DVB-T
`
`Hierarchicalmodulation
`
`— the transmission of two independent DVB-T
`multiplexes on a single frequency
`
`Alexander Schertz and Chris Weck
`Institut f(cid:252)r Rundfunktechnik (IRT)
`
`Hierarchical modulation (cid:150) a variant of the digital terrestrial television standard,
`DVB-T (cid:150) has received relatively little attention to date in the planning processes. It
`enables the transmission of two independent DVB-T multiplexes on a single TV
`frequency channel, with different transmission qualities (high priority and low
`priority). The high-priority multiplex can be used, in particular, for portable indoor
`and mobile reception.
`On the basis of specific examples, this article shows that hierarchical modulation can
`be a worthwhile alternative to non-hierarchical modulation.
`
`In the DVB standard for digital terrestrial television (DVB-T) [1], hierarchical modulation is designated [2] as
`an alternative to the (cid:147)conventional(cid:148) modulation methods specified (QPSK, 16-QAM and 64-QAM). With the
`use of hierarchical modulation (HM), two autonomous DVB-T multiplexes can be transmitted on a single TV
`frequency channel. Up until now, relatively little attention has been paid to HM, possibly on account of the
`greater complexity of the DVB-T planning (cid:150) due to an additional degree of freedom being available when
`choosing the transmission parameters.
`This article provides an outline description of hierarchical modulation and, on the basis of several examples,
`demonstrates the possibilities offered by this transmission method. It discusses (i) the benefits and drawbacks
`of HM for portable, mobile and stationary reception from a single transmitter, (ii) the opportunities that are
`provided by HM for local and wide-scale DVB-T services and (iii) the impact of HM on self-interference in
`single-frequency networks (SFNs).
`
`1. Principles of DVB-T and hierarchical modulation
`1.1. Parameters of DVB-T
`The DVB-T standard (Digital Video Broadcasting (cid:150) Terrestrial) defines a method for transmitting MPEG-2
`encoded TV signals, adapted to the specific features of the terrestrial transmission channel (e.g. multipath
`reception). Its basis is the modulation technology COFDM (Coded Orthogonal Frequency Division Multi-
`plex) which uses thousands of narrow-band frequency carriers that are orthogonal to each other [3].
`Several parameters can be chosen for a DVB-T transmission channel:
`! The code rate
`This is the ratio of the data rate of the useful bits, to the overall data rate: typical values are 1/2, 2/3, 3/4,
`5/6 and 7/8). The greater the chosen code rate, the higher the available effective data rate but the worse
`the error protection and, consequently, the lower the coverage radius and/or the service coverage quality.
`
`EBU TECHNICAL REVIEW (cid:150) April 2003
`C. Weck and A. Schertz
`
`1 / 13
`
`Constellation Exhibit 2016
`LG Electronics, Inc. v. Constellation Designs, LLC
`IPR2023-00228
`
`

