Local Commercial Insertion in the Digital Headend

Mukta L. Kar, Laboratories, Inc. Sam Narasimhan, Motorola Broadband Communication Richard S. Prodan, Terayon Communication Systems

Abstract require significant changes to encoding methodologies to enforce predetermined Existing ad insertion systems enable cable conditions at the splice points. headends and broadcast affiliates to insert locally generated commercials and short The Society of Cable Telecommunications programs in a channel seamlessly before Engineers (SCTE) Digital Video delivery to the home. The revenue generated Subcommittee (DVS) has established a by local ads and short programs is very splicing standard DVS 253 “Digital Program significant. Current ad insertion systems are Insertion Cueing Message for Cable” to hybrid systems, where commercials are stored provide cue messaging and splicing in a more in the ad server in digitally compressed form, digital headend friendly manner, which does while splicing is accomplished using analog not require restrictions or constraints on techniques. The compressed commercial is MPEG-2-compliant transport streams. This decompressed and converted to analog format paper will present a brief overview of existing before insertion into an analog channel. The analog/hybrid program insertion systems. The complete digital technology for delivery of limitations of SMPTE 312M will be compressed digital audio, video, and data is discussed. The solutions employed in DVS superior to existing analog methods. However, 253 will be described including digital dealing with compressed video poses several headend friendly features. The implementation challenges. One such challenge is splicing into of DVS 253 to insert compressed commercials a compressed digital bitstream compliant to at the headend, including the issue of the MPEG-2 standard seamlessly, that is to invisibility from commercial killers, will be say, without adversely affecting the display addressed. As cost is related to complexity due to decoding discontinuities at the insertion and flexibility of implementation, complexity point. These applications involve insertion of issues and network operational constraints will locally generated compressed digital also be examined for various implementations commercials and short programs into a digital of digital program insertion systems and channel containing previously multiplexed constituent components. and digitally compressed programs. Introduction The Society of Motion Picture and Television Engineers (SMPTE) has defined Traditionally, the cable industry has been a standard SMPTE 312M “Splice Points and popular conduit for the delivery of premium Cue Messages for MPEG-2 Transport analog television content to the home. The Streams” to promote interoperability of access to the broadcast premium content was compressed digital video splicing equipment. controlled using an analog set top box at the SMPTE 312M is aimed at studio broadcast consumer's premises. The signal scrambled at and requires a constrained (a synonym for pre- the headend could only be descrambled using conditioned) MPEG-2 transport stream. For proprietary hardware embedded in the set top use in deployed digital cable systems, this will box. The advanced audio/video compression technology and subsequent standardization Until a few years ago most of the ad- along with advances in digital modulation has insertion systems were analog where the ad- made delivery of digital audio/video to the insertion equipment is analog and analog consumer's home a reality. The digital storage devices were used to store the ads. technology is very bandwidth efficient and However, the majority of the ad-insertion provides superior video quality compared to its systems have been replaced by a hybrid one. In analog counterpart. In the last few years the a hybrid system, the ads are stored in digital cable industry has been deploying delivery of compressed format and the ad insertion digital signals using a hybrid analog/digital set equipment is analog in nature. In the hybrid top box mainly for premium and pay-per-view system, the digitally compressed commercial is (PPV) services. being decoded and converted to analog before insertion. Over the years, cable systems have grown significantly to become a primary conduit for Analog Ad Insertion the delivery of television contents to the majority of the homes. The industry also started In the analog domain, splicing between two inserting commercials both at national and local NTSC video sources or TV programs is relative- levels in some channels to enhance their ly easy. A switch between two synchronized revenues. Currently, the revenue generated video sources at any frame boundary (i.e. the from local ads and short programs contribute vertical blanking interval) can be accomplished, significantly to the total revenue of the cable since the frames are equal in size and time industry. The revenue generated from local ads duration and are synchronized to one another. and short programs has been growing The resultant video, with the switched video significantly. source at the insertion or splice point, will appear seamless when displayed. Typically when a content is created, a few avails are included to insert commercials. A Figure 1 shows the block diagram of a typi- certain percentage of the avails are used for cal commercial insertion system in cable head- national ads and the rest for the insertion of local ends and broadcast stations. In existing analog commercials. However, the networks (such as systems, networks distribute their content to ca- CNN, TBS, etc.) fill all the avails with the ble headends and broadcast affiliates via satellite national ads before to the headends modulating a baseband bandwidth close to using satellite. The headends can replace a 10 MHz. As shown in Figure 2, the lower por- percentage of these national ads with locally tion of the transmitted signal spectrum is used inserted commercials using the network cue for NTSC video. The upper portion of the spec- signals. Basically the cue signal has three trum above 4.2 MHz contains a few FM components, the Pre-roll (setup time), the Start , which are used to transmit mono, (beginning of an avail) and the Stop (end of the stereo, multi-language audio, and data. One of avail). The Pre-roll is sent in advance to allow the subcarriers is used to transmit cue-tone downstream devices at the headend (such as the signals. The is modulated with a cue- ad-server and the splicer) time to prepare for the tone signal using frequency modulation (FM) or insertion. When the Start signal is received, the frequency shift keying (FSK). Some networks splicer switches from the network program to send cue-tones using a pair of FSK tones (e.g., the commercial. After receiving the Stop signal 25 Hz and 35 Hz) within one of the audio the splicer leaves the commercial and returns to channels. The integrated receiver-decoder (IRD) the network program. receives the RF-modulated signal, demodulates the cue-tone subcarrier signal, and provides an output signal. Either a dual-tone multiple cation of the spot is signaled by a cue-tone frequency (DTMF) audio cue-tone sequence during program broadcasts. The cue-tone con- and/or a relay contact closure signal are sists of a sequence of numbers indicating the provided to the insertion equipment. The start and the end of an insertion opportunity. For networks send their schedules and spot example, the Discovery Channel uses a DTMF availability (avail) in advance for local ad- local avail cue-tone with 826* indicating the vertising insertion in a program. The precise lo- start, and 826# indicating the end of the spot.

