`With 34 Mbitls
`
`Digital modulation methods for T V satellite transmission can offer advantages in satel-
`lite transmission power and bandwidth if the bit rate can be reduced to 34 Mbit/s (to
`34.368 Mbit/s, the bit rate for 480-channel PCM systems). A very efficient coding
`technique, differential pulse code modulation (DPCM), is described, and its application
`for T V transmission is shown. A test system for demonstrating the feasibility of digital
`T V transmission is described in detail. In this system the composite T V signal is split
`into its luminance and chrominance components. Tests via directional radio and satel-
`lite line have been carried out using a modem with 4-PSK (phase-shift-keying) as
`interface. After DPCM encoding these picture signals and accompanying sound signals
`are multiplexed and transmitted together with a code word for synchronization. On the
`receive side the signals are demultiplexed and after decoding are provided as analog
`signals. An example of digitally transmitted pictures is given to show that good picture
`quality has been achieved.
`
`Introduction
`Digital transmission of TV signals via
`satellites can offer advantages compared to
`analog
`transmission. These advantages
`come partly from the inherent properties of
`digital transmission and partly from the
`transmission link via satellite.
`In satellite transmission it is highly de-
`sirable to keep the transmission power as
`low as possible; therefore, power-saving
`modulation methods which exchange
`power consumption and bandwidth have to
`be used. In the following, satellite trans-
`mission power requirements are compared
`between frequency modulation and digital
`phase-shift-keying (PSK) modulation.
`Assuming a weighted signaunoise ratio
`of 52 dB and the parameters of a 12-GHz
`broadcast satellite link (Ref. I), the satellite
`transmission power P has been calculated
`as a function of transmission bandwidth B
`(as described in Ref. 2). The result is
`shown in Fig. I . The dashed curve shows
`the power requirements for FM transmis-
`sion assuming a bandwidth of the modulat-
`ing TV signal of 5 MHz (video bandwidth
`of PAL G).
`The FM curve shows that the bandwidth
`can be reduced at the expense of the trans-
`mission power, or vice versa, while in the
`case of digital transmission, power can be
`saved if the bit ratc can be reduced.
`The solid curve shows the satellite
`transmission power for 4-phase shift key-
`ing (4-PSK) with differential encoding and
`coherent demodulation and the transmis-
`sion bit rate as a parameter. This bit,rate
`depends on the source encoding of the
`video signal. Applying straightforward
`PCM encoding will result in a bit rate of 80
`Mbit/s for a color TV signal. The transmis-
`sion of sound further increases the bit rate.
`
`A contribution submitted on 16 April 1979 by Ro-
`land Burkhardt and Josef Wasser. Standard Elektrik
`Lorenz AG, Box 40 07 49, Hellmuth-Hirth Strasse
`42. D-7000 Stuttgart 40. Federal Republic of Ger-
`many. Copyright 0 1980 by the Society of Motion
`Picture and Television Engineers, lnc.
`Note: The work described in this report has been
`carried out under contracts with the German Minis-
`try for Research and Technology. The authors are
`solely responsible for its contents.
`
`It is evident from Fig. 1 that, in this case,
`digital transmission is inferior to FM;
`however, if the bit rate can be reduced to
`less than 50 Mbids, digital transmission
`becomes superior. This can be achieved by
`applying more sophisticated source encod-
`ing techniques than PCM. These methods
`are described in the following section.
`Before recommending the introduction
`of a digital TV transmission system,
`however, sufficient test results on picture
`quality, transmission errors, reliability,
`comparability with other systems, and indi-
`cations for possible system simplifications
`must be available. For this purpose, we at
`Standard Elektrik Lorenz AG (SEL) have
`developed a system for testing the digital
`transmission of a color TV signal including
`two broadcast sound signals at a bit rate of
`34.368 Mbids, in accordance with Refs. 2
`and 5 .
`Bit-Rate Reduction
`Television pictures contain in general a
`large amount of redundancy because of the
`statistical dependencies of the picture sam-
`ples. This high redundancy stems from the
`fact that often a picture consists of areas of
`the same brightness level.