`

`DVB-T
`
`! The method of carrier modulation
`Three methods of modulation are currently used: QPSK 1, 16-QAM 2 and 64-QAM 3. In the case of the
`higher-level methods, the useful data rate is greater but the transmission is more susceptible to interfer-
`ence (because the separation between the permissible states in the phase space is smaller). Consequently,
`the coverage radius is smaller.
`! The length of the guard interval
`The guard interval extends the duration of each transmitted symbol (cid:150) in order to guarantee a safety inter-
`val for the subsequent symbol. The ratio of the guard interval to the total interval for each symbol is typ-
`ically 1/4, 1/8, 1/16 or 1/32. The longer the guard interval (tending towards 1/4), the less the interference
`due to multipath propagation and the weaker the inherent interference (self interference) when employing
`multiple transmitters in a single frequency network (SFN). However, the number of useful bits that can
`be transmitted are fewer.
`! The number of carriers
`There are currently two modes: 2k (with 1705 carriers) and 8k (6817 carriers).
`! Hierarchical or non-hierarchical modulation
`If hierarchical modulation is chosen, it is necessary to allocate a value to the modulation parameter (cid:147)α(cid:148).
`Typical values are: α = 1 (uniform modulation) and α = 2 or 4 (non-uniform modulation).
`
`64-QAM
`
`Q
`
`100000 100010
`
`101010
`
`101000
`
`001000
`
`001010 000010 000000
`
`100001 100011
`
`101011
`
`101001
`
`001001
`
`001011 000011
`
`000001
`
`100101 100111
`
`101111
`
`101101
`
`001101
`
`001111
`
`000111
`
`000101
`
`100100 100110
`
`101110
`
`101100
`
`001100
`
`001110
`
`0001101 000100
`
`I
`
`110100 110110
`
`111110
`
`111100
`
`011100
`
`011110
`
`010110
`
`010100
`
`1.2. Principle of hierarchical modulation
`The principle of hierarchical modulation will be
`explained here using 64-QAM as an example.
`Fig. 1 shows
`the constellation diagram of
`64-QAM. Each permissible digital state is repre-
`sented by a yellow dot in the complex plane
`(amplitude, phase). Because 8 x 8 different states
`are defined, 64 possible values of 6 bits can thus
`be transmitted. The diagram shows the assign-
`ment of the binary data values to the permissable
`states (as defined in the DVB-T specification).
`[In the case of 16-QAM, there are only 4 x 4 dif-
`ferent states (and 4 transmitted bits) while, in the
`case of 4-PSK, there are 2 x 2 states (and just 2
`transmitted bits).]
`In hierarchical modulation, the possible states are
`interpreted differently than in the non-hierarchical
`case. The location of a state within its quadrant is
`regarded as special information in HM. The other
`special information used in HM is the number of
`the quadrant in which the state is located (1, 2, 3
`or 4). In this way, two separate data streams can
`be made available for transmission. Formally, we
`are still dealing with 64-QAM but, in the hierar-
`chical interpretation, it is viewed as the combina-
`tion of 16-QAM and 4-PSK modulation. This is referred to as (cid:147)4-PSK in 64-QAM(cid:148). The bit-rates of the two
`partial streams together yield the bit-rate of a 64-QAM stream. Fig. 2 shows how, in the case of hierarchical
`
`110101 110111
`
`111111
`
`111101
`
`011101
`
`011111
`
`010111
`
`010101
`
`110001 110011
`
`111011
`
`111001
`
`011001
`
`011011
`
`010011
`
`010001
`
`110000 110010
`
`111010
`
`111000
`
`011000
`
`011010 010010 010000
`
`Figure 1
`Constellation chart for 64-QAM modulation
`(with α= 2 for hierarchical modulation)
`
`1. Quadrature Phase Shift Keying, referred to in this article as 4-PSK but also known as 4-QAM.
`2. Quadrature Amplitude Modulation using two 4-step orthogonal input signals.
`3. Quadrature Amplitude Modulation using two 8-step orthogonal input signals.
`
`EBU TECHNICAL REVIEW (cid:150) April 2003
`C. Weck and A. Schertz
`
`2 / 13
`
`Constellation Exhibit 2016
`
`

`

`DVB-T
`
`MUX
`adaption
`energy
`dispersal
`
`MUX
`adaption
`energy
`dispersal
`
`Outer
`coder
`
`Outer
`interleaver
`
`Inner
`coder
`
`Outer
`coder
`
`Outer
`interleaver
`
`Inner
`coder
`
`Video coder
`
`Audio coder
`
`Data coder
`
`1
`
`
`
`Programme MUX
`
`Transport MUX
`
`
`Video coder
`
`Audio coder
`
`Data coder
`
`2
`
`
`MPEG-2 Source Coding and
`Multiplexing
`
`Inner
`interleaver
`
`Mapper
`
`Frame
`adaptation
`
`OFDM
`
`Guard
`interval
`insertion
`
`D/A
`
`Front end
`
`To antenna
`
`Pilots &
`TPS
`signals
`
`Terrestrial Channel Adapter
`
`Figure 2
`Block diagram of a DVB-T system with hierarchical modulation
`
`modulation (for which the dotted-line blocks have been added), the input comprises two completely separate
`MPEG-2 transport streams. After combining at the inner interleaver, the composite stream is sent on its way
`to the transmitting antenna.
`
`It can thus be seen that hierarchical modulation allows two data streams (cid:150) each having different transmission
`performance (cid:150) to be made available within a single TV frequency channel.
`
`A comparison between 4-PSK in 64-QAM (hierarchical modulation) and 64-QAM (non-hierarchical modu-
`lation) is appropriate here. The separation of the states in the 64-QAM scheme shown in Fig. 1 does not
`change if the states within a block are interpreted as possible states of a 16-step modulation scheme. For that
`reason, the noise sensitivity is only slightly (and insignificantly) greater in the case of hierarchical modula-
`tion, compared to 64-QAM, on account of the slightly lower encoding gain of the error protection. On the
`other hand, the noise sensitivity of the 4-PSK data stream in the HM realisation is substantially lower than
`that of the 64-QAM data stream of non-hierarchical (and also hierarchical) modulation. This is because the
`affiliation of two-bit information to a quadrant is less likely to become disturbed. If any constellation state
`does get disturbed (i.e. mixed up with another arbitrary state in the same quadrant), the quadrant information
`is still correct.
`
`In addition to obtaining two independent data streams in hierarchical modulation, the data stream with the
`lower data rate is less susceptible to noise than it would be in a non-HM scheme. At the same time, the HM
`data stream with the higher data rate is not noticeably less robust than an equivalent data stream in a non-HM
`realisation. The total data rate of the two HM streams is identical to that of a non-HM scheme. The net data
`rate will however be slightly lower because twice the MPEG-TS overhead is incurred in the case of hierarchi-
`cal modulation, on account of the two multiplexes.
`
`EBU TECHNICAL REVIEW (cid:150) April 2003
`C. Weck and A. Schertz
`
`3 / 13
`
`Constellation Exhibit 2016
`
`

