Patent Application
20240048432
Dual-carrier modulation (DCM) facilitates coded orthogonal frequency-division multiplexed (COFDM) signal being successfully received despite the respective power of some of the plurality of carrier waves being severely reduced, as can happen owing to cancellations in energy caused by multipath reception conditions. The full bandwidth of the DCM-COFDM signal is separable into a lower-frequency subband and a higher-frequency subband. Common practice is to transmit the same coded digital data (CDD) two-fold on separate COFDM carrier waves, separated by half the full bandwidth of the DCM-COFDM signal. The likelihood of the two COFDM carrier waves both being severely reduced owing to cancellations in energy caused by multipath reception is low, with one carrier wave being in the lower-frequency subband and the other carrier wave being in the higher-frequency subband. Accordingly, a DCM-COFDM signal receiver of good design can receive CDD and satisfactorily recover digital data therefrom despite multipath reception conditions that cancel energy of a number of COFDM carrier waves.
Patent application US-20170104553-A1 published 13 Apr. 2017, titled “LDPC Tone Mapping Schemes for Dual-Sub-Carrier Modulation in WLAN”, and filed 11 Oct. 2016 claiming inventorship for Jian-Han Liu, Sheng-Quan Hu, Tian-Yu Wu and Thomas Edward Pare, Jr. describes a species of split-spectrum COFDM modulation signal which utilizes dual carrier modulation (DCM). The DCM modulates the same information on pairs of carriers, which carriers in each pair of them can be separated in frequency to improve frequency diversity in OFDM systems. US-20170104553-A1 points out that such separation improves reliability of reception, especially when there are narrow-band interferences (such as occur because of multipath reception, for example).
US-20170104553-A1 further describes dissimilar respective mappings of first and second sets of square 16QAM symbol constellations transmitted concurrently in time, the primary design goal of this being to reduce the peak-to-average-power ratio (PAPR) of the DCM-COFDM signals significantly. FIGS. 4(a) and 4(b) of the drawings of US-20170104553-A1 depict the first and second patterns of labeling uniformly-spaced lattice points in these first and second sets of square 16QAM symbol constellations. These first and second patterns of labeling uniformly-spaced lattice points in first and second sets of square 16QAM symbol constellations correspond to those depicted in FIG. 1(a) of the drawings of patent application US-200800212694-A1 published 4 Sep. 2008, titled “Signal Decoding Systems”, and filed 8 May 2007 claiming inventorship for Martin Geoffrey Leach and Peter Anthony Borowski. US-200800212694-A1 refers to the DCM-COFDM signal being configured for reduction of its peak-to-average-power ratio (PAPR), but does not prescribe each pair of QAM carriers concurrently conveying the same CDD being spaced half-channel-width apart in frequency to improve the reliability of reception when narrow-band interferences occur.
Square QAM symbol constellations have been favored in the transmission of DCM-COFDM signals because the positioning of mapping labels at regularly-spaced lattice points in the QAM symbol constellations facilitate designs of receivers for DCM-COFDM signals. Analog-to-digital conversion procedures following the demodulation of QAM of the COFDM carrier waves are straightforward, and there is less likelihood of bit errors in the two sets of DDC that are subsequently combined. Analog-to-digital conversion requires less resolution than needed for phase-amplitude modulation (PAM) in which labeled lattice points are laid out uniformly along circular loci in the complex-number domain.
Square QAM symbol constellation mapping with uniformly spaced lattice points is employed in the DCM-COFDM signals described in U.S. Pat. No. 10,637,711 titled “COFDM DCM communication systems with preferred labeling-diversity formats” granted 28 Apr. 2020 to Allen LeRoy Limberg. In U.S. Pat. No. 10,637,711 Limberg speculated that DCM-COFDM signals broadcast for reception by stationary DTV receivers would most likely use square 64QAM symbol constellations, although square 256QAM symbol constellations or even square 1024QAM symbol constellations could be used instead. Going from 16QAM to 64QAM symbol constellations increases the number of bits in lattice-point labels from four to six, increasing channel capacity 50%. Going from 16QAM to 256QAM symbol constellations increases the number of bits in lattice-point labels from four to eight, doubling channel capacity, but the increase of channel capacity over that for 64QAM symbol constellations is only 33%. Going from 256QAM to 1024QAM symbol constellations increases the number of bits in lattice-point labels from eight to ten, for a 25% increase in channel capacity. The diminishing increases in channel capacity for larger than 64QAM symbol constellations may not justify the extra risk of error caused by noise. Stronger FEC coding necessary to overcome these errors will decrease channel capacity, offsetting increase in channel capacity owing to a larger QAM symbol constellation.
For a given type of quadrature amplitude modulation (QAM) of COFDM carrier waves, the data throughput of the COFDM signal of given full bandwidth is halved using DCM as compared to using individual carrier modulation in which coded digital data is transmitted only once. To compensate against such loss in data throughput, the number of labeled lattice points in QAM symbols transmitted using DCM-COFDM will customarily be larger than the number of labeled lattice points in QAM symbols transmitted using individual carrier modulation. Similar data throughput requires there be four times as many labeled lattice points in the QAM symbol constellation mapping of DCM-COFDM signal as in the QAM symbol constellation mapping of COFDM signal using individual carrier modulation.
Accordingly, the distance between labeled lattice points in the QAM symbols used in DCM-COFDM signal is halved compared to the distance between labeled lattice points in the QAM symbols used in COFDM signal using individual carrier modulation, presuming the maximum amplitude of COFDM carrier waves to be kept the same. So, the bit-error ratio (BER) per a COFDM carrier received over an additive-white-Gaussian-noise (AWGN) reception channel will be 6 dB higher for the QAM symbols used in DCM-COFDM signal than for the QAM symbols in COFDM signal using individual carrier modulation. However, supposing both of the two DCM-COFDM carrier waves transmitting the same coded digital data two-fold are received over a shared AWGN reception channel with “flat” frequency response, the two sets of the same coded digital data (CDD) can subsequently be combined to improve BER, owing to the two signals being better correlated than the respective random noise accompanying each of those signals.
Another phenomenon adversely affects BER when there is increase in the number of lattice points in the square QAM symbol constellations in the dual mapping employed in generation of the DCM-COFDM signaling. The digital information encoded in the less reliable bits of the LPLs provides a sort of “noise” for the digital information encoded in the more reliable bits of the LPLs, which “noise” may displace the amplitudes of QAM carriers in the DCM-COFDM signal from respective values where more reliable bits of recovered CDD would be at their most reliable.
To considerable degree, a DCM-COFDM signal receiver can counter both these tendencies toward increased BER as the number of labeled lattice points in QAM symbol constellations is increased. The receiver accomplishes this by maximum-ratio combining of corresponding bits in the LPLs of the respective results of QAM demapping dual carriers that convey the same CDD. Yet, there is still an appreciable increase in BER, arising from continuous noise accompanying the DCM-COFDM signal. Generally, this continuous noise is considered theoretically to be additive white Gaussian noise (AWGN). In actuality, by way of example, this continuous noise is primarily composed of (a) atmospheric noise arising during passage of COFDM signal through an over-the-air communication channel, (b) thermal noise arising in components of the COFDM signal receiver, and (c) quantization noise arising during analog-to-digital conversions in the receiver.
Maximum-ratio combining (MRC) is a method of diversity combining in which (a) the signals from each channel are added together, (b) the gain of each channel is made proportional to the rms signal level and inversely proportional to the mean square noise level in that channel, and (c) different proportionality constants are used for each channel. MRC of carriers amplitude modulated by analog signals was described being done at symbol level by Leonard Kahn in Correspondence titled “Ratio Squarer” appearing in Proceedings of the IRE. Vol. 42, Issue 11, (November 1954). MRC was subsequently performed at QAM symbol level on carriers quadrature amplitude modulated by digital signals. U.S. Pat. No. 7,236,548 titled “Bit level diversity combining for COFDM system” issued 26 Jun. 2007 to Monisha Ghosh, Joseph P. Meehan and Xuemei Ouyang describes diversity combining of digitally modulated carriers being performed at bit level, rather than at symbol level. Their work was directed to multiple-in/multiple-out (MIMO) reception of COFDM signals from plural-antenna arrays. U.S. Pat. No. 7,236,548 does not indicate labeling diversity having been used in their work. U.S. Pat. No. 7,236,548 reports BER being 2.5 dB lower when diversity combining is performed at bit level, as opposed to performing diversity combining at symbol level.
U.S. Pat. No. 10,637,711 titled “COFDM DCM communication systems with preferred labeling-diversity formats” prescribes labeling diversity of the following sort for the first and second patterns of labeling uniformly-spaced lattice points in first and second sets of square QAM symbol constellations used to convey the same CDD. The bits in the LPLs for the first pattern of mapping QAM symbol constellations, which bits are more likely to be in error owing to accompanying AWGN, correspond to ones of the bits in the LPLs for the second pattern of mapping QAM symbol constellations, which bits are less likely to be in error owing to accompanying AWGN. Furthermore, the bits in the LPLs for the second pattern of mapping QAM symbol constellations, which bits are more likely to be in error owing to accompanying AWGN, correspond to ones of the bits in the LPLs of the first pattern of mapping QAM symbol constellations, which are less likely to be in error owing to accompanying AWGN. This sort of labeling diversity tends to reduce the BER of CDD resulting from maximum-ratio combining (MRC) performed on a bit-by-bit basis.
In part, this is because, providing both COFDM carrier waves concurrently conveying the same CDD are received with reasonable strength, one of the soft bits supplied for MRC performed at bit level will be less likely to be error if the other of the soft bits supplied for MRC performed at bit level is more be likely to be error. Accordingly, an MRC bit response dominated by the soft bit less likely to be error will likely be correct even if there be disagreement of the input soft bits as to hard-bit value. Additionally, if the two soft bits individually considered have similar likelihood of error and also correspond to each other as to hard-bit value, the likelihood of the hard-bit value of the MRC soft-bit response being correct can be significantly increased. That is, the labeling diversity is designed to facilitate “bit-reliability averaging” (BRA) by the maximum-ratio combining apparatus in the DCM=COFDM signal receiver.