`Encoding of the components of the TV
`signal offers a way of achieving bit rates as
`low as 34.368 Mbit/s with good picture
`quality. This bit rate is highly attractive be-
`cause it would fit into the third-order Euro-
`pean PCM hierarchy. Figure 1 shows that
`at 34 Mbit/s the necessary bandwidth is
`about 22 MHz. With this bandwidth, a re-
`duction in transmission power by a factor
`of nearly five, compared with FM, is
`achieved.
`For applying bit-rate reduction methods
`to the digitization of TV signals, there are
`basically two approaches. The TV signal
`can either be treated as an entirety (encod-
`ing of the composite signal) or split up into
`its components, luminance and chromi-
`nance, which are then encoded separately.
`At first glance, encoding of the composite
`signal appears more attractive since it pre-
`standardized TV
`signal.
`serves
`the
`However, bit-rate reduction methods for
`color TV signals are not very efficient, re-
`
`244
`
`SMPTE Journul April I980 Volume
`
`89
`
`By ROLANDBURKHARDT
`and JOSEF WASSER
`
`quiring minimum bit rates of at least 60
`Mbit/s (see Refs. 3 and 4), and thus do not
`provide the power-saving advantages men-
`tioned above. Moreover, these methods,
`based on linear prediction, can be applied
`only to PAL and NTSC standards. They
`are not applicable to SECAM because
`SECAM chrominance signals are transmit-
`ted by frequency modulation, which is a
`nonlinear operation. Therefore, bit-rate re-
`duction methods based on elimination of
`linear statistical dependencies are due to
`fail.
`
`Principle of DPCM
`Today’s most commonly applied redun-
`dancy reduction method is DPCM (differ-
`ential pulse code modulation), which is
`especially well suited because of its high
`efficiehcy of achievable bit-rate rcduction
`and acceptable system, complexity. In
`DPCM the difference between the actual
`sample and an estimated value obtained
`from a predictor is transmitted.
`Instead of transmitting the picture sam-
`ples, after appropriate quantization and en-
`coding in PCM, a difference between
`actual picture sample x and a predicted
`value 2 is transmitted. This difference is
`quantized. It is then fed into the DPCM
`feedback loop and also transmitted to the
`receiver. As can be seen in Fig. 2, the feed-
`back loops in the transmitter and the re-
`the
`ceiver are identical. The task of
`receiver or feedback loop is to add the
`quantized difference d,, to the predicted
`value f . This will result in a reconstituted
`signal x, which approximates the input sig-
`nal P the closer, the smaller the quantiza-
`tion distortions are.
`The reason that the PCM bit rate may be
`reduced by DPCM encoding is based on
`the fact that the transmitted difference - if
`the prediction is a good one - will be
`much smaller than the actual sampled
`value. Thus, the difference can be encoded
`with fewer bits per sample.
`In a DPCM system, however, typical
`quantization errors can occur. If the small-
`est quantization step is chosen too large
`this will result in a coarse reproduction of
`picture areas with nearly constant bright-
`ness level or amplitude. This effect is
`called “granular noise” and is rather an-
`noying to the human observer since the eye
`is sensitive to noise, especially in such pic-
`ture areas. On the other hand, large quan-
`tization. steps must be available for re-
`production of large differences resulting
`from inaccurate prediction at edges or con-
`tours. If the available quantizatioq steps are
`too small, this will result in an effect called
`“slope overload” which means that the
`DPCM signal cannot follow the actual
`sample quickly enough. The effect on pic-
`ture quality is a loss of definition. These
`
`Authorized licensed use limited to: Kirkland & Ellis LLP. Downloaded on August 03,2017 at 14:34:14 UTC from IEEE Xplore. Restrictions apply.
`
`APPLE EXHIBIT 1068
`APPLE v. PMC
`IPR2016-01520
`Page 1
`
`
`
`contradictory demands for quantization
`step size in picture areas and at contours
`can be satisfied partly by a nonlinear quan-
`tization characteristic. However, coarse
`quantization of edges is liable to introduce
`another effect called “edge busyness” re-
`sulting from nonidentical reproduction of
`edges in consecutive lines. This is due to
`unavoidable noise in the picture signal
`leading possibly to different quantization
`steps at edges.
