MXPA96000925A - Coded modulation with gain of conformac - Google Patents

Abstract

The present invention relates to the apparatus comprising: means for receiving an information stream, coding means for plotting information at signaling points of a predetermined base constellation with unequal probabilities, means for processing the signaling points plotted through a Tomlinson precoder, Tomlinson encoder and precoder means are presented in such a way that the signaling points provided in the Tomlinson precoder output shows at least 1.0 dB of configuration, and means to apply a modulated signal to a transmission channel which represents the signaling points provided in the Tomlins precoder output

Description

CODIFIED MODULATION WITH GAIN OF CONFORMATION BACKGROUND OF THE INVENTION The present invention relates to the transmission of digital data and more particularly, to high-definition television. A problem occurs in the context of the transmission of high definition television (HDTV) signals in channels that are assigned to the standard transmission, the so-called NTSC TV broadcast. The problem occurs specifically in areas where channels that are unused for NTSC transmission - and thus are candidates for HDTV transmission - are used for NTSC transmissions in relatively close areas. As an example, US TV channel 3 is currently unused in the New York City metropolitan area, but is used in both the city of Philadelphia and the city of Hartford. The consequence of such proximity is that if channel 3 is going to be used by HDTV in New York City, the HDTV signal can be corrupted by the NTSC signal from channel 3 of Philadelphia or Hartford for the HDTV observers of the City of New York. On the contrary, the NTSC signal can be corrupted by the HDTV signal. Such corruption is referred to as "co-channel interference".

The co-channel interference of the HDTV signal to the NTSC signal can be satisfactorily addressed by the specification in the HDTV standards (still under consideration) of a sufficiently low transmission power level. In addition, to address the co-channel interference of the NTSC signal to the HDTV signal, it has been proposed to rely on the fact that an NTSC signal is dominated by energy concentrations at particular locations in the frequency spectrum. Thus, it has been proposed to provide an HDTV receiver with a filter, hereinafter referred to as an NTSC blocking filter, having zero or zero values at those frequency locations, thereby eliminating a significant portion of the interference NTSC signal. , while only minimally degrading the HDTV signal. The proposed HDTV standards contemplate a digital transmission format in which the current bit representing the baseband HDTV signal is mapped, by means of a so-called channel code, into channel symbols consisting of each of one or more than the so-called signal points. A potential problem in such arrangements is that the NTSC blocking filter creates so-called forced intersymbol interference (ISI) in the receiver, which corrupts the received signal points.

BRIEF DESCRIPTION OF THE INVENTION It is known in the prior art to deal with the forced ISI by using the so-called Tomlinson precoder in the transmitter, which uses a feedback signal to operate on the sequence of the signal points specified by the channel code. However, the belief that the Tomlinson precoder will invariably remove any of the so-called "conformation gain" at the signal points generated by the channel code is commonly maintained. Such conformation gain is achieved at the expense of bandwidth efficiency (measured in bits per channel symbol). The prior art thus describes separating from the notion of using a channel code having conformation gain together with a Tomlinson precoder. After all, there is no reason to sacrifice the bandwidth efficiency for the conformation gain, only to have the conformation gain removed by the Tomlinson precoder. In accordance with the present invention, however, it has been recognized that the belief mentioned in the above, commonly held is erroneous. Specifically, it has been found that if the feedback signal of the mentioned Tomlinson precoder is relatively weak compared to the input signal of the Tomlinson precoder, the conformation gain (if any) of the channel code will be substantially retained in the Tomlinson precoder output. . Thus, the general transmission scheme encompassing the principles of the present invention, comprises a channel encoder which provides a more than minimal amount of conformation gain together with a Tómlinson precoder. The invention, more particularly, contemplates the use of a channel code, which provides such a high degree of shaping gain of its signal points that the shaping gain of the signal points that are output by the Tomlinson precoder is at least 1.0 dB, when compared to the signal points generated by a channel encoder without conformation gain and with the use of the Tomlinson precoder. (Even under the restriction mentioned above, vis a vis, the feedback signal from the Tomlinson precoder will still be a small loss of the conformation gain due to the action of the Tomlinson precoder). A conformation gain of 1.0 dB is significant and is achieved only by means of large exchange in the bandwidth efficiency that, absent the teachings of the present invention, the prior art - due to the belief mentioned in the above, commonly maintained it does not allow to use, in a system which includes a Tomlinson precoder, a channel code having such a large amount of conformation gain.

BRIEF DESCRIPTION OF THE DRAWINGS FIGURE 1 is a block diagram of an HDTV transmitter embodying the principles of the present invention; FIGURE 2 is a graph of help in describing the trellis code that can be used as part of the general channel code in the HDTV system formed of the transmitter of FIGURE 1 and the receiver of FIGURE 3; FIGURE 3 is a block diagram of an HDTV receiver adapted to receive the HDTV signals generated by the transmitter of FIGURE 1; FIGS. 4-7 describe a first illustrative trellis code; FIGURES 8-10 describe a second illustrative trellis code; and FIGURE 11 is a table comparing the characteristics of the first and second trellis codes mentioned above and a third trellis code that can be used by the HDTV system.

