IP networks have become the popular, low-cost choice for transporting data, voice and video content. IP communications were originally designed to ensure data integrity, where the criterion for success is complete delivery. If data packets arrive late or are lost, protocols such as TCP exist to resend them without a major impact on the user experience.
When it comes to live video, however, completeness is not enough. Packets must be delivered reliably and on time in order to ensure uninterrupted, broadcast quality video. Unfortunately, the inherent nature of IP networks, characterized by packet jitter, packet loss, and link capacity fluctuations – makes them unreliable for professional live video broadcast.
VideoFlow has leveraged its extensive knowledge in the areas of video broadcasting and IP networking to create a breakthrough, layered technology for ensuring unprecedented service continuity for live video transport. Each of these layers protects the video stream from specific types of interferences.
Packet jitter elimination
Packet jitter is caused by changes to the network delay. These changes could be the result of changes in the connection path or network utilization. VideoFlow eliminates packet jitter, making sure packets will not be discarded by the decoder due to late arrival. We use a fixed delay buffer at the destination in order to play the stream at a constant delay.
Lost packets recovery
Standard technologies for correcting packet loss, such as Pro-MPEG FEC, result in high bandwidth overheads (up to 30%). In contrast, VideoFlow uses a patent-pending technology based on the ARQ protocol, by which the receiving DVP notifies the transmitting DVP on packet loss events. To avoid retransmission overload and to keep the bit rate overhead to a minimum, the receiving DVP splits the buffer into several detection windows, each scanning for packet arrival.
Jitter grace time
Jitter grace time allows us to distinguish between late and lost packets. The jitter grace time is preset on the destination DVP (Sentinel) to determine how long a packet is allowed to be “late”. If the jitter grace time expires and the packet was not detected, then a missing packet notification is sent to the Protector at the source. This reduces the retransmission overhead and frees more bandwidth to the video stream.
Prioritized Protection Flow (PPF)
Since some packets will always be lost, two distinct flows of packets need to be handled by the transmitter: the transport stream packet flow and the ARQ-driven packet recovery flow. Another issue caused by ARQ is the burst of retransmitted packets, which may cause a momentary bit rate surge. PPF resolves both issues by giving the transport stream packets high priority, limiting the bit rate dedicated to packet retransmission, and prioritizing the retransmitted packets based on presentation time. This results in the lowest delay and lowest overhead in the market.
Null packet deletion
Null packets are inserted by the encoder to ensure a constant transport stream bit rate. Since null packets do not contain meaningful data, the receiver is expected to ignore their contents. VideoFlow deletes the null packets at the source (Protector), thus decreasing the transport stream bit rate and allowing more bandwidth for actual video content. This also helps to lower the probability for lost packets over unstable connections, or to use lower bit rate links. VideoFlow reconstructs the null packets in the receiver (Sentinel) by re-inserting the null packets into their original locations, as required to maintain the original bit rate and accurate PCR.
Controlled Adaptive Rate (CAR)
CAR technology protects against fluctuations in the connection bit rate that can freeze the service. VideoFlow’s patent-pending CAR algorithm overcomes the difficulty in increasing the bit rate on the fly and so that the maximum possible bit rate is reached without exceeding the network capacity. Essential for contribution applications that require service continuity, CAR probes the video stream health in real time and adapts the encoder’s bit rate to the IP network conditions on the fly, ensuring uninterrupted broadcast quality video.
Stream Duplication / Load Share / Multi-ISP
Private “five 9s” networks are not available everywhere, and may be cost prohibitive for many companies. VideoFlow’s technology boosts the reliability of unmanaged IP networks, like the Internet, by trading bit rate for higher availability. Independent transport paths can be created between the transmitting and receiving DVPs using multiple ISPs, while the same stream can be sent via two or more Internet connections at the same time. Load sharing keeps the bit rate constant by splitting the stream between different paths at the source.
Input stream redundancy
To protect against source failure, VideoFlow implements two types of redundancy schemes: path redundancy and facility redundancy. In path redundancy, the same source of content (i.e., encoder) is sent by the Protector via two or more independent paths (see above). Facility redundancy refers to sending the same content from two different sources (i.e., two encoders), each via an independent path to the Sentinel, which locks onto the primary stream and fails over to the secondary stream in the case of a stream sync loss.
Box redundancy – High availability
Box redundancy offers the ultimate protection against equipment failure. Our High Availability (HA) solution lets you configure a cluster of two DVP appliances to work in active/standby mode. The active and standby DVPs are directly connected to enable the standby DVP to constantly monitor the active DVP. Should the standby DVP sense that the active DVP is not responding (for a pre-defined period of time), the standby DVP will automatically become active.