There is a basic guiding principle for reducing the PAPR of DCM-COFDM signaling in which pairs of carriers concurrently conveying the same coded digital data (CDD) are modulated in accordance with respective patterns of mapping CDD to square QAM symbol constellations, each of which patterns has a prescribed number of labeled lattice points therein. Namely, the lattice-point labels (LPLs) associated with high peak power in either one of the two patterns of mapping lattice points in QAM symbol constellations are associated with low peak power in the other one of the two patterns of mapping lattice points in QAM symbol constellations. Accordingly the DCM-COFDM signal has a strong tendency to have closer-to-average power constantly. U.S. Pat. No. 10,601,624 titled “COFDM DCM signaling that employs labeling diversity to minimize PAPR”, granted 24 Mar. 2020 to Allen LeRoy Limberg, discloses this basic guiding principle being applied to DCM-COFDM signal in which square 64QAM symbol constellations govern the modulation of COFDM carrier waves. Patent application US-20210243064-A1 of A. L. R. Limberg published 5 Aug. 2021 and titled DCM-COFDM signaling using square QAM symbol constellations with lattice-point labels having over four bits apiece” discloses this basic guiding principle being applied also to DCM COFDM signal in which square 256QAM symbol constellations govern the modulation of COFDM carrier waves. US-20210243064-A1 indicates this basic guiding principle can also be applied also to DCM COFDM signals in which square 1024QAM symbol constellations or square 4096QAM symbol constellations govern the modulation of COFDM carrier waves.
Superposition coded modulation (SCM) is important in the QAM of the carrier waves of preferred formats for DCM-COFDM signals, in which formats QAM is characterized by coded digital data (CDD) being mapped to square QAM symbol constellations. SCM mapping of square QAM symbol constellations used in DCM-COFDM signaling allows labeling diversity designed to facilitate “bit-reliability averaging” during MRC, while concurrently adhering to the above-described basic guiding principle for PAPR reduction. (The term “individual carrier modulation” is used herein, rather than the substantially equivalent term “single carrier modulation” used in other texts. Single carrier modulation is referred to as “SCM” in some texts other than this document, but hereafter in this document the acronym “SCM” will be used exclusively to refer to superposition coded modulation.)
SCM is described in detail by Li Ping, Jun Tong, Xiaojun Yuan and Qinghua Guo in their paper “Superposition Coded Modulation and Iterative Linear MMSE Detection”, IEEE Journal on Selected Areas in Communications, Vol. 27, No. 6, August 2009, pp. 995-1004. In the SCM the Ping et alli paper describes, the four quadrants of square QAM symbol constellations are each Gray mapped independently from the others and from the pair of bits in the map label specifying that quadrant. Ping et alli studied iterative linear minimum-mean-square-error (LMMSE) detection being used in the reception of SCM and found that SCM offers an attractive solution for highly complicated transmission environments with severe interference. Ping et alli analyzed the impact of signaling schemes on the performance of iterative LMMSE detection to prove that among all possible signaling methods, SCM maximizes the output signal-to-noise/interference ratio (SNIR) in the LMMSE estimates during iterative detection. The Ping et alli paper describes measurements that were made demonstrating that SCM outperforms other signaling methods when multiple-user/multiple-antenna/multipath channels have iterative LMMSE detection applied to them.
In Gray mapping of an entire square QAM symbol constellation, the lattice point labels (LPLs) of adjacent lattice points differ in just a single bit place, whether or not those two lattice points are both situated in the same (−I,+Q) quadrant, the same (+I,+Q) quadrant, the same (I,−Q) quadrant or the same (−I,−Q) quadrant. In SCM mapping of an entire square QAM symbol constellation, the lattice point labels (LPLs) of two adjacent lattice points differ in just a single bit place, whether or not those two lattice points are both situated in the same (−I,+Q) quadrant, the same (+I,+Q) quadrant, the same (I,−Q) quadrant or the same (−I,−Q) quadrant. The SCM mapping of a square QAM symbol constellation differs from the Gray mapping of such a QAM symbol constellation in that, if adjacent lattice points are situated in different quadrants of the QAM symbol constellation, their LPLs differ from each other in two bit places, rather than in just a single bit place.
These differences in two bit places facilitate all palindromic LPLs in an SCM mapping of a square QAM symbol constellation being arranged along the −45°, +45°, +135° and −135° axes of its (−I,+Q) quadrant, its (+I,+Q) quadrant, its (I,−Q) quadrant and its (−I,−Q) quadrant respectively.
Such arrangements of palindromic LPLs are not completely possible in Gray mappings of a square QAM symbol constellation. Such arrangements of all palindromic LPLs in SCM mappings of square QAM symbol constellations in DCM-COFDM signaling allow PAPR reduction together with labeling diversity that supports bit-reliability averaging during MRC. Patent application US-20210243064-A1 of A. L. R. Limberg published 5 Aug. 2021 and titled DCM-COFDM signaling using square QAM symbol constellations with lattice-point labels having over four bits apiece” provides detailed description of how palindromic LPLs can be advantageously employed in SCM mappings of square QAM symbol constellations for DCM-COFDM signaling, which description is incorporated herein by reference and provides background information useful for advantageously employing palindromic LPLs in DCM-COFDM signaling that uses square NU-QAM symbol constellations with lattice-point labels having over four bits apiece.
The labeled lattice points in each of the pair of square 16QAM symbol constellations described in US-20080212694-A1 and in US-20170104553-A1 (referenced above) are arranged in a respective SCM mapping pattern. Four of the 16 lattice points in any square 16QAM symbol constellation have respective palindromic lattice-point labels, each differing from the three other palindromic LPLs. These four palindromic LPLs are separated into first and second sets of two 4-bit LPLs for further consideration. The first set of palindromic LPLs are located in the outermost corners of two quadrants diagonally opposite from each other in a first of that pair of square 16QAM symbol constellations, and the second set of palindromic LPLs are in the innermost corners of the other two quadrants of that first pair of square 16QAM symbol constellations. The second set of palindromic LPLs are located in the outermost corners of two quadrants diagonally opposite from each other in the second of that pair of square 16QAM symbol constellations, and the first set of palindromic LPLs are in the innermost corners of the other two quadrants of that second pair of square 16QAM symbol constellations. Different, other pairs of square 16QAM symbol constellations having labeled lattice points in each of them arranged in a respective SCM mapping pattern are depicted in FIGS. 10 through 15 of patent application US-20190334755-A1 published 31 Jan. 2020, titled “COFDM DCM systems with preferred labeling-diversity formats” and claiming an original filing date of 27 Jun. 2019 for inventor Allen LeRoy Limberg.
Pairs of square 64QAM symbol constellations having labeled lattice points in each of them arranged in a respective SCM mapping pattern are depicted in FIGS. 34 through 39 of U.S. Pat. No. 10,594,537, issued 17 Mar. 2020, titled “Receivers for COFDM Signals conveying the same data in lower- and upper-frequency sidebands” and claiming an original filing date of 17 Jul. 2018 for inventor Allen LeRoy Limberg. Eight of the lattice points in any square 64QAM symbol constellation have respective palindromic lattice-point labels, each different from the seven others. These eight palindromic LPLs are separated into first and second sets of four 6-bit LPLs for further consideration. A pair of the 64QAM symbol constellations depicted in two of FIGS. 34 through 39 of U.S. Pat. No. 10,594,537 exhibit lessened PAPR in a DCM-COFDM signal employing them, providing the following conditions obtain. Respective ones of the first set of four palindromic LPLs label the corners of the four quadrants of one of the SCM-mapped 64QAM symbol constellations, which corners are furthest from constellation center; and those same four palindromic LPLs label the corners of the four quadrants of the other of the SCM-mapped 64QAM symbol constellations, which corners are closest to constellation center. Respective ones of the second set of four palindromic LPLs label the corners of the four quadrants of said one of the SCM-mapped 64QAM symbol constellations, which corners are closest to constellation center; and those same four palindromic LPLs label the corners of the four quadrants of said other of the SCM-mapped 64QAM symbol constellations, which corners are furthest from constellation center.
Another pair of square 64QAM symbol constellations having labeled lattice points in each of them arranged in a respective SCM mapping pattern are depicted in FIGS. 30 and 31 of U.S. Pat. No. 10,601,624 issued 24 Mar. 2020, titled “COFDM DCM signaling that employs labeling diversity to minimize PAPR” and claiming an original filing date of 12 Dec. 2018 for inventor Allen LeRoy Limberg. Pairs of palindromic LPLs that are in the corners of two of the quadrants of each one of the SCM-mapped square 64QAM symbol constellations, which corners are furthest from constellation center, are in the corners of two of the quadrants of the other one of the SCM-mapped square 64QAM symbol constellations, which corners are closest to constellation center.
Besides describing SCM-mapped square 64QAM symbol constellations being used in DCM-COFDM signaling, US-20210243064-A1 titled “DCM-COFDM signaling using square QAM symbol constellations with lattice-point labels having over four bits apiece” describes SCM-mapped square 256QAM symbol constellations being used in DCM-COFDM signaling. Square 256QAM symbol constellations each have eight palindromic LPLs.
In the documents referred to supra, each of the square QAM symbol constellations specifically described for use in DCM-COFDM signaling presumably has its labeled lattice points arranged in uniformly-spaced columns and in uniformly-spaced rows between. None of these documents calls into consideration possible benefits for DCM-COFDM signaling that could be obtained using NU-QAM, wherein the labeled lattice points of square QAM symbol constellations are arranged in non-uniformly spaced columns and in non-uniformly spaced rows.
ETSI EN 300 744 V1.5.1 (2004-06) European Standard (Telecommunications series) titled “Digital Video Broadcasting (DVB); Framing structure, channel coding and modulation for digital terrestrial television” includes a sub-sub-section 4.3.5 titled “Signal constellations and mapping” that describes several varieties of non-uniform QAM employing square symbol constellations. Each of these varieties of square symbol constellations employing NU-QAM has a respective quadrant of Gray mapping in each quadrant of the complex number plane, and the labeled lattice points within each quadrant are uniformly spaced apart from each other. However, the quadrants of Gray mapping are spaced further from the in-phase (I) and quadrature (Q) axes of the complex number plane than they would be in Gray mapping in which all labeled lattice points are uniformly spaced apart from each other.