DETAILED DESCRIPTION OF THE INVENTION The following detailed description presents, in the order mentioned, the descriptions of an HDTV system of the transmitter and receiver of FIGURES 1 and 3 ("System Generalities"); the trellis codes that can be used in the HDTV system ("Trellis codes"); The characteristics of those codes ("Characteristics of the Codes"), as the gain of conformation of those codes is conserved according to the principles of the present invention ("Preservation of the Gain of Conformation"), and as the characteristics of gain of conformation of the various codes can be used as an advantage ("Diversity of Conformation Gain").

SYSTEM OVERVIEW FIGURE 1 shows a television transmitter encompassing the principles of the invention. A television signal - illustratively an HDTV signal - is provided by the signal source 11. The signal source 11 includes the circuitry for compressing the TV signal and placing it in an HDTV format, as well as some standard modem circuitry such as circuitry which randomizes the bit stream - the so-called "encoder". The HDTV signal generated in this form is applied to the concatenated encoder 13, which includes the serial combination of the Reed-Solomon encoder 131; the RS symbol interleaver 134, which reorders the sequence of the Reed-Solomon symbols generated by the encoder 131 to provide protection against so-called "burst" errors introduced either in the television channel or in the receiver; the N-dimensional trellis 136 encoder; and the N-dimensional constellation mapper 139. Illustratively, the Reed-Solomon code implemented by the encoder 131 is the so-called RS code (208, 188) in a finite field GF (256) and also illustratively, N = 4. The combination of the trellis encoder 136 and the constellation mapper 139 they implement a type of channel code referred to as "coded modulation". In particular, the output of the trellis encoder 136 is a sequence of data words that identify a sequence of four-dimensional symbols. Each of the four-dimensional symbols will be transmitted in the form of a sequence of four signal points of one dimension. For this purpose, the data words identifying the four-dimensional symbols are applied within the encoder 13 concatenated to the four-dimensional constellation mapper 139, whose output for each identified symbol is a sequence of four signal points of a dimension. Since the signal points are of one dimension, each signal point is simply represented as a signed number. There are M possible values of the signal point. FIGURE 2 graphically shows the relationship between the signal points of a dimension mentioned in the above and the four-dimensional symbols. Each four-dimensional symbol generated by the constellation mapper 139 is formed of a sequence of four signal points. Each signal point is a point in a base constellation of one dimension, predetermined. The constellation of a dimension illustratively, has six signal points located at coordinates -5, -3, -1, 1, 3 and 5. The four-dimensional symbol is provided during the so-called symbol interval or equivalently, four of the so-called signaling intervals , one signal point in each signaling interval. The assembly or assembly of all four-dimensional symbols that can be output by the constellation mapper 139 is referred to as the four-dimensional constellation. A particular symbol of four dimensions - (3.1, -1.5) - is represented in FIGURE 2 by means of an enlarged signal point of each of the constellations of a dimension.

Returning to FIGURE 1, the signal point values generated by the constellation mapper 139 are applied to the signal point interleaver 15 of the type described in U.S. Patent No. 5,056,112 issued October 8, 1991, which reorders the sequence of the values of the signal point. The combination of this interleaver with a corresponding deinterleaver in the receiver, advantageously causes the 'noise in the received signal to be bleached before being decoded in Viterbi in the receiver as mentioned in the following. The interleaved signal points are applied to the Tomlinson precoder 17, which compensates in advance for the so-called forced intersymbol interference (ISI) which is input to the receiver by the NTSC blocking filter of the receiver. The output of the Tomlinson precoder is a sequence of values which no longer take the finite number of values -5, -3, -1, 1, 3 and 5, but rather, a set of continuous values within a limited range between -6 and +6. These values are applied to the vestigial sideband modulator or VSB. The signal VSB generated by the modulator 19 is applied to a television channel, such as a channel in the air or a cable and is received by the receiver of FIGURE 3.

Within the receiver, the received signal VSB is applied to the demodulator 39 VSB. Its output is processed by the NTSC reject filter 37, which has notches corresponding to those regions of the NTSC television signal spectrum, where the signal energy is concentrated. In this way, this filter advantageously eliminates a main source of interference in the HDTV signal when the HDTV transmitter of FIGURE 1 is in relatively close proximity to the broadcast of the NTSC transmitter in the same television channel. The resulting signal is equalized by the equalizer 36 to compensate for the intersymbol interference introduced by the channel. The output of the equalizer 36 is the best equalizer approximation of the values of the sequence of the interleaved signal points that was generated by the interleaver 15 of signal points. The output of the equalizer is deinterleaved first in the signal point deinterleaver 35 and then applied to the concatenated decoder 33, which includes the 336 Viterbi decoder; the RS symbol deinterleaver 334, which reorders the Reed-Solomon symbols; and the Reed-Solomon 331 decoder, which provides a retrieved version of the HDTV signal. The latter is decompressed and deformed within the television apparatus 31 and displayed therein. (Some of the values that are output by the equalizer 36 may be displaced by a constant value of "12" or "-12" as a result of the action of the Tomlinson precoder 17. Such a shift, however, is compensated by the decoder Viterbi). (Although not explicitly shown or described here, it should be mentioned as it is well known to those skilled in the art, that the so-called synchronization signal points are periodically inserted by the transmitter into the signal point stream that is received from the transmitter. signal point interleaver 15. The receiver recognizes these synchronization signal points and in response, generates a synchronization control signal which is used, in conventional manner, by various components of the receiver (e.g., deinterleaver 334) for synchronize their operations with those of the corresponding components in the transmitter (e.g. interleaver 134) The internal structures of the various transmitter and receiver components of FIGURES 1 and 3 are generally similar to those known to people with skill in the art and do not need to be described in more detail in the present.