`

`DVB-T
`
`1.00E+00
`
`HP
`
`LP
`
`1.00E-01
`
`1.00E-02
`
`BER
`
`1.00E-03
`
`1.00E-04
`
`1.3. High- and Low-priority streams
`As outlined above, the data streams of hierarchical modulation vary in their susceptibility to noise. In other
`words, the service coverage areas differ in size. The better-protected data stream is referred to as the High-
`Priority (HP) stream; the other one is referred to as the Low-Priority (LP) stream. Compared with non-hier-
`archical modulation, the HM data stream with the lower data rate can be used to supply a larger coverage area,
`whereas the coverage area of the data stream with the higher data rate is only insignificantly smaller than for
`the corresponding non-HM variant. This subdivision alone can be of practical benefit.
`It is possible to enlarge the coverage area of the HP stream even further by changing the modulation parameter
`α at the expense of the robustness of the LP stream. This is the case for the constellation shown in Fig. 1.
`Here, the neighbouring dots which are located in adjoining quadrants have twice the spatial separation of those
`neighbours located within a single quadrant (modulation parameter α = 2), i.e. they can be less easily confused
`when reception is impaired by noise. In Fig 1, the first two bits indicate the quadrant. Consequently, these
`bits are more reliably transmitted than the other four. With α = 4, the protection of the two leading bits would
`be even better. On the other hand,
`the reliability of the differentiation
`between the states decreases in the
`course of decoding, on account of
`the reduction in the separation of
`the states within a quadrant. In
`each case, the enhanced protection
`for the first two bits is at the
`expense of the remaining four bits.
`When α = 4,
`the
`separation
`between the protection levels is
`correspondingly greater than for
`α = 2.
`The bits of the HP stream can addi-
`tionally be
`transmitted with a
`greater error protection than those
`of the LP stream, i.e. with a lower
`code rate (e.g. 1/2 or 2/3). This
`might be necessary, e.g. for mobile
`reception (see Section 2.3.). Using
`these error-protection techniques,
`the HP stream may require a con-
`siderably lower C/N (carrier-to-
`noise) ratio in order to achieve the
`same bit error ratio (BER) as the LP stream. This is shown in Fig. 3.
`Thus, three parameters are required to describe a hierarchical modulation scheme: α, and the code rates for the
`HP and LP streams.
`
`1.00E-05
`0.00
`
`2.00
`
`4.00
`
`6.00
`
`8.00
`
`10.00 12.00 14.00 16.00
`
`18.00
`
`C/N
`
`Figure 3
`System behaviour for low- and high-priority streams, in the case of
`hierarchical modulation
`(α = 2; high priority = 4-PSK with code rate 2/3; low priority = 16 QAM
`with code rate 3/4)
`
`1.4. The benefits of hierarchical modulation
`Originally in the development of the DVB specification, careful consideration was given to using the principle
`of hierarchical modulation, in order to deal with a fundamental problem of digital transmission (cid:150) the abrupt
`breakdown in reception below a critical field strength level. If, in the case of MPEG source encoding, a coarse
`and a fine resolution were to be divided up into two data streams (SNR scalability) then, with appropriate hier-
`archical modulation, the reception would not fail at a single stroke when the field strength diminished at the
`reception location. Instead, the resolution of the source encoding would initially decrease by one stage.
`This possibility was not actually used in the DVB-T standard because it would have increased the hardware
`costs of the source decoder in the receiver and would not have been compatible with standard MPEG decoder
`chips.
`
`EBU TECHNICAL REVIEW (cid:150) April 2003
`C. Weck and A. Schertz
`
`4 / 13
`
`Constellation Exhibit 2016
`
`