This increased spacing between the quadrants of Gray mapping in the SCM mapping of square NU-QAM symbol constellations has been found to offer unexpected improvement in the general sort of DCM-COFDM signaling disclosed in the inventor's U.S. Pat. No. 10,637,711. In accordance with what is noted supra regarding QAM, the digital information encoded in the less reliable bits of the lattice point labels (LPLs) of SCM-mapped square NU-QAM symbol constellations provides a sort of “noise” for the digital information encoded in the more reliable bits of the LPLs, which “noise” may displace the amplitudes of NU-QAM carriers in the DCM-COFDM signal from respective values where more reliable bits of recovered CDD would be at their most reliable. The increased spacing between the quadrants of Gray mapping in the SCM mapping of square NU-QAM symbol constellations increases the energy of AWGN needed to cause error in the most reliable bits of the LPLs of SCM-mapped square NU-QAM symbol constellations, these most reliable bits of the LPLs being the pair of bits that specify which quadrant of the SCM-mapped square NU-QAM symbol constellation is being accessed by a NU-QAM carrier wave. The increased spacing between the quadrants of Gray mapping in the SCM mapping of square NU-QAM symbol constellations has been further found to offer unexpected improvement in the general sort of DCM-COFDM signaling disclosed in the inventor's patent application US-20210243064-A1.
The invention in certain of its aspects is embodied in methods for generating DCM-COFDM signaling that conveys the same coded digital data two-fold, as arranged in two concurrent sets of successive square NU-QAM symbol constellations. The square NU-QAM symbol constellations are superposition-coded-modulation (SCM) mapped, rather than being Gray mapped. The pair of square NU-QAM symbol constellations exhibit labeling diversity that supports a DCM-COFDM signal receiver in its providing bit-reliability averaging during maximum-ratio combining coded digital data transmitted two-fold. The invention in certain other of its aspects is embodied in electronic apparatus configured for processing the DCM-COFDM signaling that conveys the same coded digital data two-fold, as arranged in two concurrent sets of successive square NU-QAM symbol constellations.
By way of example, the electronic apparatus configured for processing the DCM-COFDM signaling is transmitter apparatus designed for transmitting the DCM-COFDM signaling that conveys the same coded digital data two-fold, as arranged in two concurrent sets of successive square NU-QAM symbol constellations. Particularly, the dual mapping of CDD to NU-QAM carrier waves is designed to exhibit the novel labeling diversity that is characteristic of various aspects of the invention and that keeps the PAPR of the DCM-COFDM signal minimal. The resultant low PAPR of the DCM-COFDM signaling permits the linear radio-frequency power amplifier to be of simpler design, which no longer needs be of Doherty type to avoid excessive standby power consumption.
By way of further example, the electronic apparatus configured for processing DCM-COFDM signaling is receiver apparatus designed for receiving the DCM-COFDM signaling that conveys the same coded digital data two-fold, as arranged in two concurrent sets of successive square NU-QAM symbol constellations. The invention is embodied in apparatus for demodulating dual-mapped NU-QAM signals that have lattice-point labeling diversity between them that benefits diversity combining of their soft bits of coded digital data (CDD). First and second NU-QAM symbol demappers in this receiver apparatus demap first and second sets, respectively, of successive square NU-QAM symbols. The same CDD is conveyed in both the first and second sets of successive NU-QAM symbols. The respective demapping results from the first and second NU-QAM symbol demappers are maximum-ratio combined at bit level to supply CDD to a decoder that recovers the digital data.
The
In accordance with a first mapping pattern, the successive LPLs are mapped to a first set of successive superposition-coded-modulation (SCM) non-uniform quadrature-amplitude-modulation (NU-QAM) symbols in step S3A. In accordance with a second mapping pattern, the successive LPLs are mapped to a second set of successive SCM NU-QAM symbols in step S3B. The SCM NU-QAM symbols are defined by square NU-QAM symbol constellations having uniform spacing between adjacent labeled lattice points within each quadrant thereof (or, alternatively. just within each sub-quadrant thereof). There is labeling diversity between the first and second patterns of mapping LPLs to the first and the second sets of successive SCM NU-QAM symbols respectively.
Preferably, the first and second patterns that SCM map successive segments of coded digital data to square NU-QAM symbol constellations exhibit labeling diversity between them that is described more particularly in the next five paragraphs. The preferred first and second patterns for SCM mapping successive segments of coded digital data to square NU-QAM symbol constellations benefit a step S7 of the
A first pair of the 2N bits in the LPLs of the first SCM mapping pattern (e.g., the initial two bits) indicate which quadrant of the SCM symbol constellation that LPL prescribes, and the correspondingly positioned first pair of bits in the correspondingly valued LPLs of the second SCM mapping pattern indicate which sub-quadrant of that quadrant of the SCM symbol constellation that LPL prescribes.
A second pair of the 2N bits in the LPLs of the second SCM mapping pattern (e.g., the final two bits) indicate which quadrant of the SCM symbol constellation that LPL prescribes, and the correspondingly positioned second pair of bits of the correspondingly valued LPLs of the first SCM mapping pattern indicate which sub-quadrant of that quadrant of the SCM symbol constellation that LPL prescribes.
If N be an even at least four, there will be further pairs of bits in the LPLs of the first SCM mapping pattern and corresponding further pairs of bits in the LPLs of the second SCM mapping pattern. Further pairs of bits in the LPLs of the first SCM mapping pattern more likely to be in error if contaminated with AWGN of given power are arranged to correspond to further pairs of bits in the LPLs of the second SCM mapping pattern less likely to be in error than the aforementioned further pairs of bits if contaminated with AWGN of given power. Further pairs of bits in the LPLs of the second SCM mapping pattern more likely to be in error than if AWGN of given power are arranged to correspond to further pairs of bits in the LPLs of the first SCM mapping pattern less likely to be in error than the aforementioned further pairs of bits if contaminated with AWGN of given power.
If N be an odd integer at least three, there will be a third pair of bits in the LPLs of the first SCM mapping pattern (e.g., the central two bits) and a corresponding third pair of bits in the LPLs of the second SCM mapping pattern. These third pairs of bits will correspond in value and will be equally likely to be in error if contaminated with AWGN of given power.
If N be an odd integer at least five, there will be still further pairs of bits in the LPLs of the first SCM mapping pattern and corresponding still further pairs of bits in the LPLs of the second SCM mapping pattern. Still further pairs of bits in the LPLs of the first SCM mapping pattern more likely to be in error than the aforementioned further pairs of bits if contaminated with AWGN of given power are arranged to correspond to still further pairs of bits in the LPLs of the second SCM mapping pattern less likely to be in error than the aforementioned further pairs of bits if contaminated with AWGN of given power. Still further pairs of bits in the LPLs of the second SCM mapping pattern more likely to be in error than the aforementioned further pairs of bits if contaminated with AWGN of given power are arranged to correspond to still further pairs of bits in the LPLs of the first SCM mapping pattern less likely to be in error than the aforementioned further pairs of bits if contaminated with AWGN of given power.
Modulating the carriers of the DCM-COFDM signal according to the NU-QAM symbol constellations is carried out by parallel steps S4A and S4B of the general method illustrated in
In step S4A COFDM carriers in the lower half of the frequency spectrum of the communication channel for the DCM-COFDM signal are modulated in prescribed order according to respective ones of the first set of successive NU-QAM symbols supplied by the mapping step S3A.
In step S4B COFDM carriers in the upper half of the frequency spectrum of the communication channel for the DCM-COFDM signal are modulated in prescribed order according to respective ones of the second set of successive NU-QAM symbols supplied by the mapping step S3B. Step 5 for generating a full-frequency-spectrum DCM-COFDM signal computes the inverse-Fourier transform of all the COFDM carriers supplied by steps 4A and 4B of modulating carriers.
The
The first set of successive NU-QAM symbols supplied in the
In the
In the
Bit-reliability-averaging in the step S7 of diversity combining avoids the step S8 of decoding the forward-error correction coding of the CDD being presented with bits with low reliability of being correct when there is accompanying AWGN. This increases the capability of the decoding of FEC coding to recover correct digital data at higher levels of accompanying AWGN. This increased capability is more pronounced as NU-QAM symbol size is made larger. The advantages of the general method for demodulating dual-mapped NU-QAM signals illustrated in the
The data throughput for a COFDM signal using dual-carrier modulation (DCM) of its carriers is half the data throughput for a COFDM signal the carriers of which are modulated individually, presuming the carriers in both signals are all quadrature amplitude modulated in accordance with the same-size square NU-QAM symbol constellations. The data throughput of DCM-COFDM signal can be doubled by squaring the number of labeled lattice points in the square NU-QAM symbol constellations that describe the modulation of the carriers of the DCM-COFDM signal used to convey data. This is because squaring the number of labeled lattice points in a square NU-QAM symbol constellation doubles the number of bits in each lattice-point label (LPL). However, increasing the even number of bits in the LPLs of square NU-QAM symbol constellations increases bit error rate (BER) of coded data recovered during reception of a COFDM signal over an additive-white-Gaussian-noise (AWGN) channel. This increase in BER is problematic, especially at lower received signal strengths.
Each quadrupling of the number of lattice points in the square NU-QAM symbol constellations halves the amplitude of AWGN that will be of a threshold value small enough not to cause error in any bit of the lattice-point labeling of the NU-QAM symbol constellations recovered by a DCM-COFDM signal receiver. As the amplitude of AWGN increases more and more above that threshold value, increasingly more of the bits in the lattice-point labels of NU-QAM symbols will be susceptible of error.
The bit of a particular bit position in the lattice-point label of a square NU-QAM symbol constellation is more likely to be in error when the location of the lattice point approaches the boundary of the bin for the value of that bit position with the bin for the other value of that bit position. The bit of a particular position in the lattice-point label of a square NU-QAM symbol constellation is least likely to be in error when the location of the lattice point is in an outside edge of the NU-QAM symbol constellation which includes the boundary of the bin for the value of that bit position.
If the bin for the value of a particular bit position in the lattice-point label of a square NU-QAM symbol constellation does not have a boundary for such value that is in an outside edge of the NU-QAM symbol constellation, the likelihood of a bit in that position in the lattice-point label being correct will be greatest when a component of carrier amplitude modulation terminates nearest the center of that bin. (The component of carrier amplitude modulation defined by a bit will usually not be at exact center of the bin it defines, owing to the offset introduced by bits defining bins smaller than its own bin.) The likelihood of a bit in a position in the lattice-point label being correct when a component of carrier amplitude modulation terminates nearest the center of a bin for that bit position is directly proportional to the distance to the edge of such bin.
Increasing the even number of bits in the LPLs of SCM-mapped square NU-QAM symbol constellations to more than four increases BER of coded data recovered during reception of a DCM-COFDM signal over an AWGN channel. As pointed out supra, increased BER is especially problematic at lower received signal strengths. In regard to the bits if LPLs associated with bins of smaller size, the increase in BER associated with those bits despite AWGN being quite low in power is attributable to the AWGN moving the amplitude of the carrier out of the bin associated with the correct value for that LPL bit when correct and into an adjoining bin associated with the opposite (and incorrect) value of that LPL bit.