Trellis codes A first coded modulation scheme that can be implemented by the combination of the trellis encoder 136 and the constellation mapper 139 is shown in FIGS. 4-7. This scheme encodes 8 bits per four-dimensional symbol of a four-dimensional constellation predetermined. FIGURE 4 shows the structure of the 136 trellis encoder. As shown in FIGURE 1, the trellis encoder 136 receives the stream of serial bits supplied by the RS symbol interleaver 134 and collects them in groups of eight parallel bits (for example, by a serial to parallel converter, not shown), represented by Yln to Y8n , where "n" represents the current signaling interval. The bits Yln and Y2n are applied to a 2/3 speed convolutional encoder 41, which provides three output bits. One of them, represented by Y0n, is generated by delay elements and exclusive OR circuits, which comprise the convolutional encoder. The encoder 41 is the so-called "systematic" convolutional encoder, in such a way that its other two output bits are simply its two input bits Yln and Y2n.

The operations of the convolutional encoder 41 can be described explicitly as follows. In each symbol interval, designated by the index "n" for its first signaling interval, the encoder makes a transition from its current state ln 2n W3n to a next state Wln + 4 W2n + 4 W3n + 4 and outputs three bits Y2n, Yln and Y0n, where Wln, W2n and 3n are the bits stored in the delay elements at the beginning of the symbol interval and wln + 4 'W2n + 4 v W3n + 4 are - * - os kits stored in the elements delay at the end of the symbol interval, and YOn - W3n W2n + 4 - W ln T 2n T Y ln W3n +4 s W2n? Y2n Each of the eight possible, different bit patterns represented by the three output bits of the convolutional encoder 41 identifies a respective subset of the symbols of the four-dimensional constellation. The remaining six bits, called "unencrypted", Y3 to Y8n also select a particular symbol of the four-dimensional subset, identified. In particular, the symbols of the four-dimensional constellation are divided into the eight subsets mentioned in the above, based on a division of their base constellations of a constituent dimension. FIGURE 7 shows how the constellation of six points of a dimension is divided into two subsets A and B, each subset having three signal points of a dimension. The four-dimensional constellation is then divided into eight subsets of four dimensions, 0, 1, ..., and 7. As shown in FIGURE 5, each subset of four dimensions consists of two sequences of four subsets of a dimension. For example, subset 2 of four dimensions consists of two sequences of the subset of one dimension (A, A, B, B) and (B, B, A, A), which means that if a four-dimensional symbol is a member of subset 2, then any of one of two criteria are met. With reference again to FIGURE 2, one possibility is that its first signal point constituent of a dimension is taken from a subset A of the first constituent of the one-dimensional constellation; its second constituent, the signal point of a dimension is taken from the subset A of the second constituent of the constellation of a dimension; this third constituent of the signal point of a dimension is taken from the subset B of the third constituent, the constellation of a dimension; and its fourth constituent a signal point of a dimension is taken from the subset B of the fourth constituent, the constellation of a dimension. The other possibility is that its first constituent, the signal point of a dimension, is taken from the subset B of the first constituent, the constellation of a dimension; its second constituent, the signal point of a dimension is taken from the subset B of the second constituent, the constellation of a dimension; its third constituent, the signal point of a dimension is taken from the subset A of the third constituent, the constellation of a dimension; and its fourth constituent, the signal point of a dimension is taken from the subset A of the fourth constituent, the constellation of a dimension. The selection of a particular symbol of the identified four-dimensional subset proceeds as follows: The three bits Y2n, Yln and Y0n of the convolutional encoder 41, together with an unencrypted input bit Y3n, are first converted to a sequence selector 42 of the subset of one dimension to four other bits Z0n, Z0n + 1, Z0n + 2 and Z0n + 3 • FIGURE 5 shows the detail of this conversion. The effect of this operation is to select one of the two possible sequences of the subset of a dimension of the four-dimensional subset identified by the bits Y0n to Y2n - for example (A, A, B, B) or (B, B, A, A) of the subset 2. The bits Y4n to Y8n are then used to select a symbol of the sequence of the subset of a dimension, identified. There are really 81 possible symbols in each sequence of a subset of a dimension, as can be seen from the fact that each subset of a dimension has three signal points and 34 = 81. However, since 'the five bits Y4n to Y8n can represent only 32 different bit patterns, not all 81 symbols really will be used On the contrary, it is advantageous for the search table of FIGURE 6 to map the 32 input bit patterns into 32 smaller energy symbols (the energy of a symbol is simply given by the sum of the squares of the coordinates of its constituent, the signal points of a dimension). For this purpose, the bits Y4n to Y8n are applied to the bit converter 45 which implements the look-up table shown in FIGURE 6. The first pair of output bits of the converter 45 - represented Z2n and Zln - select a signal point of the first subset of the sequence of the subset of a dimension, which is identified by the bit Z0n. The second pair of output bits of the converter 45 - represented Z2n + 1 and Zln + 1 - select a signal point of the second subset of the sequence of the subset of a dimension, which is identified by the bit Z0n + 1, etc. FIGURE 7 shows the mapping by which the bit values of Z2m, Zlm and Z0m for m = n, n + 1, n + 2 and n + 3 identify a signal point of a particular dimension. A second coded modulation scheme that can be implemented by the combination of the trellis encoder 136 and the constellation mapper 139, is shown in FIGS.

