of users, the RIST main profile is ready for prime time, and it is on the right track for becoming the next industry equivalent to ASI. Pouring hundreds of years of experience to the RIST technical recommendation (TR), RIST’s simple and main profiles include the first line of defense to deal with bandwidth instability problems. RIST advanced features offer the use of multiple links for link bonding or load share.
Using multiple links decreases the probability for the bitrate to drop below a minimum required level but cannot eliminate it. Also, there are cases where multiple connections are not available. At the end of the day, once the bit rate is not there then the only way to keep the service up and running is by adaptation. The problem of insufficient bitrate intensifies when it comes to content distribution where the main issue is how to disconnect the bit rate dependency of each receiver from the rest of the receivers.
The RIST working group identified the importance of adaptivity to the broadcast industry hence RIST members, including VideoFlow, proposed to add adaptivity capabilities to a new RIST profile – the enhanced profile. VideoFlow has developed and deployed several adaptation technologies since 2015. These techniques will be offered for consideration by the RIST working group.
The adaptivity is invoked when ARQ (used in the simple profile) is no longer sufficient for error-free delivery or when a network probe detects that the bandwidth is decreasing. The adaptation will take place in the source (or sender). The source should change the stream’s attributes and reduce the effective bitrate to a level meeting the reduction in the network bitrate, and increase the stream’s bitrate whenever network conditions allow. Three different use cases can benefit from adaptive techniques:
Content Contribution – A RIST sender controls an encoder to change encoding attributes to meet the network conditions.
SPTS distribution to multiple receivers – in this type of application, a single source is preconditioned to multiple encoding profiles. These profiles can then be delivered to one or more RIST receiver destinations with each destination treated differently, by way of source profile switching.
MPTS distribution to multiple receivers – Breaking the bit rate dependency between receivers is more complicated when it comes to the distribution of MPTS. The main challenge is to create a simple and inexpensive solution to interoperate with existing equipment. The answer is to assign a priority level to each service in the MPTS per destination. Lower priority services will be dropped from the MPTS in case of bitrate reduction, giving way to premium services. The services will be added according to their priority when the network conditions allow. With this technique, a RIST sender can match the number of services in the MPTS per RIST receiver independent of the other receivers.
Adding these adaptivity capabilities to the RIST Enhanced profile will make it stand out and on the cutting edge of video delivery over IP technologies.
Adi Rozenberg will discuss the benefits of each technique in his presentation on the September 16 RIST day during IBC 2019.