A DCM-COFDM signal receiver using the
With regard to bit errors caused by accompanying AWGN, the successive bits of lattice-point labels for the square symbol constellations of the first set of successive NU-QAM symbols, as read in a prescribed order, have likelihoods to be in error that are complementary to the likelihoods to be in error of the successive bits of lattice-point labels for the square symbol constellations of the second set of successive NU-QAM symbols, when read in the same prescribed order. That is, with regard to a pair of corresponding bits in the two sets of NU-QAM symbols, the more likely one of those bits is to be in error because of accompanying noise of given level, the less likely the other of those bits is to be in error because of accompanying noise of the same given level,
The concurrent NU-QAM symbols in each successive pair of them are subjected to a step S7 of diversity combining that utilizes maximum-ratio combining (MRC) at bit level. Customarily, a “soft” bit of CDD is expressed as a “hard” bit accompanied by bits expressing a logarithmic likelihood ratio (LLR) of the hard bit being correct. When one of the corresponding soft bits of the LPLs of these concurrent NU-QAM symbols is less likely to be in error than the other, the soft bits of coded digital data (CDD) generated by this MRC depend more heavily on this bit than on the other bit more likely to be in error. The LLR of a soft bit resulting from MRC at bit level is adjusted downward or upward from that of the more-reliable soft bit input that would be chosen in straightforward selective combining.
If the hard-bit portions of the corresponding bits differ, there is some downward adjustment of that LLR, the LLR of the less reliable bit being differentially combined with the LLR of the more reliable bit. If the LLRs being differentially combined are close in value, the LLR of the soft bit resulting from MRC at bit level being correct is very low. Knowledge of this bit almost certainly being in error may benefit the step S8 of decoding FEC coding of the CDD. If CDD decoding involves repeated trial-and-error attempts to determine correct digital data, decoding attempts which assume the bit may be correct can be eschewed.
If the hard-bit portions of the corresponding bits are both ONE or both ZERO, there is some upward adjustment of the LLR of a soft bit resulting from MRC at bit level, the LLR of the less reliable bit being additively combined with the LRR of the more reliable bit. If the LLRs being additively combined are close in value, the LLR of the soft bit resulting from MRC bit level is essentially doubled, reducing the BER of its hard bit by 6 dB (or perhaps 2.5 dB more, as suggested by U.S. Pat. No. 7,236,548).
The coding of digital data supplied by the diversity-combining step S7 to the decoding step S8 entails, at least in part, some sort of forward-error-correction (FEC) coding.
At the time this document is filed for patenting, concatenated BCH/LDPC coding composed of Bose-Chaudhuri-Hocquenghem (BCH) outer block-coding and low-density parity-check (LDPC) inner block-coding is favored for digital television broadcasting. Typically, the coded digital signal resulting from diversity combining has bit interleaving and/or time interleaving, so is appropriately de-interleaved to generate the signal offered for step S8 decoding to recover the original digital data supplied to the DCM-COFDM signal transmitter apparatus.
Together,
A scheduler 10 for interleaving time-slices of services to be broadcast to stationary DTV receivers is depicted in the middle of
The physical layer pipe PLP0 continues following the scheduler 10, with
The physical layer pipe PLP1 continues following the scheduler 10, with
The physical layer pipe PLPn continues following the scheduler 10, with
The input stream synchronizers 2, 12, 22 etc. are operable to guarantee Constant Bit Rate (CBR) and constant end-to-end transmission delay for any input data format when there is more than one input data format. Some transmitters may omit ones of the input stream synchronizers 2, 12, 22 etc. or ones of the compensating delay units 3, 13, 23 etc.
For some Transport-Stream (TS) input signals, a large percentage of null-packets may be present in order to accommodate various bit-rate services in a constant bit-rate TS. In such case, to avoid unnecessary transmission overhead, the null-packet suppressors 4, 14, 24 etc. identify TS null-packets from the packet-identification (PID) sequences in their packet headers and remove those TS null-packets from the data streams to be scrambled by the BBFRAME scramblers 9, 19, 29 etc. This removal is done in a way such that the removed null-packets can be re-inserted in the receiver in the exact positions they originally were in, thus guaranteeing constant bit-rate and avoiding the need for updating the Program Clock Reference (PCR) or time-stamp. Further details of the operation of the input stream synchronizers 2, 12, 22 etc.; the compensating delay units 3, 13, 23 etc.; and the null-packet suppressors 4, 14, 24 etc. can be gleaned from ETSI standard EN 302 755 V1.3.1 for DVB-T2.
Conventional practice for over-the-air broadcasting of COFDM television signals without DCM has been to use 16QAM or 64QAM symbol constellations to facilitate reception by mobile DTV receivers and by DTV receivers with indoor antennas. When DCM with labeling diversity is employed, 256QAM symbol constellations have to be broadcast to achieve data throughput similar to that when 16QAM symbol constellations is used in COFDM signals without DCM. When DCM with labeling diversity is employed, 4096QAM symbol constellations have to be broadcast to achieve data throughput similar to that when 64QAM symbol constellations is used in COFDM signals without DCM.
The respective output ports of the pair 34 of NU-QAM mappers are connected for supplying first and second sets of successive NU-QAM symbol constellations to respective input ports of parsers 35 of the first and second sets of successive of NU-QAM symbols into respective successions of half COFDM symbols. These successions of half COFDM symbols are supplied to first and second input ports of a COFDM symbol assembler 36, which responds to their half COFDM symbols alternately to generate complete COFDM symbols. The output port of the COFDM symbol assembler 36 is connected to a respective input port of the assembler 20 of a serial stream of effective COFDM symbols, in which frames of PLP responses from the various physical layer pipes are time-division multiplexed. Together, the parsers 35 and the COFDM symbol assembler 36 combine to provide a COFDM symbol generator for arranging successive ones of the first set of NU-QAM symbols in first prescribed order in initial halves of successive COFDM symbols and arranging successive ones of the second set of NU-QAM symbols in second prescribed order in final halves of successive COFDM symbols.
The respective output ports of the pair 44 of NU-QAM mappers are connected for supplying first and second sets of successive NU-QAM symbol constellations to respective input ports of parsers 45 of the first and second sets of successive of NU-QAM symbols into respective successions of half COFDM symbols. These successions of half COFDM symbols are supplied to first and second input ports of a COFDM symbol assembler 46, which responds to their half COFDM symbols alternately to generate complete COFDM symbols. The output port of the COFDM symbol assembler 46 is connected to a respective input port of the assembler 20 of a serial stream of effective COFDM symbols, in which frames of PLP responses from the various physical layer pipes are time-division multiplexed. Together, the parsers 45 and the COFDM symbol assembler 46 combine to provide a COFDM symbol generator for arranging successive ones of the first set of NU-QAM symbols in first prescribed order in initial halves of successive COFDM symbols and arranging successive ones of the second set of NU-QAM symbols in second prescribed order in final halves of successive COFDM symbols.
The respective output ports of the pair 54 of NU-QAM mappers are connected for supplying first and second sets of successive NU-QAM symbol constellations to respective input ports of parsers 55 of the first and second sets of successive of NU-QAM symbols into respective successions of half COFDM symbols. These successions of half COFDM symbols are supplied to first and second input ports of a COFDM symbol assembler 46, which responds to their half COFDM symbols alternately to generate complete COFDM symbols. The output port of the COFDM symbol assembler 46 is connected to a respective input port of the assembler 20 of a serial stream of effective COFDM symbols, in which frames of PLP responses from the various physical layer pipes are time-division multiplexed. Together, the parsers 55 and the COFDM symbol assembler 46 combine to provide a COFDM symbol generator for arranging successive ones of the first set of NU-QAM symbols in first prescribed order in initial halves of successive COFDM symbols and arranging successive ones of the second set of NU-QAM symbols in second prescribed order in final halves of successive COFDM symbols.
Customarily there is a number of other physical layer pipes besides PLP0, PLP1 and PLPn, which other physical layer pipes are identified by the prefix PLP followed by respective ones of consecutive numbers two through (n−1). Each of the PLPs, n+1 in number, may differ from the others in at least one aspect. One possible difference between these n+1 PLPs concerns the natures of the FEC coding these PLPs respectively employ. The current trend is to use concatenated BCH coding and LDPC block coding for the FEC coding, but concatenated Reed-Solomon coding and convolutional coding have been used in the past. EN 302 755 V1.3.1 for DVB-T2 specifies a block size of 54,800 bits for normal FEC frames as a first alternative, and a block size of 16,200 bits is specified for short FEC frames as a second alternative. Also, a variety of different LDPC code rates are authorized. PLPs may differ in the number of OFDM carriers involved in each of their spectral samples, which affects the size of the DFT used for demodulating those OFDM carriers. Another possible difference between PLPs concerns the natures of the NU-QAM symbol constellations (or possibly other modulation symbol constellations) they respectively employ.
Clipping methods of PAPR reduction necessarily involve distortion that tends to increase bit errors and thus tax iterative soft decoding of error-correction coding more. Furthermore, the PAPR reduction method using a complementary-power pair of NU-QAM mappers suppresses occasional power peaks, which the various clipping methods of PAPR reduction rely upon to be markedly effective. Even so, most COFDM transmitter apparatus permits some clipping of power peaks that tend to occur infrequently, even where the power amplifier is of Doherty type. This is permitted in recognition of practical limitations on linearity in COFDM receiver apparatuses. However, band-limit filtering designed to suppress widening of the frequency spectrum caused by such clipping should follow the power amplifier for final-radio-frequency COFDM signal.
(The RF oscillator 66 combines with the SSB amplitude modulator 65 to constitute a generator of DCM-COFDM radio-frequency (RF) signal. Owing to arrangements of first and second sets of successive NU-QAM symbols in the frequency spectrum carried out by at least one preceding generator of COFDM symbols, the lower-frequency subband of this RF signal conveys the first set of successive NU-QAM symbols and the upper-frequency subband of this RF signal conveys a second set of successive NU-QAM symbols. The amplitude modulator 65 supplies RF analog COFDM signal from an output port thereof to the input port of a linear power amplifier 67. Linear power amplifier 67 can be of Doherty type, which type conventionally is used to reduce the likelihood of clipping on peaks of RF signal amplitude. Using DCM in accordance with the invention reduces PAPR of COFDM signals significantly, however, so a simpler type of linear power amplifier 67 may be used.