`

`DVB-T
`
`4-PSK
`FA, PA
`
`2 bit/carrier
`
`MPEG-TS
`MUX A
`
`COFDM
`
`Ch.A
`
`4-PSK
`Ch.A
`
`MPEG-TS
`MUX B
`
`COFDM
`
`Ch.B
`
`4 bit/carrier
`
`16-QAM
`Ch.B
`
`STB
`
`16-QAM
`FB, PB
`
`Figure 4
`4-PSK and 16-QAM separated over two TV channels using
`non-hierarchical modulation
`
`However, the transmission-side appli-
`cation of the HM principle is incorpo-
`rated in the DVB-T standard. Here,
`the variable robustness of constella-
`tions is used for the transmission of
`different programme multiplexes
`with unequal priority (high priority,
`low priority) over one frequency
`channel. Figs. 4 and 5 show the rele-
`vant alternatives which the DVB-T
`standard thus provides: 4-PSK and
`16-QAM via two separate TV chan-
`nels (in the case of non-hierarchical
`modulation) or 4-PSK in 64-QAM
`via one single channel (in the case of
`hierarchical modulation).
`As illustrated above, in the case of
`hierarchical modulation, the coverage
`areas for the two programme multi-
`plexes are of different sizes (cid:150) for
`α = 1 (uniform modulation) (cid:150) owing
`to the different susceptibility to noise
`of the two streams. In the LP case, the
`coverage is roughly as large as in the
`non-HM case, but is even larger in the
`HP case. The HP coverage area can
`be further enlarged at the expense of
`the LP area, by choosing appropriate
`values for α if this is desired.
`Of course, different-sized coverage
`areas are not always desirable for the
`programme multiplexes broadcast by
`a transmitter. But should this really
`be a fundamental objection against
`using hierarchical modulation? Keep
`in mind that even without hierarchi-
`cal modulation, different sized cover-
`age areas are inevitable because the
`radius depends on the power con-
`straints at any considered frequency.
`(Because of national and interna-
`tional frequency co-ordination, in
`most cases DVB-T multiplexes cannot be operated on different channels with the same transmission power).
`Coverage areas also depend on the selected method of modulation and on the code rate. It is therefore advis-
`able to verify, in each specific case, whether hierarchical modulation would be a better alternative under the
`existing constraints.
`
`4-PSK
`FA, PA
`
`2 bit/carrier
`
`MPEG-TS
`MUX HP
`
`COFDM
`
`MPEG-TS
`MUX LP
`
`COFDM
`
`Ch.A
`
`2 + 4 bit/carrier
`
`4-PSK in16-QAM
`FA, PA
`
`HP
`
`4-PSK
`Ch.A
`
`2 bit/carrier
`
`16-QAM
`Ch.A
`
`4 bit/carrier
`
`STB
`
`LP
`
`Figure 5
`4-PSK in 64-QAM over one TV channel using hierarchical
`modulation
`
`1.5. Definition of comparison criteria
`How can we gauge whether hierarchical or non-hierarchical modulation is the right choice for a specific appli-
`cation? With hierarchical modulation, two MPEG multiplexes with a differing data rate and differing robust-
`ness (i.e. coverage radius) can be broadcast on one TV frequency channel. A shortage of transmission
`channels is therefore an argument in favour of hierarchical modulation. Whether differently sized coverage
`areas are acceptable or even desirable depends on the individual case.
`
`EBU TECHNICAL REVIEW (cid:150) April 2003
`C. Weck and A. Schertz
`
`5 / 13
`
`Constellation Exhibit 2016
`
`