FIGURES 8-10. This scheme encodes an average of 8.5 bits per four-dimensional symbol. The modulation scheme of FIGS. 8-10 is constructed in the same underlying convolutional code as the first coding modulation scheme. As such, the encoder 81 and the selector 82 shown in FIGURE 8 are identical to the encoder 41 and the selector 42 shown in FIGURE 4, so again, three input bits are used to identify a sequence of four subsets of one dimension for each symbol interval. The remaining bits 5.5 without coding per symbol are used to select a symbol of the identified subset sequence. Obviously, it is not possible to operate in a semibit. However, this second coding modulation scheme uses the invention set forth in U.S. Patent No. 4,941,154 issued July 10, 1990 to perform the non-integral bit rate or the so-called "fractional" bit rate. In particular, the encoder 136 Trellis collects from the RS symbol interleaver 134, 17 bits in two successive symbol intervals. Two groups of three, that is six, of those bits are used to identify the sequence of the subset of a dimension for the respective two successive symbol intervals. The remaining eleven uncoded bits are used to jointly select a pair of four-dimensional symbols from the two identified one-dimensional stib-set sequences. The way in which this is done is shown in the FIGURE 8. Of the 11 uncoded bits, three are applied to a fractional bit encoder 83, implemented as the lookup table of FIGURE 9. As shown in FIGURE 9, the output of the encoder 83 is in the form of two pairs of bits, each of which can be taken in one of three possible bit patterns - 00, 01 and 10. For each symbol interval one of the two pairs of the output bits of the encoder 83 - represented Y9n and Y8n - is combined with four of the remaining eight uncoded bits to provide a six-bit input to the bit converter 85. The latter, similar to the first scheme, is used to select a four-dimensional symbol of the first sequence of the subset of a dimension, identified by the search table shown in FIGURE 10. Then, the second of the two pairs of bits of encoder 83 output - which will be the bits Y9n + 4 and Y8n + 4 - will be combined with four remaining uncoded bits to provide a second six-bit input to the bit converter 85, such that a second four-dimensional symbol will be selected from the second sequence of subsets of a dimension, identified. Note that because the bits Y9n and Y8n do not take the value 11, the total number of different bit patterns represented by the bits Y4n to Y9n is 48. The search table in FIGURE 10, similar to the search table of FIGURE 6 maps the 48 input bit patterns into the 48 lowest energy symbols. (It can also be observed that because the 00, 01 and 10 patterns of the Y9n and Y8n bits do not appear with equal probability - as can be directly verified from FIGURE 9 - the 48 different bit patterns of the Y9n bits Y4n also do not appear with equal probability, so the table in FIGURE 10 is advantageously constructed in such a way that the first 32 patterns - each of which occurs with an equal probability of 3/128 - map to the 32 symbols of lowest energy The remaining 16 patterns - each of which occurs with an equal probability of 2/128 - map to the remaining 16 symbols). Any number of other coding modulation schemes, based on the same convolutional code and the same base constellation of 6 signal points, can be implemented for example, by varying the number of bits that are collected per symbol interval and accommodating the diversity of numbers of uncoded bits by means of different mapping strategies following the basic concepts illustrated in the above. For example, an average bit rate of 8.25 bits per symbol, can be accommodated to perform coding in four four-dimensional symbol intervals - that is 33 bits per four four-dimensional symbol intervals - mapping nine of the 21 uncoded bits in four groups of three bits which, for each symbol interval, are combined with three of the remaining twelve unencoded bits to address a bit converter similar to the bit converter 85. In such a scheme, there will be a total of 40 different four-dimensional symbols in each subset sequence of a dimension.pair.