The frame preambles inserted by the frame preambles insertion unit 61 convey the conformation of each COFDM frame structure and also convey the dynamic scheduling information (DSI) produced by the scheduler 10. This information is conveyed using at least some of OFDM carriers also used for conveying the baseband OFDM information in the input signals to the frame preambles insertion unit 61. The OFDM carriers supplied by the bootstrap signal generator 63 are apt to have different frequencies than OFDM carriers used for conveying the baseband OFDM information in the input signals to the frame preambles insertion unit 61. The OFDM carriers supplied by the bootstrap signal generator 63 are constrained to a narrower bandwidth than the OFDM carriers used for conveying the baseband OFDM information in the input signals to the frame preambles insertion unit 61. The bootstrap signal conveys basic information as to the standard to which OFDM broadcasts conform, the bandwidth of the RF channel, and the size of the I-FFT used in the broadcasting of groups of OFDM frames, for example. If bootstrap signals are not used in the standard used for COFDM broadcasting, the elements 62 and 63 will be omitted, and the output port of the frame preambles insertion unit 61 will connect directly to the input port of the digital-to-analog converter 64.
The output port of the first NU-QAM mapper 71 is connected for serially supplying the complex coordinates of a first set of NU-QAM symbols to the input port of the serial-input/parallel-output register 73, which is capable of storing the complex coordinates of NU-QAM symbols for inclusion in the lower-frequency-subband half of each COFDM symbol. The output port of the second NU-QAM mapper 72 is connected for serially supplying the complex coordinates of a second set of NU-QAM symbols to the input port of the serial-input/parallel-output register 74, which is capable of storing the complex coordinates of NU-QAM symbols for inclusion in the higher-frequency-subband half of each COFDM symbol. The parallel output ports of the serial-input/parallel-output registers 73 and 74 are connected for delivering complex coordinates of respective first and second sets of NU-QAM symbols as half COFDM symbols to the parallel input ports of the parallel-input/serial-output register 75, the output port of which connects to a respective input port of the assembler 20 in
Following custom, each labeled lattice point of the NU-QAM symbol constellation maps considered in this specification and its accompanying drawing is plotted respective to an in-phase (I) axis and a quadrature (Q) axis. Each NU-QAM symbol constellation map is composed of four quadrants: a −I,+Q quadrant, a +I,+Q quadrant, a +I,−Q quadrant and a −I,−Q quadrant. In this document each of these four quadrants is considered to consist of four sub-quadrants arranged by column and row within that quadrant. An “innermost” of these sub-quadrants is closest of the four to a point of origin at which the I and Q axes cross, and an “outermost” of these sub-quadrants is furthest of the four from that point of origin. There are two “flanking” sub-quadrants in each quadrant besides the “innermost” and “outermost” sub-quadrants.
In some types of DCM-COFDM signaling the same coded digital data is conveyed by lattice-point labels (LPLs) of successive pairs of concurrently transmitted square 16NU-QAM symbol constellations. The next three paragraphs infra treat procedures that minimize the PAPR of such DCM-COFDM signaling.
In such DCM-COFDM signaling the coded digital data conveyed by the respective LPL of each of the four outermost labeled lattice points of each of the concurrently transmitted pair of square 16NU-QAM symbols is the same as the coded digital data conveyed by the LPL of a respective one of the four innermost labeled lattice points of the other of the concurrently transmitted pair of square 16NU-QAM symbols. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (2δ2+4δ+2)+2δ2=4δ2+4δ+2.
Coded digital data conveyed by the respective LPL of any of the eight flanking labeled lattice points of each of the concurrently transmitted pair of square 16NU-QAM symbols is caused to be the same as the coded digital data conveyed by the LPL of a respective one of the eight flanking labeled lattice points of the other of the concurrently transmitted pair of square 16NU-QAM symbols.
Accordingly, the combined peak powers associated with each of these eight pairs of flanking labeled lattice points has a value equal to (2δ2+2δ+1)+(2δ2+2δ+1)=4δ2+4δ+2.
The combined peak powers associated with coded digital data conveyed by DCM-COFDM signaling using square 16NU-QAM symbol constellations per
The peak power of the
Fortunately, the fact that the distance δ may be the full unit distance between labeled lattice points in the same quadrants of the 16NU-QAM signal, rather than ½ such distance, doubles the level of noise accompanying a received 16NU-QAM signal that will cause errors in the two bits of the LPLs that identify the quadrants the LPLs are positioned within. There is an overall increase of 3.5 dB in the level of noise accompanying a received 16NU-QAM signal, as compared to the level of noise accompanying a received 16QAM signal of like power, needed to cause errors in the two bits of the LPLs that identify the quadrants the LPLs are positioned within. Preferably, the two bits of the LPLs that identify the quadrants the LPLs are within in one of the two sets of successive 16NU-QAM symbols conveying the same data two-fold in a DCM-COFDM signal are positioned differently within those LPLs than the two bits of the LPLs that identify the quadrants the LPLs are positioned within in the other of the two sets of successive 16NU-QAM symbols. Maximum-ratio combining the two sets of similar coded digital data, will result in coded digital data that is generally more reliable than that retrieved from a received DCM-COFDM signal of like power that employs 16QAM of its carrier waves, rather than 16NU-QAM.
Dual mapping of digital data in DCM-COFDM signaling using the
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the four outermost labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols is the same as the coded digital data conveyed by the LPL of a respective one of the four innermost labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (2δ2+12δ+18)+2δ2=4δ2+12δ+18.
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the eight labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+6δ+9 is the same as the coded digital data conveyed by the LPL of a respective one of the eight labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols also having a peak power of 2δ2+6δ+9. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (2δ2+6δ+9)+(2δ2+6δ+9)=4δ2+12δ+18.
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the eight labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+4δ+2 is the same as the coded digital data conveyed by the LPL of a respective one of the eight labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+8δ+8. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these four pairs of labeled lattice points has a value equal to (2δ2+4δ+2)+(2δ2+8δ+8)=4δ2+12δ+10.
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the eight labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+10δ+13 is the same as the coded digital data conveyed by the LPL of a respective one of the eight labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+2δ+1. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (2δ2+10δ+13)+(2δ2+2δ+1)=4δ2+12δ+14.
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the eight labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+6δ+5 is the same as the coded digital data conveyed by the LPL of a respective one of the eight labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+6δ+5. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (2δ2+6δ+5)+(2δ2+6δ+5)=4δ2+12δ+10.
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the eight labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+4δ+4 is the same as the coded digital data conveyed by the LPL of a respective one of the eight labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+8δ+10. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (2δ2+4δ+4)+(2δ2+8δ+10) 4δ2+12δ+14.
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the eight labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+2δ+12 is the same as the coded digital data conveyed by the LPL of a respective one of the eight labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+2δ+1. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (2δ2+10δ+13)+(2δ2+2δ+1)=4δ2+12δ+14.
Consideration of the contents of the eight paragraphs immediately foregoing supports the following conclusion. There is reasonably consistent peak power in DCM-COFDM signaling in which the same coded digital data is conveyed by lattice-point labels (LPLs) of successive pairs of concurrently transmitted square 64NU-QAM symbol constellations mapped as shown in
In an SCM-mapped square NU-QAM symbol constellation with suitably positioned palindromic lattice-point labels (LPLs), two bits of the LPLs identify the quadrants that the LPLs are positioned within, and another two bits of the LPLs identify the sub-quadrants of those quadrants that the LPLs are positioned within. Palindromic LPLs are so positioned in the first through twelfth SCM maps of 64NU-QAM symbol constellations depicted in
Each pair of labeled lattice points adjacent to each other and both within any one of the sixteen sub-quadrants of the square 64NU-QAM symbol constellation depicted in
Each pair of proximate labeled lattice points in separate ones of adjoining sub-quadrants of each of the quadrants of the square 64NU-QAM symbol constellation respectively depicted in
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the four outermost labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols is the same as the coded digital data conveyed by the LPL of a respective one of the four innermost labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (18δ2+24δ+8)+2δ2=20δ2+24δ+8.
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the eight labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 10δ2+12δ+4 is the same as the coded digital data conveyed by the LPL of a respective one of the eight labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols also having a peak power of 10δ2+12δ+4. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (10δ2+12δ+4)+(10δ2+12δ+4)=20δ2+24δ+8.
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the eight labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+4δ+2 is the same as the coded digital data conveyed by the LPL of a respective one of the eight labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 18δ2+12δ+2. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these four pairs of labeled lattice points has a value equal to (2δ2+4δ+2)+(18δ2+12δ+2)=20δ2+16δ+4.
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the eight labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 18δ2+18δ+5 is the same as the coded digital data conveyed by the LPL of a respective one of the eight labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+2δ+1. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (18δ2+18δ+5)+(2δ2+2δ+1)=20δ2+20δ+6.
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the eight labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 10δ2+8δ+2 is the same as the coded digital data conveyed by the LPL of a respective one of the eight labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 10δ2+8δ+2. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (10δ2+8δ+2)+(10δ2+8δ+2)=20δ2+16δ+4.
In such DCM-COFDM signaling, the coded digital data conveyed by the respective LPL of each of the eight labeled lattice points of each of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+4δ+4 is the same as the coded digital data conveyed by the LPL of a respective one of the eight labeled lattice points of the other of the concurrently transmitted pair of square 64NU-QAM symbols having a peak power of 2δ2+8δ+10. Accordingly, the combined peak powers associated with the coded digital data conveyed by each of these eight pairs of labeled lattice points has a value equal to (2δ2+4δ+4)+(2δ2+8δ+10) 4δ2+12δ+14.
Consideration of the contents of the eight paragraphs immediately foregoing supports the following conclusion. There is reasonably consistent peak power in DCM-COFDM signaling in which the same coded digital data is conveyed by lattice-point labels (LPLs) of successive pairs of concurrently transmitted square 64NU-QAM symbol constellations mapped as shown in
The peak power of a 64NU-QAM symbol constellation with unit distance “1” between adjacent labeled lattice points only within each of its quadrants (as depicted in
The peak power of a 256NU-QAM symbol constellation with unit distance “1” between adjacent labeled lattice points only within each of its quadrants, but has twice that unit distance between adjacent labeled lattice points in different ones of adjacent quadrants. is larger by a factor of 162/152=256/225 than the peak power of a 256QAM symbol constellation with unit distance “1” between all adjacent labeled lattice points therein. Reducing the outside dimensions of the 256NU-QAM symbol constellation by a factor of 225/256 to keep its the peak power similar to that of the 256QAM symbol constellation tends to increase BER of coded digital data received via an AWGN channel. As compared with a 256QAM signal of similar power (with uniform spacing between adjacent labeled lattice points) there tends to be a 0.56 dB reduction in the level of noise accompanying a received 256NU-QAM signal, at which level the accompanying noise will cause a significant number of bits to be in error in each 256NU-QAM symbol detected at a receiver. Presuming that both carriers of the DCM-COFDM signaling that convey the same LPL are received, the gain in SNR provided by the LPLs being maximum-ratio combined will overcome the 0.56 db loss in SNR attendant upon the use of 256NU-QAM, rather than 256QAM with uniform spacing between adjacent labeled lattice points.