`In the test system these disadvantages
`have been largely avoided by using a con-
`trolled quantizer, which applies a chosen
`quantizing characteristic according to the
`picture contents. This will be discussed in
`detail in a section to follow on luminance
`signal processing.
`Concept of the Test System
`The block diagram of Fig. 3 shows how
`this DPCM method is applied separately to
`the processing of the luminance signal and
`the two chrominance signals (coding of the
`components).
`On the transmitting side the video com-
`ponents Y (luminance), R - Y, B - Y
`(chrominance) are digitized and the bit
`rates reduced in DPCM systems. A sync
`code word generator provides synchroniza-
`tion code words for identifying the frame
`start of the video signal. Together with two
`digital sound signals TI, T z , video and syn-
`chronization information is multiplexed.
`Because protection against transmission er-
`rors is of much greater importance for sig-
`nals with reduced redundancy
`than for
`
`PCM encoded signals, error protection will
`generally be necessary. The error protec-
`tion equipment has not been developed by
`SEL because several possibilities are al-
`ready knownIO for reducing bit error rate
`(BER).
`The digital TV signal can be transmitted
`by various transmission media, such as sat-
`ellite or directional radio systems, coaxial
`lines, and optical fibers. An adaptation to
`the operating mode of the chosen transmis-
`sion medium by an appropriate interface
`circuit - for instance, 4-PSK for satellite
`transmission - is necessary. With a bit
`rate of 34.368 Mbit/s, terrestrial links with
`the third-order European PCM hierarchy
`(PCM 480) may also be used. On the re-
`ceiving side, the synchronization code
`word is extracted, and the necessary timing
`information for error decoding and de-
`multiplexing is derived. After error correc-
`tion and demultiplexing, the video signals
`are reconstituted in DPCM receivers to dig-
`ital PCM signals. DigitaVanalog converters
`provide analog video and sound signals.
`The test system has been designed to be
`as flexible as possible because the technol-
`ogy is new and there are many variants on
`the DPCM principle that could improve the
`picture quality. So far there is still not com-
`to which variant
`plete agreement as
`(method and parameters) is best suited for
`maximizing picture quality while minimiz-
`ing transmission error insensitivity and sys-
`tem complexity.
`After PCM coding, picture and sound
`signals are processed completely by digital
`
`FM tronsmissim
`\
`
`\
`\
`\
`\
`
`,
`
`W
`
`500 ””
`
`d i g i
`
`t T T ,
`
`,
`
`40 MHz 50 B
`Fig. 1. Comparison of satellite transmission
`power (P) and bandwidth (B) for FM and for
`4-PSK transmission.
`
`means. This approach offers advantages of
`accuracy and reproducibility of the video
`signal as well as allowing for integration of
`functional units in MSI and LSI circuits.
`The exact reproducibility of digital circuits
`also simplifies the problem that in DPCM
`systems some functional units such as the
`adder and predictor - which are commor,
`to the transmitter and receiver unit - have
`to provide absolutely identical signals for a
`correct reconstruction of the transmitted
`signal. It is furthermore necessary for an
`exact prediction to synchronize the sam-
`pling frequencies and the picture process-
`ing clocks with the picture signal and to
`keep them so stable that the temporal sam-
`pling error in the picture signal is small
`compared to the sampling interval. This re-
`quirement is met in the test system by lock-
`ing all clock frequencies in a phase locked
`loop with a voltage controlled crysta1 os-
`
`C coder
`0 quantizer
`Edecoder
`P predlctor
`Fig. 2. Principle of differential pulse code modulation
`(DPCM).