`

`DVB-T
`In the following paragraphs, the coverage areas for various different HM alternatives are compared with the
`most important variant of non-hierarchical modulation (16-QAM, code rate: 2/3) which will be used in Ger-
`many. For that purpose, universally usable benchmarks are essential when making comparisons.
`
`One benchmark used here pertains to the population:
`
`Data Supply Capability = Supplied Population * Data Rate [User * Mbit/s]
`
`In accordance with this criterion, the maximum traffic volume is attained when all accessible users retrieve all
`the transmitted data.
`
`The second benchmark is related to the coverage area:
`
`Data Area Supply Capability = Supplied Area * Data Rate [km† * Mbit/s]
`
`In accordance with this, the maximum traffic volume occurs when all transmitted data are retrieved by each
`unit of the covered area.
`
`In the case of hierarchical modulation, the contributions of the HP and LP data streams have to be added. The
`results of a comparison using these two benchmarks naturally depend on the geographical and demographic
`circumstances.
`
`2. Example cases of hierarchical modulation
`2.1. Service coverage by a single transmitter
`
`For this investigation, coverage provided by the Alexanderplatz transmitter in Berlin (on channel 27) was
`assessed over a square-shaped region of side length 120 km. A transmission power of 150 kW was assumed.
`
`Three different implementations of hierarchical modulation were compared with the non-HM scheme:
`16-QAM, code rate 2/3. The size of the coverage areas depends on the C/N ratio which has to be achieved in
`order not to exceed a certain bit error rate. The DVB-T specification assumes a bit error rate of 2∗10-4 in
`accordance with the Viterbi decoder 4. C/N ratios based on simulations can be found in tables in the DVB-T
`specification. In this context, a distinction must be drawn between the C/N ratio for indoor reception using a
`directional rooftop aerial (Rice channel) and portable indoor reception using a small omnidirectional tele-
`scopic or rod antenna (Rayleigh channel). Table 1 shows the theoretical values which are detailed in the DVB-
`T specification for the cases considered here.
`
`For planning in accordance with the Chester Conference 5 [4], implementation losses of 3 - 4 dB were
`assumed. Subsequent measurements performed by the IRT have demonstrated that these values are rather
`optimistic. However, it did not seem advisable to simply employ the IRT(cid:146)s measured values for the supply cal-
`culations. Rather, in the case of the reference, the planned value agreed at Chester was retained while, for the
`three HM cases considered, the difference between the measured values for the HM and the non-HM cases was
`added as shown in the following expression:
`C/Nhierarchical = C/Nnon-hierarchical (Chester) + {C/Nhierarchical (measured) (cid:150) C/Nnon-hierarchical (measured)}
`Accordingly, the differences between the measured values as stated in Table 2 were to be added to the planned
`values of the reference. Thus, the results of the analysis demonstrate what has changed in the HM case, rela-
`tive to the non-HM reference, and they retain their information value even if the reference value is changed for
`the non-hierarchical case.
`
`4. Block and convolution encoding is used in the case of DVB-T. In order to “de-convolute”, a Viterbi decoder
`is used in the receiver.
`In 1997, a European conference of postal and telecommunications administrations (CEPT) was held in Ches-
`ter, England, and it adopted a convention for the conversion of analogue television to DVB-T.
`
`5.
`
`EBU TECHNICAL REVIEW (cid:150) April 2003
`C. Weck and A. Schertz
`
`6 / 13
`
`Constellation Exhibit 2016
`
`

`

`Table 1
`C/N ratios necessary for reception in the analysed non-hierarchical and hierarchical modulation cases
`(theoretical values are in accordance with the DVB-T specification, without implementation losses)
`
`DVB-T
`
`Modulation
`
`Code
`rate
`
`Hierarchical
`(4-PSK in
`64-QAM)
`
`α
`
`Priority
`
`Reference
`Case 1
`
`Case 2
`
`Case 3
`
`16-QAM
`64-QAM
`64-QAM
`64-QAM
`64-QAM
`64-QAM
`64-QAM
`
`2/3
`2/3
`2/3
`2/3
`2/3
`1/2
`1/2
`
`no
`yes
`yes
`yes
`yes
`yes
`yes
`
`1
`1
`2
`2
`2
`2
`
`HP
`LP
`HP
`LP
`HP
`LP
`
`C/N (dB)
`reception
`using roof
`antenna
`11.6
`
`17.6
`
`19.5
`
`14.9
`
`C/N (dB)
`portable
`indoor
`reception
`14.2
`14.8
`19.4
`11.7
`21.7
`11.4
`16.4
`
`Table 2
`Differences between the C/N ratios required for reception and the corresponding values of the reference
`(non-hierarchical, 16-QAM, code rate: 2/3)
`
`Modulation
`(hierarchical)
`
`Code rate
`
`Case 1
`
`Case 2
`
`Case 3
`
`64-QAM
`64-QAM
`64-QAM
`64-QAM
`64-QAM
`64-QAM
`
`2/3
`2/3
`2/3
`2/3
`1/2
`1/2
`
`α
`
`1
`1
`2
`2
`2
`2
`
`Table 3
`Coverage of the area considered (location probability: 95 %)
`
`Reference: 16-QAM, R = 2/3
`D = 13.27 Mbit/s
`Case 1: 4-PSK in 64-QAM
`α = 1, RHP = 2/3, RLP = 2/3
`DHP = 6.64 Mbit/s
`DLP+HP = 19.91 Mbit/s
`
`Case 2: 4-PSK in 64-QAM
`α = 2, RHP = 2/3, RLP = 2/3
`DHP = 6.64 Mbit/s
`DLP+HP = 19.91 Mbit/s
`
`Case 3: 4-PSK in 64-QAM
`α = 2, RHP = 1/2, RLP = 1/2
`DHP = 4.98 Mbit/s
`DLP+HP = 14.93 Mbit/s
`
`EBU TECHNICAL REVIEW (cid:150) April 2003
`C. Weck and A. Schertz
`
`Portable indoor
`
`HP portable indoor
`LP portable indoor
`LP stationary
`HP portable indoor
`LP portable indoor
`LP stationary
`HP portable indoor
`LP portable indoor
`LP stationary
`
`Priority
`
`C/N (dB)
`reception
`using roof
`antenna
`
`HP
`LP
`HP
`LP
`HP
`LP
`
`+ 9
`
`+ 11
`
`+ 6
`
`Area
`51 %
`
`51 %
`29 %
`95 %
`61 %
`22 %
`94 %
`67 %
`37 %
`96 %
`
`C/N (dB)
`portable
`indoor
`reception
`+ 0
`+ 8
`- 3
`+ 11
`- 5
`+ 5
`
`Population
`90 %
`
`90 %
`84 %
`99 %
`92 %
`79 %
`99 %
`94 %
`86 %
`99 %
`
`7 / 13
`
`Constellation Exhibit 2016
`
`