As compared with a 1024QAM signal of similar power (with uniform spacing between adjacent labeled lattice points) there tends to be a 0.28 dB reduction in the level of noise accompanying a received 1024NU-QAM signal, at which level the accompanying noise will cause a significant number of bits to be in error in each 1024NU-QAM symbol detected at a receiver. Presuming that both carriers of the DCM-COFDM signaling that convey the same LPL are received, the gain in SNR provided by the LPLs being maximum-ratio combined will overcome the 0.56 db loss in SNR attendant upon the use of 1024NU-QAM, rather than 1024QAM with uniform spacing between adjacent labeled lattice points.
As compared with a 4096QAM signal of similar power (with uniform spacing between adjacent labeled lattice points) there tends to be a 0.14 dB reduction in the level of noise accompanying a received 4096NU-QAM signal with doubled spacing between adjacent labeled lattice points in different ones of adjacent sub-quadrants of the symbol constellation map, at which level the accompanying noise will cause a significant number of bits to be in error in each 4096NU-QAM symbol detected at a receiver. Presuming that both carriers of the DCM-COFDM signaling that convey the same LPL are received, the gain in SNR provided by the LPLs being maximum-ratio combined will overcome the 0.56 db loss in SNR attendant upon the use of 4096NU-QAM, rather than 4096QAM with uniform spacing between adjacent labeled lattice points.
The peak power of the 64NU-QAM symbol constellation as depicted in 45A, 45B, 45C and 45D, with unit distance “1” between adjacent labeled lattice points only within each of its sub-quadrants, is larger by a factor of 100/49 than the peak power of a 64QAM symbol constellation with unit distance “1” between all adjacent labeled lattice points therein. Reducing the outside dimensions of the 64NU-QAM symbol constellation by a factor of 7/10 to keep its the peak power similar to that of the 64QAM symbol constellation tends to increase BER of coded digital data received via an AWGN channel. There tends to be a 3.10 dB reduction in the level of noise accompanying a received 64NU-QAM signal, which noise will cause a significant number of bits to be in error in the 64NU-QAM signal detected at a receiver.
As compared with a 256QAM signal of similar power (with uniform spacing between adjacent labeled lattice points) there tends to be a 1.58 dB reduction in the level of noise accompanying a received 256NU-QAM signal with doubled spacing between adjacent labeled lattice points in different ones of adjacent sub-quadrants of the symbol constellation map, at which level the accompanying noise will cause a significant number of bits to be in error in each 256NU-QAM symbol detected at a receiver. Presuming that both carriers of the DCM-COFDM signaling that convey the same LPL are received, the gain in SNR provided by the LPLs being maximum-ratio combined will overcome the 1.58 db loss in SNR attendant upon using this sort of 256NU-QAM, rather than 2566QAM with uniform spacing between adjacent labeled lattice points.
As compared with a 1024QAM signal of similar power (with uniform spacing between adjacent labeled lattice points) there tends to be a 0.80 dB reduction in the level of noise accompanying a received 1024NU-QAM signal with doubled spacing between adjacent labeled lattice points in different ones of adjacent sub-quadrants of the symbol constellation map, at which level the accompanying noise will cause a significant number of bits to be in error in each 1024NU-QAM symbol detected at a receiver. Presuming that both carriers of the DCM-COFDM signaling that convey the same LPL are received, the gain in SNR provided by the LPLs being maximum-ratio combined will overcome the 0.80 db loss in SNR attendant upon using this sort of 1024NU-QAM, rather than 1024QAM with uniform spacing between adjacent labeled lattice points.
As compared with a 4096QAM signal of similar power (with uniform spacing between adjacent labeled lattice points) there tends to be a 0.40 dB reduction in the level of noise accompanying a received 4096NU-QAM signal with doubled spacing between adjacent labeled lattice points in different ones of adjacent sub-quadrants of the symbol constellation map, at which level the accompanying noise will cause a significant number of bits to be in error in each 4096NU-QAM symbol detected at a receiver. Presuming that both carriers of the DCM-COFDM signaling that convey the same LPL are received, the gain in SNR provided by the LPLs being maximum-ratio combined will overcome the 0.40 db loss in SNR attendant upon using this sort of 4096NU-QAM, rather than 4096QAM with uniform spacing between adjacent labeled lattice points.
The foregoing descriptions of 64NU-QAM symbol constellations are offered primarily to acquaint the reader with how palindromic lattice point labels (LPLs) having more than four bits are preferably positioned in SCM maps of pairs of square NU-QAM symbol constellations governing DCM-COFDM signaling, thus to keep the PAPR of the DCM-COFDM signaling low. Arriving at such acquaintance is facilitated beginning with 64NU-QAM symbol constellations rather than with square NU-QAM symbol constellations with more than sixty-four labeled lattice points.
As noted supra in Background of the Invention, for a given type of quadrature amplitude modulation (QAM) of COFDM carrier waves, the data throughput of the COFDM signal of given full bandwidth is halved using DCM as compared to using individual carrier modulation in which coded digital data is transmitted only once. To compensate completely against such loss in data throughput, the number of labeled lattice points in NU-QAM symbols transmitted using DCM-COFDM is double the number of labeled lattice points in NU-QAM symbols transmitted using individual carrier modulation. The labeled lattice points in square 256NU-QAM symbol constellation maps use 8-bit binary-number lattice point labels (LPLs), which can implement DCM-COFDM with the same data throughput as COFDM using 16NU-QAM in individual carrier modulation. COFDM using 16QAM in individual carrier modulation has been widely used, which is one reason that DCM-COFDM using 256NU-QAM is of particular interest.
In an SCM map of a square 256NU-QAM there are 64 labeled lattice points in each quadrant and 16 lattice points in each of the four sub-quadrants within a quadrant. I. e., the quadrant considered as a bin for LPLs has eight columns and eight rows of LPLs, twice the number of columns and twice the number of rows as a sub-quadrant.
As read from left to right, the initial two bits of each lattice-point label (LPL) identify the quadrant in which the LPL is positioned. The initial two bits of an LPL being 10 indicates that it is positioned within the −I,+Q quadrant depicted in full in
The palindromic label 10000001 is applied to the lattice point located in the innermost corner of the −I,+Q quadrant in the
The palindromic label 00000000 is applied to the lattice point located in the innermost corner of the +I,+Q quadrant in the
The palindromic label 01000010 is applied to the lattice point located in the innermost corner of the +I,−Q quadrant in the
The palindromic label 11000011 is applied to the lattice point located in the innermost corner of the −I,−Q quadrant in the
In DCM-COFDM signaling, the same LPLs are concurrently transmitted via pairs of carriers. Consider DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the first SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the second SCM map of 256NU-QAM. The LPLs in the outermost sub-quadrants of the first SCM map of 256NU-QAM symbol constellations have peak powers higher than average associated with them, but corresponding LPLs in the innermost sub-quadrants of the second SCM map of 256NU-QAM have peak powers lower than average associated with them. The LPLs in the outermost sub-quadrants of the second SCM map of 256NU-QAM symbol constellations have peak powers higher than average associated with them, but corresponding LPLs in the innermost sub-quadrants of the first SCM map of 256NU-QAM have peak powers lower than average associated with them. Accordingly, peak-to-average-power ratio is constrained to be about 6 dB.
In DCM-COFDM signaling, the same LPLs are concurrently transmitted via pairs of carriers. Consider DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the second SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the third SCM map of 256NU-QAM.
In DCM-COFDM signaling, the same LPLs are concurrently transmitted via pairs of carriers. Consider DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the first SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the fourth SCM map of 256NU-QAM.
Consider DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the fifth SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the sixth SCM map of 256NU-QAM. The LPLs in the outermost sub-quadrants of the fifth SCM map of 256NU-QAM symbol constellations have peak powers higher than average associated with them, but corresponding LPLs in the innermost sub-quadrants of the sixth SCM map of 256NU-QAM have peak powers lower than average associated with them. The LPLs in the outermost sub-quadrants of the sixth SCM map of 256NU-QAM symbol constellations have peak powers higher than average associated with them, but corresponding LPLs in the innermost sub-quadrants of the fifth SCM map of 256NU-QAM have peak powers lower than average associated with them. Accordingly, peak-to-average-power ratio (PAPR) is constrained to be about 6 dB, so long as carriers concurrently conveying the same coded digital data in the DCM-COFDM signaling are both received at normal relative strengths.
Although not depicted in the drawings, there can be a seventh SCM map of 256NU-QAM symbol constellations in which map the succession of bits in each LPL mirrors the succession of bits in the correspondingly positioned LPL in the fifth SCM map of the square 256NU-QAM symbol constellation depicted in
Although not depicted in the drawings, there can be an eighth SCM map of 256NU-QAM symbol constellations wherein the succession of bits in each LPL mirrors the succession of bits in the correspondingly positioned LPL in the sixth SCM map of the square 256NU-QAM symbol constellation depicted in
Although not depicted in the drawings, there can be a ninth SCM map of 256NU-QAM symbol constellations constructed per the same general schema as the ninth SCM map of 64NU-QAM symbol constellations depicted in
Also, although not depicted in the drawings, there can be a tenth SCM map of 256NU-QAM symbol constellations constructed per the same general schema as the tenth SCM map of 64NU-QAM symbol constellations depicted in
Although not depicted in the drawings, there can be a eleventh SCM map of 256NU-QAM symbol constellations in which the order of bits in each LPL of this eleventh SCM map of mirrors the order of bits in the correspondingly positioned LPL in the ninth SCM map of square 256NU-QAM symbol constellations. Peak-to-average-power ratio is constrained to be about 6 dB DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the tenth SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the eleventh SCM map of 256NU-QAM.
