Turning a generic IP network into a video network

IP networks offer many technical and financial benefits to broadcasters who need to transport SD and HD video content in real-time to remote studios, affiliates and distribution partners
Publish date:
Social count:

Broadcasters may have a lot of experience with generic IP WAN networks for moving data. However, if the goal is to use these same networks to deliver TV signals, especially live HDTV programs, then these IP networks must first be enhanced to support high-quality video. The key is controlling the variables.

Reliability and quality of service (QoS) have to be tracked, measured and managed differently. Throwing more bandwidth at the problem is not the solution. Such techniques as forward error correction (FEC), clock synchronization and buffer control are essential tools in maintaining a persistent and predictable quality level.

Quality and QoS

Although it's common to see video files exchanged over IP networks in production and broadcast LAN environments, real-time transmission of broadcast-quality SD and HD video over WANs is just starting to take off. The combination of DTV and IP technologies now offers broadcasters a fresh opportunity to significantly improve their core business by enhancing their television transmission services while simultaneously lowering the program transmission costs.

While IP networks provide distinct cost advantages in transporting video content from source to destination, ensuring video quality and reliability over long distances requires more than just a network with a specified QoS. It requires an IP video gateway capable of preserving the important characteristics of the broadcast video signal, such as clock accuracy and data integrity. While QoS can help minimize impairments in IP networks, a well-designed IP video gateway must also handle the intrinsic impairments of packet networks.

Video over IP

To successfully turn a generic IP network into a video network, one needs to first understand the unique challenges posed by IP networking. Fundamentally, IP networks are designed for data communications and operate in a best effort mode, i.e. they do not guarantee the delivery or the correct sequence of every packet.

Traditional packet networks do not have a priority mechanism to control the amount of traffic injected into the network. Traffic is only throttled after congestion is detected in the network, by which time a large number of packets may have been dropped at the routers.

Figure 1. Packet loss and variable delay caused by router queuing. Click here to see an enlarged diagram.

In best-effort IP networks, all user packets are treated equally, meaning time-critical video traffic receives the same priority as delay-tolerant e-mail traffic. In such an environment, video packets may be queued at a router behind a long line of delay-tolerant packets. The result can be packet loss or long delays for the video packets before they arrive at the receiver. (See Figure 1.)

A common but flawed approach to solving these issues is to allocate more bandwidth for the video traffic. However, over-provisioning alone cannot guarantee uninterrupted video transmission. In over-provisioned networks, there may still be an errant sender that suddenly injects a large burst of traffic into the networks, exceeding the network capacity and causing catastrophic congestion.

To guarantee consistent, perfect video transmission, an IP-based video network has to be architected with the essential building blocks: QoS and a broadcast-quality IP video gateway.

Quality of service

QoS is essential to transporting SD or HD video over an IP network. It consists of various network techniques and protocols that enable IP networks to guarantee each user a unique set of measurable network performance parameters, such as minimum bandwidth, packet loss ratio and maximum jitter. Such guarantees are especially required by real-time multimedia applications that have minimal tolerance for such network impairments as packet loss and delay variation.

QoS is typically handled by routers and switches in the network. Several important components of these QoS techniques are described below: multi-protocol label switching (MPLS) and traffic classification, metering, shaping, and admission control policies.

MPLS enables the creation of virtual connection-oriented paths within a network. Bandwidth can be reserved for each virtual path. Network performance parameters, such as loss and jitter, can be guaranteed throughout the traffic flow. MPLS can, therefore, be used to provide bandwidth guarantees for video traffic by separating these flows from other best-effort paths assigned to Web or e-mail traffic.

Traffic metering, shaping and admission control policies are other forms of QoS techniques that limit the amount of traffic entering the network. By controlling the ingress traffic, network congestion can be avoided and such conditions as packet loss and jitter become minimized.

Other QoS techniques, such as differentiated services, traffic classifications and priority queuing mechanisms, also allow critical broadcast video traffic to be treated with higher priority than standard voice or data traffic.

While these QoS techniques described above greatly improve the performance of IP networks, they do not completely eliminate certain fundamental characteristics of packet networks, such as jitter due to queuing in the routers, occasional packet loss or out-of-sequence delivery caused by network outages.

Despite the many benefits of QoS, impairments like these may still cause video distortion, such as image macroblocking, frame dropouts and temporary interruption of services, beyond the tolerance level of most broadcasters and content owners. The most effective way to mask these residual network impairments and ensure a perfect video transmission is to use an IP video gateway.

IP video gateways

The second critical building block of a video network is an IP video gateway specifically designed to transport broadcast video signals. A video gateway, also called a video network adapter, performs conversion from one type of interface to another type.

Figure 2. IP video gateways in contribution networks. Click here to see an enlarged diagram.

Most equipment used in studios or broadcast stations use non-IP, bit-serial interfaces, such as ASIs or SDIs. An IP video gateway is required to convert those interfaces to IP interfaces, such as fast Ethernet or gigabit Ethernet. (See Figure 2.) The functions of an IP video gateway include:

  • Conversion from bit-serial interfaces to packet-based IP interfaces.
  • Clock recovery and synchronization.
  • Recovery from packet loss and out-of-order packets.
  • Network monitoring and reporting.

One of the most important functions of an IP video gateway is to accurately reproduce the original timing (clock) information of the video signal at the receiver. This task becomes extremely challenging in IP environments because the large packet jitter (the variance in interpacket arrival times) caused by the IP networks makes it extremely difficult to estimate and track minuscule clock differences between the sender and the receiver.

In addition, due to the high bit-rate nature of broadcast contribution video, even a slight inaccuracy in clock recovery may quickly cause buffer overflow or underflow, disrupting the video signal. If not accurately corrected, the clock skew at the receiver may cause a host of problems, ranging from loss of colors to complete loss of picture.

Figure 3. Forward error correction. Click here to see an enlarged diagram.

A video over IP gateway also has to cope with potential packet loss or reordering of packets, which may cause significant picture degradation. FEC techniques, such as the one defined by the SMPTE Pro-MPEG Code of Practice #3, represent a powerful tool to combat this problem by sending repaired packets together with the original stream of packets containing the video data. When packets are lost and never make it to the receiver, the repaired packets, combined with the non-missing packets, are used to reconstruct the lost packets. (See Figure 3.)

When evaluating IP video gateways, it's important to use gateways that implement a smart buffer scheme, combining the synchronization, jitter, reorder and FEC buffers so latency at the receiver is minimized. Minimizing latency is particularly critical in an environment where large delays can be problematic, such as in live interviews and interactive TV applications.

IP video gateways should provide real-time network statistics so that appropriate actions can be taken as soon as a problem is detected. By capturing statistics such as packet loss ratio, maximum burst loss and network jitter, users can be alerted to immediate or pending problems and correct them before the conditions worsen.

These statistics also enable users to optimize the operations of the IP network, video codec and video gateways. If such operations are tuned correctly, both private and public IP networks should be capable of carrying perfect, broadcast-quality video.

IP networks offer many technical and financial benefits to broadcasters who need to transport SD and HD video content in real-time to remote studios, affiliates and distribution partners. Ensuring content quality and stream reliability depends on deploying the critical architectural components that enable this. The first critical step is to provide networks with the right QoS components in order to minimize various impairments associated with packet networks. The next important step is to select the right IP video gateway solutions that will overcome any remaining impairments. FEC is designed specifically for this purpose. By implementing both of these measures, broadcasters can enable an IP network for real-time SD and HD video transport.

The benefits

Dan McCrary is vice president of marketing and Henry Sariowan, PhD, is vice president of strategic technology planning for Path 1 Network Technologies.