`
`S
`
`11ine
`
`s i ~ l
`
`R - Y / B - Y 1
`a015 MHz 1L
`0015 MI+
`14
`
`7 2
`
`inSpsibit rotein6Lps
`h;role
` i5bondw~; ~;its/sample ~sump;rat:~~
`10.0 MHz 300 Mbit/s
`21.375Mbit/s
`6.500 Mbit/s
`21) MHZ
`8 0 Mbit/s
`0.W8 Mbitls
`3125 kHz
`0.438 Mbit/s
`3.25 kHz
`ro e
`resul ing
`31.751 Mbit/s
`error protection oncl
`additioml informtion
`2.61 7 Mbit/s
`fronsmlssion bit mte
`%.368Mbit/s
`Fig. 4. Frame structure and bit rates.
`
`transmitter
`
`Fig. 3. Block diagram of the test system for satellite and terrestrial
`transmission.
`
`predictor
`receiver
`Fig. 5. Block diagram of luminance processing.
`
`Burkhardt and Wasser: Digital TV Transmission With 34 Mhitls
`
`245
`
`3 bit .
`
`Y
`
`Authorized licensed use limited to: Kirkland & Ellis LLP. Downloaded on August 03,2017 at 14:34:14 UTC from IEEE Xplore. Restrictions apply.
`
`APPLE EXHIBIT 1068
`APPLE v. PMC
`IPR2016-01520
`Page 2
`
`
`
`- - - - - - - _ - - - - - - - - -
`-
`-
`
`n
`
`-
`
`
`
`.
`
`interlaced
`line
`
`U U I
`
`A
`
`line
`
`transmitter
`
`-
`A
`prectlctions
`
`.
`
`-
`
`.
`
`n
`
`-
`-
`_ m ,
`X
`
`G .Bta
`;=A
`;,As 2
`2
`;;=A*?
`Fig. 6. Picture element configuration and
`predictions.
`
`cillator to the line frequency of the picture
`signal.
`Figure 4 shows how the combination of
`the luminance, chrominance, sound, and
`sync signal in the lines of the digitized TV
`picture results in the bit rate of 31.7 Mbit‘s.
`The two chrominance signals are transmit-
`ted line sequentially. This results in the
`halving of the bit rate for the chrominance
`coding but not in an intolerable loss in ver-
`tical resolution. Because the synchroniza-
`is carried out once per frame,
`tion
`providing the horizontal blanking interval
`for picture signal transmission, the bit rate
`for the picture encoding is reduced accord-
`ing to the ratio of the active to the total line
`period. The resulting bit rate for the encod-
`ing of luminance, chrominance, and two
`high-quality sound channels is 31.7 Mbit‘s.
`The final bit rate for transmission is 34.368
`Mbit‘s thus allowing for an additional re-
`dundancy of 8.24% of the resulting bit rate
`which can be used for error protection. It
`will be necessary to add some useful redun-
`dancy in the form of an error correcting
`code, because the transmitted signal itself
`has largely lost its own redundancy by
`DPCM processing and is more sensitive to
`transmission errors.
`To reduce the BER from 10- to 10- by
`error protection equipment, less than 3% of
`the bit rate is necessary for error correcting
`code information. This would enable the
`test system to transmit additional informa-
`tion, such as commentary channels, within
`the transmission bit rate of 34.368 MbitJs.
`
`Functional Units of the Test System
`Processing of the Luminance Signal
`(Fig. 5 )
`The luminance signal is sampled with
`10 MHz, converted to 8-bit PCM, and
`coded with a DPCM system using two-
`directional prediction and a controlled
`quantizer.6 This means that the amount of
`the difference of two picture elements al-
`
`780
`
`208
`
`2 8
`
`72
`
`~-
`Fig. 7. Block diagram of chrominance processing.
`
`ready processed yields a control signal
`which selects the optimal quantization
`characteristic for processing the actual pic-
`ture element. Depending on the absolute
`difference between two samples, one of
`three quantization characteristics (el. Q2,
`Q3) having coarser or finer steps is chosen.
`In order to keep the transmitted difference
`signal as small as possible, the predicted
`value of the next picture element must be
`very accurate. This is obtained by the two-
`directional predictor which uses previous
`picture elements of the same line as well as
`picture elements of the previous line.
`The subtractor forms the difference
`(x - i ) between the signal of the actual
`picture sample x and the corresponding
`predicted value 2 .