Figure 1. Typical Commercial Insertion System

Figure 2. Analog Baseband Used by Networks

Digital Program Insertion completed, the Society of Motion Picture and Television Engineers (SMPTE) decided to At the time of standardization of the create a standard for splicing compressed MPEG-2 compression and system/transport digital streams. SMPTE completed the layers [1], it had been expected that switching splicing standard in 1998 and is designated as and splicing of compressed elementary SMPTE 312M [2]. SMPTE 312M is more streams and transport streams will be needed suited to post production studios. Some of the for editing commercial insertions, etc. major difficulties with SMPTE 312M are: Subsequently some hooks were provided for this purpose. After MPEG-2 standards were The splicer requires a constrained input correction needed after remultiplexing may bitstream. also be transmitted to maintain timing accuracy of the cue signal. Cue Signaling is not flexible enough. Cue signaling can only be sent to perform splicing A cue message has basically two on all the component streams of a program but components – the first schedules the insertion not on a single component if needed. points far in advance (such as the previous week). The second defines frame accurate No encryption mechanism for cue insertion points. These messages are carried signaling was provided. in an MPEG-2 private section syntax and called a splice_information section. A splice The impact of the requirement of schedule defines the insertion time using constrained bitstreams for splicing alone is coordinated universal time (GPS UTC) significant. For example, existing MPEG-2 accurate to one second, while the splice insert compliant TV contents have to be pre- uses MPEG PTS time, which enables frame conditioned before any local commercials can accurate insertion. be inserted into them. Also the installed encoders either have to be modified or Another important feature added to DVS replaced by the new ones. In essence, the 253 is the capability to encrypt cue messages. standard is not very cable friendly for insertion This will enable selective authorization of of local compressed commercial in a cable splicing devices in multiple headends. If a centric statistical multiplexing environment headend is not authorized, its splicing device [3]. The Society of Cable Tele- will not be able to identify the insertion spot. communications Engineers (SCTE) Digital The headends, which are authorized, will get Video Subcommittee (DVS) reviewed the the correct information about an avail. SMPTE standard in detail and decided to work out a separate, more cable-friendly In addition to these two message standard. SCTE established the digital components, each splice command enables program insertion ad-hoc group to work out splicing of complete programs or individual such a standard in 1998. After working for components of a program (such as Video or more than a year, the ad-hoc completed their Audio or Data). Today’s systems use only the work, which has become DVS 253 titled program level splicing where all the “Digital Program Insertion Cueing Message components of a program are replaced at the for Cable” [5]. splice point. In the future, we will see programs and advertisements that are In creating such a standard no ‘enhanced’ using data broadcast as well as precondition or constraint for the MPEG-2 programs that offer interactivity. These transport stream is assumed. The standard systems will use the component level splicing does not control the behavior of the splicing where only selected components of a program device in any way. The cue message carries are replaced at the splice point. The timing information, which the splicing device component level splicing enables pre-loading may use to perform the splice. In other words, of data streams that are part of the same it may be treated as an opportunity for splicing program by splicing the data component ahead but not a command. Cue messaging, if of video/audio content. This may be done to required, may be passed on to the downstream load and run the data enhancements in the equipment such as a pass-through via a receiver’s application engine. In the future, we remultiplexer to a set-top box. The timing will also see part of the bandwidth being used for ‘opportunistic data’ carriage and these additional components even though they are components of the program are not spliced not present in the original program. At the over during the insertion. splice point, these components are enabled or disabled. Typical enhanced commercials The components of a program are could include delivery of discount coupons, identified using the component_tag defined in prize drawings, or free software. the program map table using a new descriptor called the stream_id_descriptor. In this The use of the splice event_id and command mode, each component of the equipment interfaces such as between the program can signal a different PTS time for splicer and ad_server are not defined. The insertion and the components that are not DPI group has started working on the signaled are continued in the stream. When definition of APIs between the splicer and the inserted streams (such as enhanced compressed commercial servers. DVS 253 commercials) have more components than the along with standardized APIs will allow network program, the program map table will splicers and commercial servers to be typically include the definition of these interoperable.

