Video Networks

Camden Ford

Have you noticed how those small digital islands are growing and looking more like a continent? The first wave of digital video was marked by the replacement of analog devices. Many of these newer digital devices offered new and useful features such as pre-read and self-diagnostics. As important as this wave was to the television industry, these devices did not take advantage of digital networking.

The professional video industry is well into a second wave of digital video, one in which islands are interconnected to form systems. This is a classic case of the whole being greater than the sum of its parts. With the convergence of video and computer technology now reaching a critical mass, networking can be the next big step to facility efficiency. Choosing the right technology for a given application is the key to performance and – equally important – expandability.

Converging technologies One of the first instances of convergence in broadcasting was in the form of nonlinear editing systems. Most of these systems were standalone or, at best, part of a simple network. The concept of workgroups was advanced based on the idea that all of the elements of content creation needed to be linked to enhance productivity. What developed were networks linking graphic artists and audio workstations with the nonlinear editing workstation. Artists and editors could exchange materials electronically, reducing errors and eliminating the costs of and need to care for floppy disks, tapes and other physical transfer media.

Success in this area naturally led to the implementation of shared storage systems where multiple editing workstations could have access to needed material. At the same time, computerization had already penetrated management functions, such as on-air programming databases, schedules, automation systems and other station functions. One factor limiting routine network implementation has been the size of video files. Early networking technology was slow and did not take into account the real-time needs of video. Because of this, early networks were relegated to data transactions and control functions.

Today, however, network technology has reached a reasonable performance level with respect to moving video files. It is time to consider the question of effectively networking an entire facility. The typical goal would be to move content throughout a broadcast facility electronically, without using videotape. To consider effective network strategies for broadcast facilities, a primer in network technology choices is in order.

Basic choices As broadcasters integrate more digital systems into their facilities, reliance on different types of video compression increases. In many cases, compression technology makes digital systems cost-effective. Different compression techniques (such as MPEG, DV, M-JPEG, etc.) are used in different applications based on their characteristics. As compression technologies improve, and network performance increases, many of these devices will integrate networking technologies to extend the size of these digital islands.

Some common networking technologies and topologies that are gaining acceptance in today’s broadcast facility include Fibre Channel networks, Storage Area Networks (SANs), Ethernet, and ATM. SDI/SDTI technology is used as a replacement for analog video routing systems. Although not a true networking technology, SDI/SDTI is used to connect standard videodevices such as VTRs, switchers, mixers, video servers, video monitors, etc. through a central router. Advantages of SDI/SDTI are that it can carry uncompressed ITU-R 601 video as well as compressed formats such as DV, MPEG, HDCAM and D5-HD, and of course, most broadcast manufacturers support SDI/SDTI as a standard interface on their products.

Fibre Channel (FC) networks are being installed to link multiple video servers together for transferring files at high speeds. FC itself runs at gigabit speeds, but due to packetization and protocol overhead, the actual data throughput is closer to 600Mb/s. While FC networks are growing in size due to the recent advent of intelligent FC switches, the protocols used in many cases are still proprietary. Typically, video servers that utilize FC networking for file transfers cannot transfer files in and out of the network or interoperate with non-FC devices. FC does not currently support isochronous streaming and can only be used as an asynchronous data transport. Due to this, FC networks are often utilized within a single manufacturer’s product family. To get video or data into or out of these networks, typically requires the video or data to pass through one of these devices via some other interface (usually SDI for video or Ethernet for data).

Storage area networks are a fast growing technology, especially for the data industry. Storage area networks (SANs) also use FC networking, but instead of connecting one device to another, they connect devices to a storage system. Devices connected to a SAN can share video and data files by using disc storage as a central data repository. The acceptance of this networking topology in the nonlinear production and post-production markets is growing. A SAN allows multiple nonlinear workstations to access a common pool of shared storage. Unlike the FC networks, SANs use standard FC protocols for accessing FC disc storage (typically SCSI on FC). Storage can be shared by connecting multiple workstations directly to the FC storage through a FC hub or switch. The difficulty with these networks is that to control access to the shared storage, complex, proprietary file systems and file management software may be required. In many implementations, software drivers must run on every workstation tied to the network. In addition, an external workstation may be used to synchronize access to the material stored. To attach external devices to the SAN, these devices must be capable of running this proprietary software. Typically this limits network access to workstations only. As in the FC networks previously described, moving video or data into or out of this network means the video and data must pass through one of these workstations.