`The quantizer subdivides the range of
`the difference signals (2256 are possible)
`into only eight subranges. These parts cor-
`respond to the steps of the quantization
`characteristic. These eight steps are en-
`coded in three bits and transmitted to the
`receiver. If different quantization char-
`acteristics are used, the size of the sub-
`ranges must be altered.
`The demonstration model uses up to
`three different quantization characteristics.
`Each of these can be selected by switches,
`while in the dynamic mode they may be
`chosen under control of the predictor. The
`decoder assigns to each subrange transmit-
`ted by the quantizer a certain code word
`which gives the actual value of the quan-
`tized difference signal.
`Depending on the sign bit, the adder
`unit adds or subtracts the 8-bit representa-
`tive words of the decoder and the corre-
`sponding predicted value. In addition to
`these simple arithmetic operations, the ad-
`der provides a double-sided limitation to
`“black” and “white.” The limitation is
`necessary because the output signal of the
`adder may be outside the possible 8-bit
`range of 256 quantization levels due to the
`inaccuracies introduced by the quantiza-
`tion.
`The predictor can be switched to four
`different algorithms. In the simplest one,
`the predicted value of the picture sample f
`is given by the previous sample A (Fig. 6).
`In this way, the horizontal edges in the TV
`picture are predicted optimally but this pre-
`
`diction will fail for all other directions, es-
`pecially vertical edges. This drawback can
`be overcome by a two-dimensional predic-
`tion using picture elements from the previ-
`ous line. Computer simulation and subjec-
`tive tests have shown that the prediction
`Z = A + (C - 8)/2 in general yields the
`best results. If the previous line is stored
`some other predictions are available and re-
`alizable, for example Z = (A + D)/2 or
`(A + C)/2. For testing and comparing the
`different prediction algorithms they can be
`selected manually by switches.
`Depending on the difference (D - A1
`the quantizer characteristic is selected as
`described above, and the meaning of the
`transmitted code word is changed to the
`value corresponding to the related quan-
`tizer characteristic.
`The receiver for the luminance signal is
`shown in the bottom part of Fig. 5 . The
`input signal is a transmitted 3-bit DPCM
`word. Two bits are processed in the de-
`coder. Depending on the sign bit the adder
`unit adds or subtracts. Adder and predictor
`are the same as on the transmitting side.
`Chrominance and Sound Processing
`Figure 7 is a block diagram of the trans-
`mitter and receiver sides of chrominance
`processing. After digitization into 8-bit
`samples at a 2-MHz sampling rate, the sig-
`nals are processed alternately in a DPCM
`system. This system uses a simple previous
`sample predictor. Quality
`improvement
`compared to a conventional DPCM is ob-
`tained by using two’s complement arithme-
`tic and a special quantization characteris-
`tic.’ This approach does not require
`transmission of the sign of the transmitted
`difference which is quantized to 4-bit code
`words. After the DPCM receiver, both
`components R - Y and B - Y are provided
`simultaneously, replacing one of the actual
`components by the line-delayed signal.
`Digitization of the two sound signals is
`carried out with 14-bit linear PCM at twice
`the line frequency.
`Multiplexing and Modulation
`The multiplexed signal is a dibit stream
`consisting of the following parts (Fig. 8):
`2 x 780 luminance bits are followed by
`2 x 208 chrominance bits and 2 x 28
`
`Authorized licensed use limited to: Kirkland & Ellis LLP. Downloaded on August 03,2017 at 14:34:14 UTC from IEEE Xplore. Restrictions apply.
`
`APPLE EXHIBIT 1068
`APPLE v. PMC
`IPR2016-01520
`Page 3
`
`
`
`cw
`
`/.'I01
`
`bit
`
`R-Y
`8-Y
`
`Fig. 9. Multiplex unit.
`
`L
`
`-buffer
`B
`Cw =write counter
`Cr zreod counter
`M =rnulftplex unit
`I
`.interleaving unit
`
`FM transmission 1S/N =52 OdBweighted and
`5 '
`preemphazisedt
`?!L - - - - - - - -
`- - -
`transponder vrturntion
`
`-
`
`\ , frequency deviation
`
`with 36Mbit1s
`with 36Mbit1s
`
`0 n6
`
`0 1 , 'a
`
`MHZ
`
`a1
`
`
`
`02 - 02
`
`80 -
`
`
`
`78. 78.