Although not depicted in the drawings, there can be a twelfth SCM map of 256NU-QAM symbol constellations in which the order of bits in each LPL of this twelfth SCM map of mirrors the order of bits in the correspondingly positioned LPL in the tenth SCM map of square 256NU-QAM symbol constellations. Peak-to-average-power ratio is constrained to be about 6 dB DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the ninth SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the twelfth SCM map of 256NU-QAM.
As noted supra, the labeled lattice points in square 256NU-QAM symbol constellation maps use 8-bit binary-number lattice point labels (LPLs), which can implement DCM-COFDM with the same data throughput as COFDM using 16QAM in individual carrier modulation. The distance between labeled lattice points in the 256NU-QAM symbols used in DCM-COFDM signal is more than halved compared to the distance between labeled lattice points in the 16QAM symbols used in COFDM signal using individual carrier modulation, presuming the maximum amplitude of COFDM carrier waves to be kept the same. So, the signal-to-noise ratio (SNR) of a 256NU-QAM signal received over an AWGN reception channel will be about 6.56 dB lower than the SNR of a 16QAM signal received over an AWGN reception channel. However, supposing both of the two sets of DCM-COFDM carrier waves transmitting the same coded digital data concurrently are received over a shared AWGN reception channel with “flat” frequency response, the two sets of the same coded digital data (CDD) can subsequently be combined using bit-reliability averaging (BRA) to improve SNR in the combined (coded) digital data.
Again, consider DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the first SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the second SCM map of 256NU-QAM. Reading from left to right, the first bits of LPLs are similar within each of two eight-columns-wide portions of both the first and second SCM maps of 256NU-QAM. Reading from left to right, the second bits of LPLs are similar within each of two eight-rows-deep portions of both the first and second SCM maps of 256NU-QAM. The first and second bits of the LPLs are the bits least likely to be in error owing to AWGN accompanying the DCM-COFDM signal. Reading from left to right, the seventh bits of LPLs are similar within each of four four-columns-wide portions of both the first and second SCM maps of 256NU-QAM. Reading from left to right, the eighth bits of LPLs are similar within each of four four-rows-deep portions of both the first and second SCM maps of 256NU-QAM. Reading from left to right, the third bits of LPLs are similar within each of four-columns-wide portions of both the first and second SCM maps of 256NU-QAM, and also within two two-columns-wide portions extending to edges of those maps. Reading from left to right, the fourth bits of LPLs are similar within each of four-rows-deep portions of both the first and second SCM maps of 256NU-QAM, and also within two two-rows-deep portions extending to edges of those maps. Reading from left to right, the fifth bits of LPLs are similar within each of two-columns-wide portions of both the first and second SCM maps of 256NU-QAM, and also within two single-column-wide portions extending to edges of those maps. Reading from left to right, the sixth bits of LPLs are similar within each of two-rows-deep portions of both the first and second SCM maps of 256NU-QAM, and also within two single-row-deep portions extending to edges of those maps. The fifth and sixth bits of the LPLs are the bits most likely to be in error owing to AWGN accompanying the DCM-COFDM signal.
Maximum-ratio combining the fifth bits of the pairs of like-valued LPLs conveying the same coded digital data in the DCM-COFDM signal improves their SNR 3 dB, and maximum-ratio combining the sixth bits of the pairs of like-valued LPLs conveying the same coded digital data in the DCM-COFDM signal improves their SNR 3 dB. This is because the values of the concurrent conveyances of similar bits are correlated, whilst the respective accompanying AWGN is random. The resultant 3 dB improvement in SNR provided by bit-reliability averaging will offset the 6.56 dB or so lower SNR that the QAM symbols used in DCM-COFDM signal have as compared to the 16QAM symbols in COFDM signal using individual carrier modulation.
Also, bit-reliability averaging (BRA) can provide a 3 dB improvement in SNR of 256NU-QAM symbols recovered from DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the fifth SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the sixth SCM map of 256NU-QAM. Also, BRA can provide a 3 dB improvement in SNR of 256NU-QAM symbols recovered from DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the fifth SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the second SCM map of 256NU-QAM. Also, BRA can provide a 3 dB improvement in SNR of 256NU-QAM symbols recovered from DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the sixth SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the first SCM map of 256NU-QAM. Also, BRA can provide a 3 dB improvement in SNR of 256NU-QAM symbols recovered from DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the sixth SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the first SCM map of 256NU-QAM. Also, BRA can provide a 3 dB improvement in SNR of 256NU-QAM symbols recovered from DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the ninth SCM map of 256NU-QAM (not depicted in the drawings) and the other of that pair of carriers is modulated in accordance with the tenth SCM map of 256NU-QAM (not depicted in the drawings).
Again, consider DCM-COFDM signaling in which one of each pair of carriers is modulated in accordance with the second SCM map of 256NU-QAM, and the other of that pair of carriers is modulated in accordance with the third SCM map of 256NU-QAM. Reading from left to right, the seventh bits of LPLs are similar within each of the two eight-rows-deep portions of the third SCM map of 256NU-QAM. Reading from left to right, the eighth bits of LPLs are similar within each of the two eight-columns-wide portions of the third SCM map of 256NU-QAM. The seventh and eighth bits of the LPLs are the bits least likely to be in error owing to AWGN accompanying the DCM-COFDM signal. Reading from left to right, the first bits of LPLs are similar within each of four four-rows-deep portions of the third SCM map of 256NU-QAM. Reading from left to right, the second bits of LPLs are similar within each of four four-columns-wide portions of the third SCM map of 256NU-QAM. Reading from left to right, the third bits of LPLs are similar within four-rows-deep portions of third SCM map of 256NU-QAM, and also within two two-rows-deep portions extending to edges of those maps. Reading from left to right, the fourth bits of LPLs are similar within four-columns-wide portions of the third SCM map of 256NU-QAM, and also within two two-columns-wide portions extending to edges of those maps. Reading from left to right, the fifth bits of LPLs are similar in two-rows-deep portions of the third SCM map of 256NU-QAM, and also within two single-row-deep portions extending to edges of those maps. Reading from left to right, the sixth bits of LPLs are similar in two-columns-wide portions of of the third SCM map of 256NU-QAM, and also within two single-column-wide portions extending to edges of those maps. The fifth and sixth bits of the LPLs are the bits most likely to be in error owing to AWGN accompanying the DCM-COFDM signal.
The fifth bits of the LPLs in the third SCM map of 256NU-QAM are prone to significant error at an AWGN level half as large as the AWGN level that will cause significant error in the fifth bits of the LPLs in the second SCM map of 256NU-QAM. Maximum-ratio combining the fifth bits of a pair of like-value LPLs concurrently transmitted in a DCM-COFDM signal by respective subcarriers of similar average strength, one subcarrier modulated per the second SCM map of 256NU-QAM and the other subcarrier per the third SCM map of 256NU-QAM, provides a resultant increase in SNR of a little over 6 dB in regard to the sixth bits of the LPLs in the MRC results.
The sixth bits of the LPLs in the third SCM map of 256NU-QAM are prone to significant error at an AWGN level half as large as the AWGN level that will cause significant error in the sixth bits of the LPLs in the second SCM map of 256NU-QAM. Maximum-ratio combining the sixth bits of a pair of like-value LPLs concurrently transmitted in a DCM-COFDM signal by respective subcarriers of similar average strength, one subcarrier modulated per the second SCM map of 256NU-QAM and the other subcarrier per the third SCM map of 256NU-QAM, provides a resultant increase in SNR of a little over 6 dB insofar in regard to the sixth bits of the LPLs in the MRC results.
The MRC results obtainable in a well-designed receiver of DCM-COFDM signaling, in which signaling a first set of transmitted subcarriers is modulated per the second SCM map of 256NU-QAM and a second set of transmitted carriers is modulated per the third SCM map of 256NU-QAM, will have an SNR about the same as that obtainable from reception of COFDM signal using 16QAM of individual subcarriers. For reasons analogous to those set forth in the three paragraphs immediately preceding this one, the MRC results obtainable in a well-designed receiver of DCM-COFDM signaling, in which signaling a first set of transmitted subcarriers is modulated per the first SCM map of 256NU-QAM and a second set of transmitted carriers is modulated per the fourth SCM map of 256NU-QAM, will also have an SNR about the same as that obtainable from reception of COFDM signal using 16QAM of individual subcarriers.
Also, the MRC results obtainable in a well-designed receiver of DCM-COFDM signaling, in which signaling a first set of transmitted subcarriers is modulated per the seventh SCM map of 256NU-QAM and a second set of transmitted carriers is modulated per the sixth SCM map of 256NU-QAM, will also have an SNR about the same as that obtainable from reception of COFDM signal using 16QAM of individual subcarriers. Also, the MRC results obtainable in a well-designed receiver of DCM-COFDM signaling, in which signaling a first set of transmitted subcarriers is modulated per the eighth SCM map of 256NU-QAM and a second set of transmitted carriers is modulated per the fifth SCM map of 256NU-QAM, will also have an SNR about the same as that obtainable from reception of COFDM signal using 16QAM of individual subcarriers. Also, the MRC results obtainable in a well-designed receiver of DCM-COFDM signaling, in which signaling a first set of transmitted subcarriers is modulated per the eleventh SCM map of 256NU-QAM and a second set of transmitted carriers is modulated per the tenth SCM map of 256NU-QAM, will also have an SNR about the same as that obtainable from reception of COFDM signal using 16QAM of individual subcarriers. Also, the MRC results obtainable in a well-designed receiver of DCM-COFDM signaling, in which signaling a first set of transmitted subcarriers is modulated per the twelfth SCM map of 256NU-QAM and a second set of transmitted carriers is modulated per the ninth SCM map of 256NU-QAM, will also have an SNR about the same as that obtainable from reception of COFDM signal using 16QAM of individual subcarriers.
There is a clear advantage when the resolution of constellation space provided by the bits of LLRs in one of the two SCM maps of square NU-QAM symbol constellations used in DCM-COFDM signaling mirrors the resolution of constellation space provided by the bits of LLRs in the other of the two SCM map of square NU-QAM symbol constellations. U.S. Pat. No. 10,637,711 characterizes this sort of labeling diversity as follows. “The bits in the LPLs for the first pattern of mapping QAM symbol constellations, which bits are more likely to be in error owing to accompanying AWGN, correspond to ones of the bits in the LPLs for the second pattern of mapping QAM symbol constellations, which bits are less likely to be in error owing to accompanying AWGN. Furthermore, the bits in the LPLs for the second pattern of mapping QAM symbol constellations, which bits are more likely to be in error owing to accompanying AWGN, correspond to ones of the bits in the LPLs of the first pattern of mapping QAM symbol constellations, which are less likely to be in error owing to accompanying AWGN.” This sort of labeling diversity benefits bit-reliability averaging (BRA) in the maximum-ratio combining (MRC) of the bits of pairs of like LPLs recovered from QAM symbol constellations that are concurrently transmitted in accordance with the first and second patterns. There is significant reduction of BER of the CDD resulting from MRC performed on a bit-by-bit basis.