`

`DVB-T
`The coverage areas are shown in Fig. 6, while Table 3 gives an overview of the service coverage achieved in
`each case.
`
`High-res GIF (850 KB)
`
`High-res GIF (850 KB)
`
`Reference:
`16-QAM
`R = 2/3
`13.3 Mbit/s
`
`Portable Indoor
`
`Stationary
`
`Case 1:
`4-PSK in 64-QAM
`α = 1
`RHP = 2/3
`DHP = 6.64 Mbit/s
`
`RLP = 2/3
`DLP+HP = 19.91 Mbit/s
`
`HP Portable Indoor
`
`LP Portable Indoor
`
`LP Stationary
`
`High-res GIF (850 KB)
`
`High-res GIF (850 KB)
`
`Case 2:
`4-PSK in 64-QAM
`α = 2
`RHP = 2/3
`DHP = 6.64 Mbit/s
`
`RLP = 2/3
`DLP+HP = 19.91 Mbit/s
`
`Case 3:
`4-PSK in 64-QAM
`α = 2
`RHP = 1/2
`DHP = 4.98 Mbit/s
`
`RLP = 1/2
`DLP+HP = 14.93 Mbit/s
`
`Figure 6
`Coverage by the Alexanderplatz transmitter: reference (16-QAM), hierarchical modulation and 64-QAM cases
`
`EBU TECHNICAL REVIEW (cid:150) April 2003
`C. Weck and A. Schertz
`
`8 / 13
`
`Constellation Exhibit 2016
`
`

`

`DVB-T
`
`HP
`
`LP
`
`HP
`
`LP
`
`Data Supply Capability in 1,000,000 N*Mbit/s
`
`0
`
`10
`
`20
`
`30
`
`40
`
`50
`
`60
`
`70
`
`80
`
`90
`
`Data Area Supply Capability in 1,000 km²*Mbit/s
`
`Berlin 16-QAM
`
`HM Fall 1
`
`HM Fall 2
`
`HM Fall 3
`
`64-QAM
`
`Berlin 16-QAM
`
`HM Fall 1
`
`HM Fall 2
`
`HM Fall 3
`
`64-QAM
`
`Case 2:
`
`Case 3:
`
`A brief summary, drawn from these three tables, is as follows:
`Case 1: For the HP multiplex, the portable indoor reception is unaltered because the C/N ratio has
`remained the same. By contrast, however, the covered area is significantly smaller for the LP
`multiplex because the necessary C/N ratio is 8 dB higher.
`If α is changed from 1 to 2, the difference between the HP and LP coverage areas increases. The
`HP area is now larger than the reference area (-3 dB) and the LP area is considerably smaller
`(+11 dB).
`If the code rate for HP and LP is also reduced, i.e. the error protection is increased, then both cov-
`erage areas increase in size. For HP, the effect is smaller than for LP (2 dB less in one case, 6 dB
`in the other) so that the coverage area sizes again draw closer together. Nonetheless, the data rate
`is lower than in the first two cases.
`In all three cases, the population having stationary reception is completely covered by the LP multiplex.
`On account of the differing data rates of the MPEG-2 multiplexes under consideration (reference, HP and LP),
`these findings on their own do not support a conclusion on which is the best alternative. However, from the
`coverage data in Table 3, it is now possible to deduce the coverage capability according to the criterion given
`in Section 1.5. The total area is 14,400 km†, the modulation is hierarchical (4-PSK in 64-QAM), coverage is
`provided by the Alexanderplatz transmitter for both non-hierarchical and hierarchical modulation, and the
`total population covered is 4,774,402.
`That yields (cid:150) for the reference case:
`Data Supply Capability
`= Supplied population ∗ data rate [users ∗ Mbit/s]
`= 0.90 ∗ 4,774,402 ∗ 13.27 [users ∗ Mbit/s]
`= 57,020,683 [users ∗ Mbit/s]
`Fig. 7 provides an overview of the findings.
`It can be seen that all the cases of hierarchical
`modulation yield higher overall supply capa-
`bility than the reference. The largest increase
`in the data supply capability with respect to
`the population is 43 % in Case 1. For com-
`parison purposes, the supply capability was
`also calculated in the case of non-hierarchical
`modulation for 4-PSK and 64-QAM with a
`code rate of 2/3 (Fig. 8 shows the coverage
`areas). For this purpose, as in the case of the
`calculations for hierarchical modulation, the
`measured deviations between the C/N values
`required for reception and the theoretical val-
`ues were considered.
`With 4-PSK, a considerably lower supply
`capability is obtained than with the reference
`16-QAM. With 64-QAM, the Data Area
`Supply Capability is roughly 6 % lower than
`in the case of the reference. However, the
`Data Supply Capability with respect to the
`population is 42 % greater, thus nearly as
`high as in Case 1 of hierarchical modulation.
`
`0
`
`10
`
`20
`
`30
`
`40
`
`50
`
`60
`
`70
`
`80
`
`90
`
`100
`
`110
`
`Figure 7
`Supply capability for the reference (16-QAM), hierarchical
`modulation and 64-QAM schemes
`
`2.2. Single-frequency network (SFN) having strong self interference
`As an example of an SFN, the coverage area provided by 18 transmitters of the national TV station ZDF in
`northern Germany, using channel 40, was considered. According to the Chester Agreement of 1997, the
`
`EBU TECHNICAL REVIEW (cid:150) April 2003
`C. Weck and A. Schertz
`
`9 / 13
`
`Constellation Exhibit 2016
`
`