Characteristics of the Codes FIGURE 11 presents various properties and characteristics of the two coding modulation schemes explicitly described in the foregoing, as well as a third 9-bit scheme per symbol interval based on the same convolutional code and base constellation, than the code that is described in U.S. Patent Application Serial No. 08 / 226,606 filed April 12, 1994, co-pending, incorporated herein by reference. For each of the modulation schemes - represented by convenience as I, II and III - the signaling speed is the same, namely 10.76 x 106 signal points per second or 10.76 Mbaud. The trellis code inputs the bit rates of 8, 8.5 and 9 bits per symbol interval corresponding to the so-called "loads" - that is, the bit rates at the input to the concatenated encoder 13 - of 19.5, 20.7 and 21.9 Mbps, given the use of the Reed-Solomon RS code (208, 188) mentioned in the above. It should be noted that the signal points of a base constellation of a dimension at the output of the constellation mapper 139 do not appear with equal probability. Thus, as shown in FIGURE 11 for scheme I, for example, the signal points at coordinates -5, -3, -1, 1, 3, and 5 appear with the probabilities of 0.06, 0.17, 0.27, 0.27. , 0.17 and 0.06, respectively. A similar effect is observed for the other two schemes as well. Note that for each scheme, the probability of use decreases with an increase in the energy of the signal point, that is, a square coordinate value. A consequence of this fact is that for a given distance between the signal points of the base constellation, the average energy of the selected signal points is less than it would be if they were used with equal probabilities. Equivalently, for a given average energy that is assigned by a icular transmission environment, the distance between the signal points of the base constellation can be increased, thus providing increased immunity to noise. (Indeed, the conventional circuit set in 'the VSB modulator automatically adjusts the actual average transmission power to correspond to that in which the system wishes to operate). Such a coding modulation scheme is to exhibit the "conformation gain", the term "conformation" is used to reflect the fact that the probability distribution of the signal point is "conformed" instead of being a straight line. Quantitatively, the shaping gain of a given scheme at the output of the constellation mapper 139 is a function of an X / Y ratio - typically expressed in dB - where X is the average signal point energy of the base constellation, which would result if the signal points appear with equal probability and Y is the energy of the average signal point in the output of the constellation output 139 for the scheme under consideration. Thus, for example, by measuring the conformation gain at the output of the constellation mapper 139, it will be noted that the value of X for each of the three schemes shown in FIGURE 11 is 11.67. (= [12+ (-1) 2 + 32 + (-3) 2 + 52 + (-5) 2] / 6). The value of Y for example, for scheme I is 6.75 (= 0.27xl2 + 0.27x (-l) 2 + 0.17x32 + 0.17x (-3) 2 + 0.06x52 + 0.06x (-5) 2). The conformation gain of this form is 11.67 / 6.75 = 1.73, which is equivalent to 2.38 dB as shown in FIGURE 11. The conformation gains for schemes II and III are 1.67 dB and 0.67 dB, respectively. In this way, it is seen that, through the use of different schemes of these coding modulation schemes, increased levels of noise immunity in the exchange for a decreased load can be achieved. And advantageously, it will be noted that varying the shaping gain amounts are all achieved by using a icular constellation, ie a base constellation of one dimension, of 6 signal points which is, advantageously, the smallest constellation that can be used for any of the schemes I, II and III to support the input bit rates of the trellis encoder. In addition, the actual amount of conformation gain achieved, for example, by schemes I and II is really very significant in terms of the operation of the real-time system.

Preservation of Conformation Gain First referring again to FIGURE 1 for a brief description of the (well-known) structure of the Tomlinson precoder. In particular, the input signal of the Tomlinson precoder is received on the cable 151 of the interleaver 15 of the signal point. That signal is added by the combiner 175 to a feedback signal provided on the cable 174. The resulting signal is processed by the module device 171. The function of the device of the module is to ensure that the output of the Tomlinson precoder as a whole in the cable 177 is kept within a predetermined range of values - in this case in the range of between -6 and +6. Conceptually this is done in a very simple way, performing the function -12 of conventional module. That is, add and subtract repetitively the value "12" from the output of the combiner until the result falls within that range. The feedback signal mentioned above on the cable 174 is a function of the Tomlinson precoder outputs passed on the cable 177. More particularly, it is provided by the filter 173, which is illustratively a finite impulse response filter (FIR). ), which have a transfer function Z given by? = l where Z "1 represents a delay element with a delay amount of i signaling intervals In an illustrative implementation of the Tomlinson precoder, k = 36 and h- to h3g have values in the range between -0.0865 and 0.0603. In the foregoing, the Tomlinson precoder 17 is provided in the transmitter of FIGURE 1, to compensate for the forced intersymbol interference (ISI) that is created by the NTSC filter 37 in the receiver, although the use of a Tomlinson precoder will eliminate the ISI. Forced, the belief that the Tomlinson precoder will eliminate any "conformational gain" that should have been provided by the preceding coder is commonly held This belief is a consequence of the fact that in the prior art understanding of the Tomlinson precoder operation , it is assumed that each value of the allowed precoder output will appear with the same probability as any other allowed value. In this example, then, it would generally be believed that this is the case in which the values of the signal at the output of the Tomlinson precoder 17 on the cable 177 would be uniformly distributed across the range of -6 to +6. Another consequence of this assumption is that a design of a coded modulation scheme to be used in conjunction with a Tomlinson precoder would have the specific design objective of using each signal point with as little conformation gain as possible (ie, with the signal point probabilities that are as equal as possible) because the conformation gain is achieved at the expense of the bandwidth efficiency and there is no reason to sacrifice the bandwidth efficiency for the conformation gain , only to have the conformation gain removed by the Tomlinson precoder. In accordance with the present invention, it has been recognized that the aforementioned common understanding is erroneous. Specifically, it has been recognized that if the average power of the feedback signal on the cable 174 is relatively weak compared to the average energy of the input signal of the Tomlinson precoder on the cable 151, such as approximately 12 dB weaker, it is provided enough conformation gain (if any) for the channel code that will be preserved in the Tomlinson precoder output. In effect, the illustrative values mentioned above for the coefficients h will provide such a relatively weak feedback signal. In this way a general transmission scheme exemplifying the principles of the present invention comprises a channel encoder which provides a more than minimal amount of conformation gain together with a Tomlinson precoder. The conformation gain of a given scheme at the output of the Tomlinson precoder 17 is similarly defined to that of the constellation mapper output 139. It is defined as an X / V ratio, where X is as in the above, the average signal point energy of the base constellation, which would result if the signal points appeared with equal probability and V is the energy of the average signal point in the precoder output 17 Tomlinson scheme under consideration. With this definition, the conformation gains of Schemes I, II and III at the Tomlinson precoder 17 output are 1.94, 1.34 and 0.46 dB, respectively, as shown in the last column of FIGURE 11. More particularly, the invention it contemplates the use of a channel code which provides such high degree conformation gain of its signal points as the conformation gain of the signal points at the Tomlinson precoder output, which is at least 1.0 dB. The conformation gain of 1.0 dB is significant and is achieved only by means of such a large exchange in bandwidth efficiency that, in the absence of the teachings of the present invention, the prior art - due to the commonly held belief mentioned above - it does not lead to utilization, in a system which includes a Tomlinson precoder, a channel code having such a large amount of conformation gain. (It may be noted that even under the restriction mentioned in the foregoing, the feedback signal of the Tomlinson precoder, there will still be a small loss of conformation gain due to the action of the Tomlinson precoder, so that the gain of conformation provided by the channel code itself, ie before the Tomlinson precoder, must be at least one bit greater than 1.0 dB).