`
`76.
`76
`
`io 22
`71
`16
`i0
`MHz
`81 F
`Fig. 10. Results of transmission test via Symphonie
`satellite.
`
`i e
`
`30 L A
`
`iL
`
`i6
`
`sound bits. The rcmaining 2 X 72 hits are
`reserved for rcdundancy bits as a protection
`against transmission errors and for addi-
`tional information. The total bit rate of the
`dibit stream is 17.184 Mbit/s; this results in
`a transmission bit rate of 34.368 Mbit/s.
`Figure 9 is a block diagram of thc mul-
`tiplexer. In order to obtain a lower process-
`ing bit rate in the memories the total
`information is spread across four interlcav-
`ing channels with 8.5 Mbit/s. Thc DPCM
`chrominance signal is already dividcd into
`four channels by the 4-bit encoding whilc
`the 3-bit encoded luminance information
`has to be converted from three channels
`with 520 bitslline to four channels with 390
`bits/line. The reading counter distributes
`the address bits to the buffers and simulta-
`neously, in its function as a central clock
`unit, providcs several timing pulses neces-
`sary for the correct filing of the signals in
`the multiplex frame.
`For transmission via satellite, thc dibit
`stream at the output of the multiplexer (2 X
`17.184 Mbit/s) modulates a 4-PSK modula-
`tor which operatcs with a 70-MHz carrier.
`The phase-shift is controlled by thc differ-
`ence of consecutive dibits.
`DPmoditliitiun iind Dmitltiplexing
`In the receiver the 4-PSK signal is de-
`modulated first to a dibit stream with the
`same bit rate as in the transmitter. Becausc
`of the differential cncoding in the modula-
`
`tor, the pha.e ambiguity of tP/2 (for ti = 0,
`. . . 3) can be eliminated.
`The following demultiplexer divides the
`incoming dibit stream into four channcls
`which are first buffered. After converting
`the divided bit streams into parallel form
`for luminance, chrominancc, and sound,
`the information is processed in DPCM dc-
`codcrs for luminance and chromin'ance.
`The sound signals are DIA-convertcd and.
`aftcr low-pass filtering, become available
`at the output of the receivcr.
`Both chrominance signals H - Y and
`R - Y are processed in the samc DPCM
`decoder because of their line scquential
`transmission. D/A conversion and low-
`pass filtering are necessary for monitoring
`the transmitted video signal with an analog
`monitor. Thc synchronization c(dc word is
`transmitted once each TV framc. It con-
`trols, together with a line sync combination
`and a central clock, the processing of the
`signals in the receiver. Bccause the func-
`tions of the multiplexcr are nearly inverse
`to those of thc demultiplexer it is not neces-
`sary to show a block diagram of the latter.
`
`Results and Future Work
`A test system for the digital transmis-
`sion of color TV signals and two broadcast
`sound signals was devcloped and realized
`in mid-1978. The picture quality attainable,
`after transmission over the complete sys-
`
`tcni, scorcs between 4 and 5 on the intcma-
`tionally used 1 to 5 picture quality scale.
`After conducting
`transmission
`tcsts
`ovcr coaxial links as well as directional-
`radio and fiber-optic channels. mcasure-
`mcnts wcrc rnadc in June 1978 in a trial
`broadcast via thc Symphonie satellite (Ger-
`man/Frcnch communications satellite for
`test purposes), from Raisting. the German
`ground terminal. The results achicvcd were
`wcll in agrccment with the thcoretical val-
`ues shown in Fig. I for the relation between
`bandwidth and transmission power for dig-
`ital (4-PSK)- and frequency-modulation.