`

`DVB-T
`power of each transmitter was reduced by
`7 dB. Again the reference was 16-QAM
`modulation with a code rate of 2/3.
`Case 3 as described in Section 2.1. was
`assumed to be the hierarchical modulation
`case (4-PSK in 64-QAM, with α = 2,
`RHP = 1/2 and RLP = 1/2). Fig. 9 shows
`the coverage provided by the reference
`system. The red frame contains large
`areas that are not covered because of self-
`interference problems (the corresponding
`coverage provided by an MFN network
`demonstrates this).
`Table 4 shows the calculated coverage
`probabilities, within the red frame only.
`Fig. 10 indicates the data supply capabili-
`ties deduced from these. The overall area
`of the test rectangle is 8,058 km† and it
`contains a population of 1,863,838.
`In this case too, the data supply capabili-
`ties compared with the non-hierarchical
`reference are improved on account of the
`hierarchical modulation.
`
`2.3. Mobile reception of the
`HP stream
`In a field trial performed in Munich in
`1999, the IRT tested mobile reception of
`the HP multiplex in the case of hierarchi-
`cal modulation using 4-PSK in 64-QAM,
`α = 1, RHP = 1/2 and RLP = 2/3. This
`mode is a variation of Case 1 from
`Section 2.1. with more robust transmission
`of the HP multiplex (code rate 1/2 instead
`of 2/3). With three different first-genera-
`tion receivers (BBC, ITIS and NTL /
`DMV), mobile reception was assessed for
`the 8k-FFT case which, at that time, was
`regarded as extremely critical for mobile
`reception. Fig. 11 shows the minimum
`field strengths required for reception,
`according to this analysis.
`The overall representation is for outdoor
`reception. A, B and C denote the three
`receivers tested. Horizontal lines indi-
`cate the minimum field strengths in the
`case of non-mobile reception for LP and
`HP. As a result of the greater interfer-
`ence sensitivity,
`the minimum field
`strengths for
`the LP multiplex are,
`depending on the receiver, up to 14 dB
`above those for the HP multiplex. The
`
`10 / 13
`
`QPSK
`
`High-res GIF (650 KB)
`
`16-QAM
`
`High-res GIF (700 KB)
`
`Portable – Indoor coverage, 95%
`
`64-QAM
`
`High-res GIF (700 KB)
`
`Figure 8
`Variations of non-hierarchical modulation with code rate 2/3
`
`High-res GIF (700 KB)
`
`95% of all locations
`
`70% of all locations
`
`Figure 9
`Portable indoor coverage by an SFN of 18 transmitters with 16-QAM
`modulation, code rate 2/3.
`Noise interference for 5% of the time.
`
`EBU TECHNICAL REVIEW (cid:150) April 2003
`C. Weck and A. Schertz
`
`Constellation Exhibit 2016
`
`