Diversity of Conformation Gain The fact that different loads are achieved by the coding modulation schemes described in the above, by means of the use of coding modulation schemes having different amounts of conformation gain, but which are within the same family, ie, using the same base constellation and the same underlying convolutional code can take advantage of achieving something called "conformation gain diversity".

By this, the support of different loads is understood simply by switching between different coding modulation schemes within the family, each having a different amount of conformation gain. Such conformation gain diversity can be implemented for example, in the time domain such that a radio transmitter can exchange the load for noise immunity in response to: a) changing the channel conditions (based on either long or short term) or b) change the service requirements. As an example of the first, the quality of the received signal could be measured at a remote sensing site and fed back via a wireless telemetry channel or wired to the broadcast site. If the quality deteriorates significantly, the transmitter can "fall backwards" to a scheme that is more immune to noise, with a lower bit rate. As an example of the latter, a broadcaster may decide to increase a single HDTV broadcast service, which offers to include relatively low bit rates or another data service (for example stock price). By switching from one of the encoded modulation schemes to another, the additional load required to supply such an auxiliary service can be accommodated. Indeed, it will be appreciated from a consideration of these two examples, that not all transmitters operating in a given geographical area need to be using the same coding scheme at the same time. Different coding schemes may be used simultaneously for different channels. In FIGURE 1, the control cable 12 is a cable extending to a TV signal source 11 and the concatenated encoder 13. A control signal is provided on this cable that specifies the desired load speed, thus allowing the signal source and the encoder to adapt their operations at the desired speed. Of course, if broadcasters are allowed to use such diversity of conformation gain in their offered services, HDTV or other receivers need to be able to decode signals transmitted using any of the different coding schemes. Advantageously, when the conformation gain diversity is implemented by means of the coding schemes of the same "family" as defined above, a relatively small amount of additional circuitry is required to be able to do that decoding. In particular, using the same "front end" receiver, for example demodulator, equalizer, etc., in the same constellation, can be used. In addition, the main portion of the Viterbi decoder - notably the portion that carries out so-called maximum likelihood computations - will also be the same. The only change is in the circuitry, which converts the recovered sequence of the signal points back into bits. This, however, is a simple matter of using the receiver lookup tables that correspond to the search tables used in the transmitter to implement bit converters and fractional bit encoders, as described above. In the receiver, more particularly, a capacity to determine which scheme is being used, can be achieved in different ways. One way would be to monitor the distribution of cut versions of the outputs of the equalizer 36 and infer which coding scheme was used by the transmitter based on a priori knowledge of the signal point distributions of the different coding schemes as shown in FIGURE 11. Another way would be for the transmitter to explicitly encode this information in the same HDTV signal. The foregoing illustrates only the principles of the invention. For example, all the values of the parameters of the present, such as the number of signal points in the base constellation, the number of dimensions of the base constellation, the number of states and the number of dimensions in the trellis codes, bit rates , signaling speeds, etc., are all illustrative.