`Figure I0 shows the measured depcnd-
`ence of EIRP (cquivalent isotropically radi-
`atcd power) versus Blf (intcrmediatc fre-
`quency bandwidth). Naturally, 4-PSK
`transmission could he made only with the
`transmission bit rate of the systcm (34
`Mbit/s) which requires a bandwidth of
`about 22 MHz. The EIRP for scveral bit
`crror ratcs is shown in the diagram. Addi-
`tional tests via directional radio equipment
`with simulated link attenuation have been
`carried out in order to obtain an impression
`of the influence of bit errors on thc attain-
`able picture quality. All these tests and
`measurements wcre camcd out without er-
`ror protection equipmcnt.
`The next step will be to test the systcm
`with crror protection equipment which is
`capable of reducing a bit crror rate of 10-
`to a value of 104'. This value is necessary
`
`Fig. 11. The test system with the Symphonie ground terminal.
`
`Fig. 12. Digitized picture.
`
`Hurkhordt ond Wcrswr: Digitul 7 V Trunsmission Wirh 34 Mbirls
`
`247
`
`Authorized licensed use limited to: Kirkland & Ellis LLP. Downloaded on August 03,2017 at 14:34:14 UTC from IEEE Xplore. Restrictions apply.
`
`APPLE EXHIBIT 1068
`APPLE v. PMC
`IPR2016-01520
`Page 4
`
`
`
`for achieving a good picture quality. As
`shown in Fig. 10, power reduction by this
`procedure is 4.5 dB.
`Figure 11 shows a photograph of the test
`system in front of the Symphonie ground
`terminal in Raisting. Because of the flexi-
`ble design of the picture processing equip-
`ment, the system is considerably larger
`than would be necessary for a final version.
`Figure 12 shows a transmitted digitized
`picture, with 3-bit DPCM, adaptive quan-
`tization, and prediction i = A + (C -
`B)/2.
`References
`I. Siemens AG. Standard Elektrik Lorenz AG,
`Messerschmitt-Bolkow-Blohm GmbH, “Kon-
`zeptstudie
`fiir
`einen
`Fernsehrundfunk-
`satelliten.” Phase A. Schlussbericht: Gesell-
`schaft fiir Weltraumforschung mbH. (“Feasi-
`bility study of a television broadcast satellite.”
`Phase A , final report: Society for Space Inves-
`tigation, Inc.), Vertrag (Contract) No. RV 1Yl-
`V14/72-HQ-01-00.
`2. Standard Elektrik Lorenz AG, “Autbau cines
`Demonstrationsger8tes
`zur digitalen Uber-
`tragung von Fernsehsignalen iiber Satelliten-
`strecken unter besonderer Beriicksichtigung
`eines wirtschaftlichen Heimempfiingers.”
`Gesellschaft fur Weltraumforschung mbH.
`(“Design of a demonstration apparatus for the
`digital transmission of television signals over
`satellite links with special consideration of eco-
`nomic home receivers.” Society for Space In-
`vestigation, Inc.), Vertrag (Contract) No. RV
`11-V4/73 (3)-TIIO.
`3. I. E. Thompson, “Differential Coding for Dig-
`ital Transmission of PAL Color Television Sig-
`nals,’’ Proceedings of the International Broad-
`casting Convention, London. 4-8 Sept. 1972,
`Insr. of Electrical Engrs. Conf. Publ. No. 88, pp.
`2 6 3 2 .
`4. I. E. Thompson, “Predictive Coding of Com-
`posite PAL and NTSC Color Television Sig-
`nals,” Institute of Electrical and Electronics
`Engineers, International Conference on Com-
`munications, Seattle, 11-13 June 1975. No. 48.
`pp. 3 2 3 8 .
`5 . H. 1. Klutz et al., “Test System for Digital TV
`Transmission,” Electrical Communication. 51:
`No. 2, 100-106, 1976.
`6. T. Kummerow.
`“Ein DPCM-System mil
`zweidimensionalem Priidiktor und gesteuertem
`
`Quantisierer,” Nachrichtentechnische Gesell-
`schaft Symposium (“A DPCM system with
`two-dimensional predictor and controlled quan-
`tizer,” Symposium of the Society for Cornmu-
`nications Technology), Erlangen, 4-6 April
`1973. Digest, pp. 425439.