Ethernet networks are mature and relatively inexpensive. They have been used extensively in broadcast facilities for years as data and control networks. Most broadcast facilities have internal IS networks for general application and file sharing. Most automated traffic systems have Ethernet connectivity for local access. Many also have connectivity over wide area networks (WANs) to a centralized traffic and program scheduling system. Many automation systems use Ethernet networks for connectivity to central media databases and to distributed control computers. However, many times, these similar networks remain islands. Typically, these networks are not used for transferring video files, and most are standard 10Base-T or 100Base-T.

The bigger brother to Fast Ethernet (100Base-T) is Gigabit Ethernet. This relatively new technology is limited to fiber optic connections. Fiber optic connections are not yet standard on PCs; however, new NICs are available that offer these connections as standard I/O. Because Gigabit Ethernet uses the same protocols as other implementations of Ethernet, it is therefore automatically compatible with existing application software. The increased bandwidth makes Gigabit Ethernet feasible for high-speed transfer of video files between devices. Broadcasters are already starting to see implementations ofGigabit Ethernet used instead of Fibre Channel for interconnecting video servers. The data rates for file transfers are relatively similar and Gigabit Ethernet is a widely accepted standard. Not only that, but it is easy to connect to other devices such as nonlinear editing workstations, automation systems, traffic systems, etc. One disadvantage of Gigabit Ethernet is that it does not support isochronous channels, which can be used for video streaming. Because of this, bandwidth or timing cannot be guaranteed.

ATM was originally designed to handle both data and voice traffic. To do this, there is a mechanism for specifying class of service. Users can specify data priorities such that bandwidth can be guaranteed for voice (or video) applications. These come at the expense of lower class of service data such as TCP/IP data. Because this technology was designed to handle many types of data and up to five different classes of service, it can be fairly expensive to implement. It is a highly flexible and highly scalable technology that is most often used for long-haul data backbone networks such as the Internet backbone and telephone networks. Many broadcasters have used ATM on wide area networks for both data and video applications, but it is not often used for internal network operations due to its complexity and cost.

Planning for the future The dilemma facing broadcast facility owners and operators today is how to install a full-featured network without constraining opportunities for later growth, and – naturally – to stay within reasonable budget limitations. There is an overriding need for flexibility, both in the size of the network and the ability to handle video data from an ever-changing mix of input and output formats. This need for format independence also encompasses broadcasters’ requirements to get full use and value from the equipment currently in their facilities while addressing new equipment reaching the market each year.

A number of solutions have been proposed and implemented. Some are more effective than others are. As always, such decisions come down to a series of trade-offs; typically cost vs. features vs. flexibility.

Computer networks are designed to handle data; but not all video content reaches a facility as data. However, not all networking technologies have these limitations. IEEE 1394 offers the ability to route data in both an asynchronous and isochronous manner. Many people think of IEEE 1394 as a simple consumer technology used for connecting digital camcorders to PCs. Digging deeper into the IEEE 1394 specification, one finds that it is a good networking technology for use in video-intensive environments.

The current commercial implementation is IEEE 1394a. This specification is complete. Products that meet this specification have been shipping for over a year. IEEE 1394a specifies link distances of 4.5m (15 feet) with standard cables and up to 10m (35 feet) with high-quality cables. Bus speeds up to 400Mb/s are allowed. One may ask how an effective network can be built with cable distances of only 4.5m. The answer is that 4.5m link distance does not make a very good network. However, IEEE 1394 proponents are forging ahead with a new specification known as IEEE 1394b. The new specification allows IEEE 1394 devices to incorporate glass optical fiber (GOF), plastic optical fiber (POF), and unshielded twisted pair (UTP Category 5) for distances up to 500m, depending on media type. This new specification increases bus bandwidth to 800Mb/s and beyond. The IEEE 1394b specification actually specifies link bandwidths of 800Mb/s, 1600Mb/s and 3200Mb/s. This new specification will be complete shortly. Prototype devices running at 800Mb/s are already operating at industry trade shows.