Furthermore, although the signals communicated are television signals illustratively, of course, they are simple streams of bits and as such, can be derived from any source and for example, can be derived from text or any other source. Additionally, although the convolutional encoder shown and described herein is a systematic convolutional encoder, it need not be. In addition, although the various components of the transmitter and the receiver are shown as individual functional blocks, the functions of any one or more of them could be provided, for example, by an individual processor operating up to the software control; by one or more digital signal processing (DSP) chips; or by integrated special purpose circuitry. Other implementational variations are possible. For example, the trellis coder 136, the constellation mapper 139 and the signal point interleaver 15 could be replaced by a bank of trellis coders and associated constellation mappers and interleavers. Each of the trellis coders would receive a successive output of the interleaver in a round-robin fashion and each of the interleavers would provide a one-dimensional signal point in a round-robin fashion for the Tomlinson precoder 17. The bands of trellis coders mentioned and the associated constellation mappers do not need to be physically separate elements. On the contrary, the effect of such a bank could be achieved by the time shared of a single trellis coder and a constellation mapper in a form that will be apparent to those skilled in the art. In the case where an HDTV signal is not subjected to the co-channel interference of the NTSC signals, 'the pre-encoder Tomlinson will be removed from the HDTV transmitter of FIGURE 1 and his associated NTSC blocking filter will be removed from the HDTV receiver of FIGURE 3. It will be appreciated that those skilled in the art will be able to contemplate numerous arrangements which, although not shown or explicitly described herein, encompass the principles of the invention and are within its spirit and scope. It is noted that in relation to this date, the best method known by the applicant to carry out the aforementioned invention, is the conventional one for the manufacture of the objects to which it relates. Having described the invention as above, property is claimed as contained in the following:

Claims (54)

CLAIMS 1. An apparatus characterized in that it comprises: a means for receiving a stream of data, an encoder for mapping the data at signal points of a predetermined base constellation with different probabilities, and a Tomlinson precoder for processing the mapped signal points, the encoder and the Tomlinson precoder are such that the signal points provided in the Tomlinson precoder output present at least

1. 0 dB of the conformation gain.

2. The apparatus according to claim 1, further characterized in that it comprises a modulator for applying to a transmission channel a modulated signal representing the signal points provided at the output of the Tomlinson precoder.

3. The apparatus in accordance with the claim 2, characterized in that the data stream represents a television signal.

4. The apparatus according to claim 3, characterized in that the television signal is an HDTV signal.

5. The apparatus in accordance with the claim 2, characterized in that the encoder includes a trellis encoder.

6. The apparatus according to claim 2, characterized in that the encoder includes an encoder Reed-Solomon concatenated with a trellis encoder.

7. The apparatus according to claim 2, characterized in that the signal points are signal points of a dimension of a N-dimensional symbol constellation.

8. The apparatus according to claim 7, characterized in that the modulated signal is a vestigial sideband signal.

9. The apparatus according to claim 2, characterized in that the Tomlinson precoder includes a combiner, which combines the mapped signal points with a feedback signal, which is weak compared to the average energy of the mapped signal points.

10. The apparatus according to claim 9, characterized in that the feedback signal is substantially 12 db, or more, below the average energy of the mapped signal points.

11. The apparatus according to claim 9, characterized in that the Tomlinson precoder further includes a finite impulse response filter, which generates the feedback signal, the filter having as its input the signal points provided at the output of the Tomlinson precoder.

12. The apparatus in accordance with the claim 11, characterized in that the Tomlinson precoder further includes a module device which performs a module function at the output of the combiner to generate the signal points provided at the output of the Tomlinson precoder.

13. The apparatus in accordance with the claim 12, characterized in that the feedback signal is substantially 12 dB, or more, below the average energy of the mapped signal points.

14. A receiver apparatus for use in a system in which a data stream is received, in which the data is mapped by an encoder at signal points of a predetermined base constellation with different probabilities, in which the mapped signal points are processed by means of a Tomlinson precoder, the Tomlinson coder and precoder are such that the signal points provided at the Tomlinson precoder output exhibit at least 1.0 db of shaping gain and in which a modulated signal representing the points of The signal provided at the Tomlinson pre-encoder output are applied to a transmission channel, the receiver apparatus is characterized in that it comprises: a means for receiving the modulated signal from the transmission channel, and a means for recovering the data of the received signal.

15. The apparatus according to claim 14, characterized in that the means for recovery includes a blocking filter having null values at a plurality of predefined frequency locations.

16. The apparatus according to claim 15, characterized in that the blocking filter is an NTSC blocking filter.

17. The apparatus according to claim 14, characterized in that the data stream represents a television signal and wherein the apparatus further includes a television element for displaying the retrieved data.

18. The apparatus according to claim 17, characterized in that the television signal is an HDTV signal.

19. The apparatus according to claim 14, characterized in that the encoder includes a trellis encoder and wherein the recovery means includes a Viterbi decoder.

20. The apparatus according to claim 14, characterized in that the encoder includes a Reed-Solomon encoder concatenated with a trellis encoder and wherein the recovery means includes a Viterbi decoder concatenated with a Reed-Solomon decoder.

21. The apparatus according to claim 14, characterized in that the signal points are signal points of a dimension of a N-dimensional symbol constellation.

22. The apparatus according to claim 21, characterized in that the modulated signal is a vestigial sideband signal and wherein the receiver apparatus further includes a vestigial sideband demodulator.