`7. G. Bostelmann, “A Simple High-Quality
`DPCM-Codec for Video Telephony Using
`8 Mlbit per Second,” Nachrichtenfechnische
`Zeirschrifr, 27: 115-117, Mar. 1974.
`8. Standard Telecommunication Laboratories,
`“Digitalization of TV Signals,” Telecom-
`munication System Studies, ESRO Contract No.
`1765172 SW.
`9. 1. Wasser and W. Zschunkc, “Test System for
`Digital Satellite TV Transmission,” Proceed-
`ings of a GSA Symposium on Satellite Broad-
`casting Held in Stockholm, 22-24 Nov. 1976.
`ESA SP-122, Feb. 1977.
`10. R. Briiders et al., “Ein Versuchssystem zurdig-
`italen fjbertragung van Fernsehsignalen unter
`besonderer Beriicksichtigung
`von Ober-
`tragungsfehlern.” Festschrifr 50 Jahre Heinrich
`Hertz Instituf, Berlin GmbH (“A trial system for
`the digital transmission of television signals
`with special consideration of transmission er-
`rors,” Anniversary publication, 50 Years of the
`Heinrich Hertz Institute. Berlin, Inc.), Ein-
`steinufer 37, D-I000 Berlin 10.
`
`Comments Solicited by the
`Board of Editors
`
`Reviewers: It seems unlikely that the lu-
`minance TV signal can be DPCM encoded
`with 3 bits and still give good quality pic-
`tures. The authors use a fairly unsophisti-
`cated intrafield, two-dimensional predic-
`tion. According to V. G. Devereux (in his
`paper “Differential Coding of PAL Video
`Signals Using Intrafield Prediction,” in
`Proceedings IEE, Dec. 1977), 3-bit encod-
`ing will cause the picture impairment to be
`somewhat objectionable, and 5 bits per
`sample are needed for acceptable picture
`quality. One opinion is that 4 bits is the
`smallest number that can be tolerated.
`There is also question about using a 10-
`MHz sampling rate for a 5-MHz video sig-
`
`nal. Due to the finite response of the analog
`filters used with a PCM codec, it is imprac-
`tical to encode at the Nyquist sampling
`limit. The authors may be getting only
`about 4 MHz of Y response.
`The Authors: Our objective has been a
`coding method for color TV, including
`transmission, an objective sought by only a
`few other researchers.
`We agree that when using pictures for
`testing the resolution, 3 bits per sample for
`the luminance signal is not enough com-
`pared with analog transmission. But the
`CCIR Recommendation requests that test
`patterns not be used for subjective tests of
`picture quality. Our subjective tests have
`been carried out in accord with CCIR Rec.
`500, the slides being the three commonly
`used test slides titled Playboy, Strawhat
`Girl (Fig. 12), and Beach Scene, and two
`others (Cityscape and Puppet). The 25 test
`persons included five experts. In all cases
`we got a quality score of more than 4 on the
`CCIR scale, when the bit error rate was
`lo+ or less.
`The use of interframe coding would im-
`prove picture quality, but more sophisti-
`cated equipment would be required; we do
`not see a disadvantage in using an un-
`sophisticated method of encoding.
`We suggest that a 5-bit DPCM encod-
`ing for the luminance signal will not be
`standardized because of the final bit rate for
`transmission. The
`international
`trend
`seems to be for a standard of 140-Mbids
`PCM encoding in the studio area and of 34-
`Mbit/s DPCM encoding for transmission.
`In this case, a maximum of 4 bits for each
`luminance sample may be used for trans-
`mission purposes. In Webster’s “Digital
`Picture Coding’’ in Wireless World for Oc-
`tober 1978, we note advice that “Excellent
`picture quality can be achieved by means
`of a 3-bit difference signal for each picture
`element. ”
`
`248
`
`SMPTE Journal April 1980 Volume 89
`
`Authorized licensed use limited to: Kirkland & Ellis LLP. Downloaded on August 03,2017 at 14:34:14 UTC from IEEE Xplore. Restrictions apply.
`
`APPLE EXHIBIT 1068
`APPLE v. PMC
`IPR2016-01520
`Page 5
`
`