Jitter And Network Performance In Broadcast Networks

In this article, dB Broadcast discusses the causes of IP jitter and its impact on broadcast network performance. dB Broadcast has extensive experience in designing and implementing innovative IP solutions for clients such as Bloomberg Television and BBC Wales.

In an ideal IP network carrying constant bitrate data, jitter is any deviation from the regularity or "linearity" of the arrival times of IP packets at a receiver. This can be caused by improper router or switch configuration or queueing issues, but if the routers and switches are operating correctly, the most common cause of jitter is network congestion at router and switch interfaces.

As IP networks are asynchronous some jitter is inherent, but to optimise network performance and minimise jitter, senders should ideally transmit IP packets in as linear a fashion as possible. Applications within network elements are likely to require the data to be non-bursty, and to achieve this, receivers implement a "de-jitter" buffer. Packets flow out of the buffer at a regular rate, removing the jitter from the packets flowing into the buffer from the interface. The application then receives de-jittered packets from the output of this buffer rather than directly from the interface.

What is the impact of excessive packet jitter?
In the previous section, we saw that packets flow out of a receiver's de-jitter buffer at a steady rate known as the "drain rate", with the rate at which a buffer receives data being known as the "fill rate".

Selecting the size of this buffer is important for two reasons:
• If the drain rate exceeds the fill rate, then it is possible that too small a buffer could underflow, resulting in stalled packet flow. Likewise, if the fill rate exceeds the drain rate, then at some point the buffer will overflow, resulting in packet loss.
• However, if the buffer size is too large, then the network element will introduce excessive latency.

Network jitter causes the packet arrival intervals at a receiver to become non-linear, and the buffer fill rate will no longer be constant. As the jitter becomes greater, the variability of the buffer fill rate becomes greater. At some point this will lead to the condition where the buffer's fill and drain rates become so uneven that the buffer will either underflow leading to stalling, or overflow leading to packet loss. Both stalling and packet loss will lead to impaired video.


What else causes packet loss?
If the network is experiencing packet loss, and if jitter levels are excessive, then it is likely that network congestion is significant enough to be causing the packet loss. However, if jitter levels are acceptable and the system is experiencing packet loss, network congestion is unlikely to be the cause, which is more likely to be port over-subscription.


What is the impact of non-linear data flows on network performance?
With SMPTE ST 2022-6 the data flows are linear, as the data rate is that of the entire SDI payload, which is always of constant bit-rate. With SMPTE ST 2110 only the active video is transmitted and the HANC and VANC data is not transmitted. There are two possible modes of transmission defined by SMPTE ST2110-21 (which is commonly referred to as "traffic shaping"). The first is the "linear" mode or PRS (Packet Read Schedule) as it is referred to. When a linear PRS is used, packets containing only the active video are transmitted by the sender at even intervals for the duration of the whole frame, resulting in a constant bit-rate transmission of just over 1 Gbps.

The second mode is known as the "gapped" PRS, where the sender transmits the active video packets in a linear fashion, but transmits nothing during the period corresponding to the VANC interval. This of course results in an inherently bursty (non-linear) variable bit-rate transmission of between 0 Gbps and more than 1.5 Gbps. The average bit-rate is the same as for the linear PRS at just over 1 Gbps. This burstiness is, as we discussed earlier, undesirable from a network perspective, as the impact on the network is the same as if the linear data flow was subject to bursty jitter caused by network congestion.

The burstiness must be handled by buffering the flow, to ensure that the packet flow is linear enough to avoid buffer underflows and overflows. However, increasing the buffer size when compared with that required for the linear PRS, leads to a corresponding increase in latency.

When implementing broadcast networks where a SMPTE ST 2110 gapped PRS is used, system integrators need to pay special attention to network congestion and jitter in the video data flows. They must ensure that any jitter does not result in excessive non-linearity, causing the router and switch buffers to become overwhelmed, resulting in packet loss.

dbbroadcast.co.uk

This article is also available in the August edition of Broadcast Film & Video. Available here.