Satellite

Program 1 • • QPSK Program N Multiplexer Modulator & Transmitter Encryptor Data

Network (HBO, CNN, TBS, etc.) Uplink Center Cue Message Generator

Figure 3. Digital Cue Messaging System (per DVS 253)

Figure 3 shows a typical block diagram multiplexed with the other elementary streams depicting how cue messages will be to form an MPEG-2 compliant transport multiplexed with a program before stream (TS). This baseband TS then can transmission to the headend via satellite. A digitally modulate a RF carrier and be Cue Message Generator generates a cue-signal distributed to headends. per DVS 253. The digital cue message gets Satellite

Baseband Other Multiplex signals with cue *2 Mux 1 Integrated message • Remux Receiver Splicer 1 • QAM HFC • or Transcoder Mux N Modulator Rate remux •

• Combiner • • STB Baseband Multiplex • • • with cue message • • Splicer N • Remux Ad_Server 1 • TV • or Encryptor & • Modulator • Rate remux • Local Cue Message Ad_Server N Generator MSO Headend

Figure 4. Typical Commercial Insertion System per DVS 253

Figure 4 shows the block diagram of a The cue message standard DVS 253 does digital headend where local commercials are not specify how to splice between two inserted into a program utilizing the cue bitstreams. The techniques and resultant message multiplexed in the transmitted constraints for splicing compressed streams program. The satellite integrated receiver have been left to the innovation of the splicing demodulates the RF signal to its baseband equipment manufacturer. Splicing may be MPEG-2 transport stream and decrypts it. designed to splice seamlessly, near- The cue message, multiplexed in the transport seamlessly, or in a non-seamless way. Also, stream is detected and decoded by the splicer. the compressed bitrate of the inserted ad may If it is a "pre-roll" it communicates with the need to be reduced to the original program ad_server to prepare. When cue message with compressed bitrate to provide a seamless a "splice_insert" is detected, the splicer transition (i.e. the spliced stream cannot switches from the network program to the violate the MPEG-2 TSTD buffer model). local ad. At the end of the avail, the splicer Figure 5 shows the typical trajectory of a switches back to the network program. A decoder buffer fullness. The constant headend may have multiple splicers and transmission rate of the compressed video ad_servers interconnected among them so that bitstream fills the buffer and the decoder the same commercial need not be stored in removes variable size compressed frames more than one server. If for any reason a (such as I, P or B) every display period. For network cue signal is not detected or decoded the MPEG-2 main profile at main level properly, a local cue signal generator may be (MP@ML), the minimum size of the decoder used instead. The local cue signal generator buffer must be 1.8 Mbits. Decoder buffer may be used as a backup system. overflow or underflow should be avoided so that no audible or visual artifact is created. Compressed Video Uncompressed Video Video Buffer (variable frame size) Decoder (fixed frame size)

Splice Point Splice Point (out) (in or out) Buffer Fullness (bits) I P I B B B B B

Start-up Delay Splice Decoding Time Delay

A Typical Buffer Trajectory

Figure 5. Simplified Representation of a Video Decoder

operator selected programs such as local ads. Splicing for Unconstrained, Seamless A few of the important functions performed Insertion by a remultiplexer are:

Statistical multiplexing of compressed Demultiplexes the input multiplexed video streams helps to utilize digital channels streams (unbundles the individual programs); better than constant bitrate encoded streams, assuming that the peak demand for bits from Creates a new multiplex out of the all video encoders does not coincide. The operator-selected programs and includes PSI important constraint here is that all the videos for the new multiplex; need to be encoded while multiplexing. A previously compressed stream typically had to Maintains the bitrate constraint such that be decompressed and re-encoded before it can sum of all elementary stream bitrates, be inserted in a statistical multiplex. An including PSI does not exceed the alternative to such an approach to splicing transmission channel rate; compressed video streams is to use remultiplexing or “grooming” so that new Removes jitter in Program Clock services such as digital ad insertion may be Reference (PCR) time stamp values and added more efficiently without compromising maintains A/V synchronization within the visual quality due to cascaded decompression programs; and recompression. Provides perceptually seamless switching A remultiplexer receives one or more capability from one program to the other multiplexed streams as input and creates a without any audible or visual artifacts. new output multiplexed stream from local Nominally, a remultiplexer does not alter bitrate of the local compressed commercial the bitrates while constructing a new multiplex with that of the remotely transmitted program, out of the input multiplexed streams. The or creating and storing different bitrate technology that deals with the multiplexing of versions of the commercials in the ad server. compressed video streams along with other streams (audio and data), and trims the Related Standards Development resulting multiplex to match an assigned constant total transmission channel bitrate, is Currently two work items are in progress known as rate-remultiplexing [4]. Rate- in the Digital Program Insertion Ad-Hoc remultiplexing meets the latter constraint by Group of the SCTE DVS Subcommittee – transcoding individual video stream within the development of a standard splicing API and a output multiplex. Transcoding is the technique splicing Guide for the use of DVS 253. by which a compressed video stream is translated to a lower bitrate strictly within the The goal of the API is to create a common compressed domain. Thus without cascaded interface for communication between the Ad compression, degradation in picture quality is Servers and Splicers for inserting ads into any not noticeable with occasional or moderate multiplex. The API will be flexible enough to reductions in the average bitrate of individual support multiple Ad Servers attached to one video streams. Splicer and one Ad Server attached to multiple Splicers. The above advantages of rate- remultiplexing make it a very attractive choice The Guide will explain briefly the features over standard remultiplexing. Rate- of DVS 253. This will help not only to remultiplexing combines existing minimize the chance of misunderstanding of remultiplexing technology with a new the features of the standard but will also capability known as transcoding. Transcoding reduce the miscommunication between the can reduce the bitrate of MPEG-2 compressed creators of DVS 253 messages (the networks) video without fully decoding and re-encoding and the ad insertion equipment manufacturers. a bitstream with the attendant loss of picture ATSC specialist group T3/S8 has just quality inherent in such cascaded completed the specifications for ‘Directed compression. Channel Change’ (DCC). This is an extension In creating compressed commercials, to PSIP standard and specifies the signaling to content producers can produce one high- automatically switch a ‘virtual channel’ based quality version and store it in a server. But, on the user preference at appropriate times. based on the availability of digital bandwidth The specification also signals the end of this in the multiplex at various times and at various event so that normal programming can systems, the commercial’s bitrate will have to resume. This feature is expected to be used for be reduced during insertion. It is also possible applications such as targeted advertisements to create different versions of commercials as well as enhanced advertisements. The time with different bitrates. However, storing reference in this specification is based on different versions of the same commercial UTC/GPS time and is only accurate to 1 could be redundant if rate-remultiplexing is second and the specification is very similar to employed. Rate-remultiplexing technology the splice schedule command. Acquisition of provides the capability to insert compressed the new ‘virtual channel’ as well seamless digital commercials into digital channels at the transition to the new channel requires splicer headend and removes the need to match the architectures that are very similar to what is presented in this paper. In this model, the S. Prodan, Mukta Kar, and Majid Chelehmal, streams for advertisement are expected to be NCTA Technical Papers, 1999. within the same transport multiplex identified by a different virtual channel number so that 5. Digital Program Insertion Cueing the channel change can occur at the receiver. Message for Cable, SCTE Standard DVS 253, In the cable headend model the server is December, 1999. resident at the headend and the switching occurs in the headend instead of at the end Acknowledgements user equipment. The authors would like to thank Carol Conclusion Derr of AT&T, Joe Davis of Atlantek Electronic Development Corp. and other In summary, the problems and solutions members of the Digital Program Insertion for flexible, unconstrained insertion of local (DPI) Adhoc group of SCTE for their digital compressed programming such as local participation in the various digital program ads into remotely downlinked digital video insertion technology development and multiplexes has been described. The evolution standardization activities. of technology from previous analog and hybrid ad insertion systems to this new digital insertion/remultiplexing technology has been described. The features of the new digital program insertion cueing message standard for cable were summarized. Splicing technology using rate-remultiplexing for unconstrained, seamless local insertion and grooming of remotely compressed digital video programming was described. Related standards developments were noted.

References

1. CODING OF MOVING PICTURES AND ASSOCIATED AUDIO, ISO/IEC 13818-1 (Systems), ISO/IEC 13818-2 (Video).

2. Splice Points for MPEG-2 Transport Streams, SMPTE Standard 312M, December 1997.

3. Digital Program Insertion for Local , by Mukta Kar, Majid Chelehmal and Richard S. Prodan, NCTA Technical Papers, 1998.

4. Rate-Remultiplexing: An Optimum Bandwidth Utilization Technology, by Richard