Application of FPGA in Broadcast Video Processing

1. Timing

The demand for broadcast video systems is increasing significantly year by year around the world due to a combination of factors:

The increase in broadcast channels available to viewers. Worldwide, more viewers have gone from a few channels to hundreds of channels. The development of cable and satellite TV transmission networks. The development of Internet broadcasting has brought us into the era of "millions of channels". The development of small-scale broadcast video applications within companies and on a small scale for information exchange, training and security. There has also been great progress in the entertainment industry, such as restaurant video on demand and ultra-large screen electronic display screens. Delayed video playback and quasi-video on demand have greatly increased the convenience of users, making users need more channels. Today, large-screen TVs are developing from ordinary TVs to high-definition TVs, resulting in the need to install more broadcast equipment. The content contained in videos is increasing rapidly. Additional special effects, multiple windows, subtitle overlays, replays and other functions require higher-performance broadcast equipment. The abundance of video resources. Compared with the previous production in special studios and reporting of major events, it can now be produced in every community and business. The number of video resources has grown rapidly, and a new low-end broadcast equipment market has emerged. The rapid transition to digital television has led to the replacement of old analog signal systems. Due to the digital television revolution, there is a strong demand for video encoding conversion between various digital signal specifications, which has led to the need to add broadcast equipment, including: Basic specifications: non-compression, time compression, MPEG2, MPEG4, H.264 compression, Windows Media9, Real Networks, etc.; Image size: standard definition (SD), high-definition 720P, high-definition 1080P; Transmission/packaging specifications: ATSC/DVB, SDI, IP, etc.; Audio specifications: AES3, etc.; International specifications: NTSC, PAL; Equipment interface: 1394 FireWire interface, DVI interface, GigE Ethernet interface;

The latest changes above have made powerful broadcast video equipment necessary. Broadcast equipment requires stronger processing capabilities, more output channels, more output specifications, and a wider range of applications.

This paper describes a completely new product platform that will meet future needs. The product and development costs of this platform will be lower than existing systems.

2. Video system platform

First, let's survey some of the video broadcast systems and video platforms in use today. In general, the basic processes used for video production and distribution are as follows:

Content production: camera production and computer production; editing/image overlay/image mixing; special effects; video quality enhancement; video signal format conversion; video storage.

Video extraction and publishing: video extraction specification conversion; insertion of program information and advertisements; broadcast video specification conversion; distribution: normal, delayed playback, video on demand.

Today, the platforms used to complete the above video functions can be roughly divided into the following types:

Content production platform

Because the content production platform is a non-real-time system, there is no high performance requirement, even if the processing time is longer than the actual time. Therefore, ordinary computers and workstations are sufficient. Usually, these platforms are used with a video accelerator board, but this accelerator board is only used to improve the work efficiency of video producers, not for real-time processing needs.

Video ingestion and publishing platform

The video ingestion and publishing platform must be synchronized with the video broadcast, and all processing must be completed within the line rate and frame rate of the video broadcast, so there are absolute high performance requirements. Reliability is also an inevitable requirement for these processes.

To meet these requirements, most of the video ingestion and distribution platforms currently used are built on the manufacturer's professional broadcast platform, which consists of several modules with special functions built into professional equipment.

Some manufacturers have begun to use low-cost personal computers for real-time video processing, but this is limited to processing some limited peripheral functions and has not become the mainstream of video extraction and distribution systems.

It is already clear that PC-based platforms will play an increasingly important role in video ingestion and distribution systems. These platforms are poised for a number of reasons:

Outstanding performance-price ratio brought by large-capacity processing; Standard platform that can be used for different application software manufacturers, better functions, lower costs; Higher performance-price ratio and more functions;

3. Advantages and limitations of personal computers

The capabilities of personal computers have been continuously improved, and now they can process a single video stream in real time, compress a single video stream, store it on a hard disk, and decompress and play a standard quality stream. However, the bandwidth of a personal computer is far from being able to handle video processing of many uncompressed video streams, processing of terminal video distribution, live broadcast processing, and other high-quality video processing.

At present, the difficulty of uncompressed video processing lies in solving the contradictions of numerous professional processing boards from various manufacturers. These professional processing boards can complete their specific functions, but cannot flexibly add new functions. When you need multiple functions, you need to coordinate completely different processing boards from different manufacturers.

What the market needs is a flexible, uncompressed video processing platform on a PC that can provide the processing power of professional equipment like high-end systems but at the low price of low-end PC systems.