23. The apparatus according to claim 18, characterized in that the Tomlinson precoder includes a combiner, which combines the mapped signal points with a feedback signal, which is weak compared to the average energy of the mapped signal points.

24. The apparatus according to claim 23, characterized in that the feedback signal is substantially 12 db or more, below the average energy of the mapped signal points.

25. The apparatus according to claim 23, characterized in that the Tomlinson precoder further includes a finite impulse response filter, which generates the feedback signal, the filter having as its input the signal points provided at the Tomlinson precoder output.

26. The apparatus in accordance with the claim 25, characterized in that the Tomlinson precoder further includes a module device which performs a module function at the output of the combiner to generate the signal points provided at the output of the Tomlinson precoder.

27. The apparatus in accordance with the claim 26, characterized in that the feedback signal is substantially 12 dB below the average energy of the mapped signal points.

28. A method characterized in that it comprises the steps of receiving a data stream, mapping the data in an encoder at signal points of a predetermined base constellation with different probabilities, and processing the signal points mapped through a Tomlinson precoder, the encoder and the Tomlinson precoder are such that the signal points provided at the output of the Tomlinson precoder have at least 1.0 dB of the conformation gain.

29. The apparatus according to claim 28, further characterized in that it comprises a modulator for applying to a transmission channel a modulated signal representing the signal points provided at the output of the Tomlinson precoder.

30. The apparatus according to claim 28, characterized in that the data stream represents a television signal.

31. The apparatus according to claim 30, characterized in that the television signal is an HDTV signal.

32. The apparatus according to claim 28, characterized in that the encoder includes a trellis encoder.

33. The apparatus according to claim 28, characterized in that the encoder includes a Reed-Solomon encoder concatenated with a trellis encoder.

34. The apparatus according to claim 28, characterized in that the signal points are signal points of a dimension of a N-dimensional symbol constellation.

35. The apparatus according to claim 34, characterized in that the modulated signal is a vestigial sideband signal.

36. The apparatus in accordance with the claim 30, characterized in that the Tomlinson precoder includes a combiner, which combines the mapped signal points with a feedback signal, which is weak compared to the average energy of the mapped signal points.

37. The apparatus according to claim 36, characterized in that the feedback signal is substantially 12 db, or more, below the average energy of the mapped signal points.

38. The apparatus according to claim 36, characterized in that the Tomlinson precoder further includes a finite impulse response filter, which generates the feedback signal, the filter having as its input the signal points provided at the output of the Tomlinson precoder.

39. The apparatus according to claim 38, characterized in that the Tomlinson precoder further includes a module device which performs a module function at the output of the combiner to generate the signal points provided at the output of the Tomlinson precoder.

40. The apparatus according to claim 38, characterized in that the feedback signal is substantially 12 dB, or more, below the average energy of the mapped signal points.

41. A method for being used in a system in which a data stream is received, in which the data is mapped by an encoder at signal points of a predetermined base constellation with different probabilities, in which the mapped signal points are processed by means of a Tomlinson precoder, the Tomlinson coder and precoder are such that the signal points provided at the Tomlinson precoder output exhibit at least 1.0 db of shaping gain and in which a modulated signal representing the signal points provided at the output of the Tomlinson precoder are applied to a transmission channel, the method is characterized in that it comprises the steps of: receiving the modulated signal from the transmission channel, and recovering the data of the received signal.

42. The method according to claim 41, characterized in that the recovery step includes the step of processing the modulated signal by means of a blocking filter having null values at a plurality of predefined frequency locations.

43. The method in accordance with the claim 42, characterized in that the blocking filter is an NTSC blocking filter.

44. The method according to claim 41, characterized in that the data stream represents a television signal and wherein the method further comprises the step of displaying the retrieved data.

45. The method according to claim 44, characterized in that the television signal is a signal HDTV

46. The method according to claim 41, characterized in that the encoder includes a trellis encoder and wherein the recovery step includes the Viterbi decoding step.

47. The method according to claim 41, characterized in that the encoder includes a Reed-Solomon encoder concatenated with a trellis encoder and wherein the recovery step includes a Viterbi decoding stage concatenated with a Reed-Solomon decoding stage.

48. The method in accordance with the claim 41, characterized in that the signal points are signal points of a dimension of a constellation of N-dimensional symbols.

49. The method according to claim 48, characterized in that the modulated signal is a vestigial sideband signal and wherein the recovery step further includes a vestigial sideband demodulation step.

50. The method according to claim 41, characterized in that the Tomlinson precoder includes a combiner, which combines the mapped signal points with a feedback signal, which is weak compared to the mapped signal points.

51. The method according to claim 50, characterized in that the feedback signal * is substantially 12 db or more, below the average energy of the mapped signal points.

52. The method according to claim 50, characterized in that the Tomlinson precoder further includes a finite impulse response filter, which generates the feedback signal, the filter having as its input the signal points provided at the output of the Tomlinson precoder.

53. The method in accordance with the claim 52, characterized in that the Tomlinson precoder further includes a module device which performs a module function at the output of the combiner to generate the signal points provided at the output of the Tomlinson precoder.

54. The compliance method runs the claim 53, characterized in that the feedback signal is substantially at least 12 dB below the average energy of the mapped signal points.