`

` Table 4
`Service coverage of the area considered (location probability: 95 %)
`
`Reference
`
`Case 3
`
`16-QAM
`R = 2/3
`D = 13,27 Mbit/s
`4-PSK in 64-QAM
`α = 2,
`RHP = 1/2, RLP = 1/2
`DHP = 4.98 Mbit/s
`DLP+HP = 14.93 Mbit/s
`
`Portable Indoor
`
`HP Portable Indoor
`(95 % of the locations)
`
`LP Portable Indoor
`(95 % of the locations)
`
`DVB-T
`
`Population
`39 %
`
`57 %
`
`29 %
`
`Area
`17 %
`
`42 %
`
`7 %
`
`interesting possibility for mobile recep-
`tion is the HP multiplex. The chart indi-
`cates for the three receivers how the
`minimum field strength for HP reception
`increases with increasing speed. In the
`case of high speeds (100 - 150 km/h), it
`reaches the value which is necessary for
`stationary reception of the LP multi-
`plex. In other words: at high speeds, the
`coverage area for mobile reception of HP
`is roughly the same as the coverage area
`for portable outdoor reception of LP. In
`this context, it is essential to know that
`the receivers used were not optimized for
`mobile reception and
`the maximum
`speed for the 16-QAM reference, accord-
`
`Data Supply Potential in 1,000,000 N*Mbit/s
`
`0
`
`2
`
`4
`
`6
`
`8
`
`10
`
`12
`
`Data Supply Catchment Area Potential in 1,000 km²*Mbit/s
`
`HP
`
`LP
`
`HP
`
`LP
`
`SFN 16-QAM
`
`SFN HM Fall 3
`
`SFN 16-QAM
`
`SFN HM Fall 3
`
`0
`
`5
`
`10
`
`15
`
`20
`
`25
`
`Figure 10
`(Upper) Data Supply Capability and (lower) Data Area Supply
`Capability (cid:150) for (cid:147)reference(cid:148) and (cid:147)higher modulation (Case 3)(cid:148)
`
`A: LP stationary
`
`B: LP stationary
`
`C: LP stationary
`
`A: HP mobile
`
`B: HP mobile
`
`C: HP mobile
`
`A: HP stationary
`
`B: HP stationary
`
`C: HP stationary
`
`0
`
`50
`
`100
`
`150
`
`Speed km/h
`
`-60
`
`-65
`
`-70
`
`-75
`
`-80
`
`-85
`
`-90
`
`Input power dBm
`
`Figure 11
`Minimum field strength depending on driving velocity in the case of mobile reception of the HP multiplex of
`hierarchical modulation
`
`EBU TECHNICAL REVIEW (cid:150) April 2003
`C. Weck and A. Schertz
`
`11 / 13
`
`Constellation Exhibit 2016
`
`

`

`DVB-T
`
`4-PSK 4-state Phase Shift Keying (see QPSK)
`BER
`Bit-Error Ratio
`C/N
`Carrier-to-Noise ratio
`COFDM Coded Orthogonal Frequency Division
`Multiplex
`D/A
`Digital-to-Analogue
`DVB
`Digital Video Broadcasting
`DVB-T DVB - Terrestrial
`HM
`(DVB-T) Hierarchical Modulation
`HP
`High-Priority
`IEC
`International Electrotechnical Commission
`
`Abbreviations
`ISO
`International Organization for Standardization
`LP
`Low-Priority
`MPEG (ISO/IEC) Moving Picture Experts Group
`MUX
`Multiplex / multiplexer
`OFDM Orthogonal Frequency Division Multiplex
`QAM
`Quadrature Amplitude Modulation
`QPSK Quadrature (Quaternary) Phase-Shift Keying
`SFN
`Single-Frequency Network
`SNR
`Signal-to-Noise Ratio
`STB
`Set-Top Box
`TS
`(MPEG) Transport Stream
`
`ing to measurements made by the IRT at that time, was only approx. 55 km/h in the 8k mode (broadcasting
`via channel 43).
`
`3. Applications of hierarchical modulation
`Using specific examples, it has been demonstrated here that hierarchical modulation enables greater overall
`data supply capability than non-hierarchical modulation. However, HM should only be used if it fulfils the
`broadcasters(cid:146) existing service requirements. For example, using the HP multiplex to broadcast major pro-
`grammes (such as the German national TV programmes ARD, ZDF ...) and the LP multiplex to broadcast sup-
`plementary programmes (such as Eins MuXx, ZDF INFObox, ARD online) is quite conceivable. The HP
`programmes can be received on portable and mobile equipment all over the rural surrounding areas. The LP
`transmissions are suitable for stationary reception in rural areas and portable reception in the higher populated
`areas or in the proximity of the transmitter, respectively. The higher data rate of the LP stream can also be
`used to accommodate more programmes in the multiplex or, in the future, to broadcast HDTV programmes for
`displaying on high-resolution screens.
`
`4. Conclusions
`The conclusion is that, in certain realistic cases, hierarchical modulation can represent a better alternative than
`non-HM. It provides two autonomous multiplexes (HP and LP) over a single TV frequency channel. The total
`data rate (HP + LP) of 4-PSK in 64-QAM is hi