4. FPGA in Video Processing

FPGAs (field programmable gate arrays) produced by Altera or other companies are a very fast-growing semiconductor chip. In the past 10 years, these chips have been widely used and their performance has been continuously improved. Now there is a $2 billion market and they are widely used in fields such as communications, military and broadcast equipment.

FPGA chips have the ability to provide programmable software and the processing performance of professional hardware. In video applications, FPGA chips can provide up to 10 to 100 times the processing power on a personal computer platform, and when users need to change the function of the system, they only need to re-download a file to modify the user's hardware configuration, and the user's hardware upgrade can be completed in 1 second.

Although FPGA technology has long been used in the broadcast video industry, to date, such applications have mostly been limited to specific functions on professional platforms, with only a few applications for specific functions on PCI boards in personal computers.

Despite this, the processing power of FPGA has been developing rapidly in recent years. Now it has evolved to the point where one chip can perform the functions of a board, or one chip can perform the functions of a system (SoC). Because multiple FPGA chips can be installed on a PCI board, it is possible to develop a complete system equivalent to a professional device on one board.

In addition, under the existing technology, standard video processing function modules can be downloaded to a new high-processing capability FPGA chip. These modules are like multiple function modules in a professional device. In other words, now a chip can contain the entire system, instead of having to be implemented by a professional device as before.

Compared with developing professional equipment systems, this development method has the following advantages: much lower development costs; much lower system costs; extremely convenient user design and upgrades; and the ability to directly use application software on the personal computer platform.

5. Develop a "PC + FPGA" platform

When designing a video system using the "PC + FPGA" algorithm, the most important decision is which processing runs in the FPGA and which runs on the PC. Generally speaking, the best system design is for all high-speed data flows to pass through the FPGA and not through the PC. The PC can be used for data flow management and some lower data output functions, such as inserting overlay data.

The following types of processing are suitable for optimal performance through FPGA: video mixing/overlay; synchronous compression/decompression; data format conversion/transcoding; IP, DVB/ATSC specification packet/stream processing for multi-channel simultaneous output systems; video extraction, decompression, multi-channel broadcasting; video enhancement: noise reduction, color stabilization, resolution enhancement, etc.; video, audio and other data stream processing.

A single FPGA-based board can handle up to 500MB of aggregate bandwidth (equivalent to one HD uncompressed stream or four SD uncompressed streams).

If the user needs, a personal computer can carry multiple processing boards, and a tower rack-mount server can contain multiple personal computers.

This design approach is very applicable to a wide range of applications, from small personal computers that can handle a few SDI data streams to tower servers that can handle hundreds of data streams.

6. Application of “PC+FPGA” platform

Using the algorithms of the above "personal computer + FPGA" platform, low-cost, high-efficiency solutions can be provided for the following types of applications: video on demand servers for local networks, entertainment, the World Wide Web, and cable TV; scheduled replay servers; delay; live video production of sports programs and other programs; ATSC/DVB data broadcasting; community service systems, local news/weather forecasts, and cable online shopping; cooperative communication channels on the Internet; multi-channel satellite receiving equipment; program production and advertising insertion equipment; and regional video distribution systems.

7. Introduction: Tsunami on-board video platform

Avvida's Tsunami PCI board and Wave development kit enable users to quickly develop and mass produce their products.

The Tsunami broadcast processing board includes: up to 6 SDI interfaces; up to 5 FPGA chips, each with 4 high-speed memories, capable of implementing a variety of video processing functions; high-bandwidth memory access to the personal computer via the PCI bus; and a digital video input interface with a transfer rate of up to 300Mbps from the personal computer to the Tsunami broadcast board.

Avvida can provide the following video processing function modules: format conversion between SDI RAW video format and 4:4:2, YUV, YcrCb, and other synchronous compression formats; sampling from 4:4:4 to 4:4:2 format; color space conversion: RGB format to YUV format; video and audio overlay and mixing; SDI input/output processing and error detection (EDH) insertion and extraction; synchronous phase lock; MPEG2, MPEG4 and other compression/decompression specifications; SDI input signal storage; stored content replay; and others.

8. Summary

The demand for more powerful digital video broadcast systems is growing rapidly, which is both a new challenge and a new opportunity for a system with broad application prospects. A standard personal computer equipped with Avvida's powerful Tsunami FPGA PCI processing board can effectively meet the market demand. The FPGA-based video processing function module attached to the processing board can bring users lower development costs and faster time to market.