With Gigabit Ethernet and Fibre Channel already on the market and running at gigabit speeds, what makes IEEE 1394 different from the other networking technologies? IEEE 1394 provides fixed, low latency end-to-end delivery for high bandwidth streams. Designed specifically as a multimedia technology for carrying audio, video, control, and data, 1394 provides two Classes of Service, isochronous (guaranteed bandwidth) service and asynchronous service. Its most attractive feature is that it provides the ability to allocate discrete bandwidth channels for delivery of isochronous datastreams such as video and audio. This means that a device can reserve 25Mb/s of bandwidth for streaming DV25 video. Regardless of other traffic on the network, the reserved data channel is never interrupted or reduced.

Another interesting capability of this technology is that it can support up to 64 individual isochronous channels on a single link. This means that up to 64 different MPEG-2 videostreams can be simultaneously carried on a single link. In addition to the isochronous data channels, any bandwidth not reserved can be used for asynchronous traffic. There are already several commercial implementations of TCP/IP on IEEE 1394. To guarantee at least some asynchronous packet transport on the link, up to 80 percent of the available bandwidth can be reserved for isochronous data channels. The remaining bandwidth carries asynchronous packets. On a 400Mb/s link, this equates to 320Mb/s for isochronous traffic (such as video and audio), and 80Mb/s for asynchronous packets (such as TCP/IP). Using IEEE 1394b at 800Mb/s, this means that up to 640Mb/s can be used for video, leaving 160Mb/s for asynchronous packets. These numbers are a maximum for isochronous bandwidth. If less bandwidth is required for isochronous channels, any remaining bandwidth can be used for asynchronous packets (up to 50 percent of the link bandwidth). For example, using 800Mb/s IEEE 1394b with only 40 MPEG-2 streams (10Mb/s per stream), 400Mb/s of bandwidth is available for asynchronous traffic (IP traffic). This is four times the raw bandwidth of Fast Ethernet.

Connectivity For the most part, current networking technologies used in video applications do not provide an easy connection to video and audio devices. Video servers, for example, may have physical connectors for SDI video, AES audio, LTC timecode, RS-422 control, Ethernet and Fibre Channel. The SDI, AES, and LTC connections are all different data formats carried at different bit rates; but they are all serial and isochronous. The RS-422, Ethernet and Fibre Channel connections are asynchronous, also with different formats. Getting data into and out of this server requires a multitude of connections. IEEE 1394 can simplify this because it can carry all of the isochronous information and asynchronous information on the same cable. A simple example can be seen in the consumer market. For a nonlinear editing workstation to record video for editing, the workstation needs an SDI (or composite or S-video) input, and AES (or analog) audio input, a timecode reader (optional), and RS-422 (or RS-232) for control of the VTR. With these connections and the appropriate software, the nonlinear workstation can then control a professional VTR. On the consumer side, a PC with an IEEE 1394 card and a 1394-capable digital camcorder can accomplish all of the above with a single IEEE 1394 connection.

Many devices are adding IEEE 1394 ports as standard connections. Apple and Sony computers have them, as do some Compaq and NEC computers. Many lower-end digital camcorders have them, and they are even appearing on some higher-end equipment. IEEE 1394 has been specified as the preferred method of connecting set-top boxes to digital television sets, and some new HD VCRs now include 1394 connectors. Additionally, there are a multitude of scanners, digital still cameras, and even disk drives that are implementing IEEE 1394. While most devices are generally consumer or industrial grade, the high volume of these devices has driven the cost of IEEE 1394 down to very low levels, reducing the implementation costs. The advent of high-performance IEEE 1394b technology, with its capacity for long distance links on fiber and copper cables, combined with IEEE 1394 switching and mass storage technology, makes 1394’s future as a video networking technology.

There is a next generation networking and storage solution for professional broadcaster based on IEEE 1394. The recently introduced Video Area Network has the capability of recording and playing back video, audio, and data in a variety of formats, both compressed and non-compressed. This network infrastructure was built to encompass the many devices on the market today with IEEE 1394 interfaces, as well as those without. The system includes a set of network interfaces that will connect the IEEE 1394 network to SDI, AES, LTC, RS-422, Ethernet, ATM, DVB-ASI and other standard interfaces. Implementation of effective video network topologies can begin to connect those many and varied digital islands found in broadcast facilities today. The net result will be technical centers poised to take full advantage of the future and to serve their customers.

Battery development adn applications

Technology abused