Such BRA technique becomes even more advantageous as the number of LPLs in the SCM map of square NU-QAM symbol constellations increases and the number of bits in each LPL increases accordingly. Presuming the transmission power to remain the same, increasing the number of LPLs in an SCM map of square NU-QAM symbol constellations together with increasing the number of bits in each LPL engenders the following results. Some of the bits of the LPLs describe columns of less width or rows of less depth in the QAM of subcarriers in the DCM-COFDM signaling. These bits are more susceptible of error caused by AWGN accompanying the DCM-COFDM signaling. So, there is increased utility in using bit-reliability averaging to combine the “soft” values of these bits recovered from a first set of subcarriers of the DCM-COFDM signaling, which have their respective QAM governed by one SCM map of square NU-QAM symbol constellations, with the “soft” values of bits with corresponding “hard” values but less probability of error. These bits with corresponding “hard” values but less probability of error are recovered from a second set of subcarriers which have their respective QAM governed by another SCM map of square NU-QAM symbol constellations. The two SCM maps of square NU-QAM symbol constellations have the sort of labeling diversity espoused by U.S. Pat. No. 10,637,711 and championed supra. Bit-reliability averaging using maximum-ratio combining on a per bit basis will allow DCM-COFDM signaling to make practical use of square 1024NU-QAM symbol constellations that have suitable labeling diversity. BRA using MRC on a per bit basis will allow DCM-COFDM signaling to make practical use of square 4096 NU-QAM symbol constellations that have suitable labeling diversity.
The central portion of the
A method of DCM-COFDM signaling, which method embodies the invention in one of its aspects, utilizes a pair of SCM maps of square NU-QAM symbol constellations both of the same size. Each of the pair of SCM maps contains 22N different labeled lattice points (LPLs), N being an integer greater than unity. Each of the 22N LPLs will have 2N bits. Amongst the 22N different LPLs expressed as binary numbers extending in horizontal direction, there will be 2N palindromic LPLS, the bits of each palindromic LPL reading the same from right to left as from left to right. An initial SCM map of square NU-QAM symbol constellations can be developed that provides a basis for developing the final pair of SCM maps to be used in a method of DCM-COFDM signaling that embodies the invention in one of its aspects.
That initial SCM map can be generated as described following. The 2N palindromic LPLS are separated into four same-size sets of LPLS, each for Gray mapping onto a diagonal of a respective innermost sub-quadrant of the four quadrants of the initial SCM map of square NU-QAM symbol constellations. A respective set of 2N-bit palindromic LPLs is Gray mapped along a diagonal axis of each of the four innermost sub-quadrants of the initial SCM map, which diagonal axis reaches from a point of origin central to the four quadrants of the initial SCM map. This point of origin is where the I and Q axes of the map cross each other.
The respective 2N-bit palindromic LPL closest to the point of origin in each quadrant will differ in just two of its bits from the 2N-bit palindromic LPLs closest to the point of origin in each of the two adjoining quadrants, which conforms with the characteristics of SCM mapping. Necessarily then, the respective 2N-bit palindromic LPL closest to the point of origin in each quadrant will differ in four of its bits from the 2N-bit palindromic LPL closest to the point of origin in the quadrant diagonally opposite across the point of origin.
Each sequence of palindromic LPLS in an innermost sub-quadrant of one of the quadrants of of the initial SCM map of square 22N NU-QAM must support Gray mapping of LPLs within that innermost sub-quadrant. The Gray maps of the four innermost sub-quadrants of the initial SCM map should support SCM map requirements of only two bits differing between adjoining LPLs in adjoining quadrants. This requirement imposes restrictions on the ordering of palindromic LPLS in adjoining sub-quadrants of those adjoining quadrants. Bits other than the pair that change between adjoining sub-quadrants need to be similarly arranged in the four innermost sub-quadrants of the initial SCM map, with regard to departure from the central point of the group of those four innermost sub-quadrants. To support SCM map requirements of only two bits differing between adjoining LPLs in adjoining quadrants, the Gray mappings of the four innermost sub-quadrants of the initial SCM map need to be twisted properly around their diagonal axes reaching from the point of origin central to the group of those four innermost sub-quadrants. The mapping of the innermost sub-quadrant of each quadrant of the initial SCM map provides a basis for mapping the other sub-quadrants of that quadrant, as detailed in the four paragraphs following, thereby to Gray map that quadrant.
The −I,+Q quadrant customarily located in the upper left corner of the initial SCM map has its innermost sub-quadrant mirrored upwards and also mirrored to its left as initial steps in generating respective flanking sub-quadrants of that −I,+Q quadrant. As an initial step in generating the outermost sub-quadrant of that −I,+Q quadrant, either its upper flanking sub-quadrant is mirrored to its left, or the left flanking sub-quadrant in that −I,+Q quadrant is mirrored upwards. (The results of the foregoing two alternatives are identical.) As the final step in generating the outermost sub-quadrant of that −I,+Q quadrant, the two bits in LPLs that identify its innermost sub-quadrant are each ones complemented so as to identify its outermost sub-quadrant. As the final step in generating the sub-quadrant flanking the outermost sub-quadrant of that −I,+Q quadrant on the right, one and only one of the two bits in LPLs that identify its innermost sub-quadrant is ones complemented, so as to identify that flanking sub-quadrant uniquely. As the final step in generating the sub-quadrant in that flanking its innermost sub-quadrant on the left, the other of the two bits in LPLs that identify its innermost sub-quadrant is ones complemented, so as to identify that flanking sub-quadrant uniquely.
The +I,+Q quadrant customarily located in the upper right corner of the initial SCM map has its innermost sub-quadrant mirrored upwards and also mirrored to its right as initial steps in generating respective flanking sub-quadrants of that +I,+Q quadrant. As an initial step in generating the outermost sub-quadrant of that +I,+Q quadrant, either its upper flanking sub-quadrant is mirrored to its right, or the right flanking sub-quadrant in that +I,+Q quadrant is mirrored upward. (The results of the foregoing two alternatives are identical.) As the final step in generating the outermost sub-quadrant of that +I,+Q quadrant, the two bits in LPLs that identify its innermost sub-quadrant are each ones complemented so as to identify its outermost sub-quadrant. As the final step in generating the sub-quadrant flanking the outermost sub-quadrant of that +I,+Q quadrant on the left, one and only one of the two bits in LPLs that identify its innermost sub-quadrant is ones complemented, so as to identify that flanking sub-quadrant uniquely. This bit is chosen so as not to cause dissimilarity in the bits of a row of bits shared with the −I,+Q quadrant. As the final step in generating the sub-quadrant in that +I,+Q quadrant flanking its innermost sub-quadrant on the right, the other of the two bits in LPLs that identify its innermost sub-quadrant is ones complemented, so as to identify that flanking sub-quadrant uniquely.
The +I,−Q quadrant customarily located in the lower right corner of the initial SCM map has its innermost sub-quadrant mirrored downwards and also mirrored to its right as initial steps in generating respective flanking sub-quadrants of that +I,−Q quadrant. As an initial step in generating the outermost sub-quadrant of that +I,−Q quadrant, either its lower flanking sub-quadrant is mirrored to its right, or its right flanking sub-quadrant is mirrored downwards. (The results of the foregoing two alternatives are identical.) As the final step in generating the outermost sub-quadrant of that +I,−Q quadrant, the two bits in LPLs that identify the innermost sub-quadrant of that +I,−Q quadrant are each ones complemented so as to identify the outermost sub-quadrant of that +I,−Q quadrant. As the final step in generating the sub-quadrant in that +I,−Q quadrant flanking its innermost sub-quadrant on the right, the other of the two bits in LPLs that identify its innermost sub-quadrant is ones complemented, so as to identify that flanking sub-quadrant uniquely. This bit is chosen so as not to cause dissimilarity in the bits of a column of bits shared with the +I,+Q quadrant. As the final step in generating the sub-quadrant flanking the outermost sub-quadrant of that +I,−Q quadrant on its left, one and only one of the two bits in LPLs that identify its innermost sub-quadrant is ones complemented, so as to identify that flanking sub-quadrant uniquely.
The −I,−Q quadrant customarily located in the lower left corner of the initial SCM map has its innermost sub-quadrant mirrored downwards and also mirrored to its left as initial steps in generating respective flanking sub-quadrants of that −I,−Q quadrant. As an initial step in generating the outermost sub-quadrant of that −I,−Q quadrant, either the lower flanking sub-quadrant in that −I,−Q quadrant is mirrored to its left, or the left flanking sub-quadrant in that −I,−Q quadrant is mirrored downwards. (The results of the foregoing two alternatives are identical.) As the final step in generating the outermost sub-quadrant of that −I,−Q quadrant, the two bits in LPLs that identify its innermost sub-quadrant are each ones complemented so as to identify its outermost sub-quadrant. As the final step in generating the sub-quadrant flanking the outermost sub-quadrant of that −I,−Q quadrant at the right, one and only one of the two bits in LPLs that identify its innermost sub-quadrant is ones complemented, so as to identify that flanking sub-quadrant uniquely. This bit is chosen so as not to cause dissimilarity in the bits of a row of bits shared with the +I,−Q quadrant. As the final step in generating the sub-quadrant in that −I,−Q quadrant flanking its innermost sub-quadrant on the left, the other of the two bits in LPLs that identify the innermost sub-quadrant of that −I,+Q quadrant is ones complemented, so as to identify that flanking sub-quadrant uniquely.
After the initial SCM map of square NU-QAM symbol constellations is developed, the following procedure can be used to continue the development of a final pair of SCM maps to be used in a method of DCM-COFDM signaling that embodies the invention. A “complementary” SCM map of square NU-QAM symbol constellations is generated with reference to the initial SCM map, which complementary SCM map can be used together with the initial SCM map in DCM=COFDM signaling having low PAPR. The LPLS arrayed in the −I,+Q quadrant of this complementary SCM map correspond to the LPLS similarly arrayed in the +I,−Q quadrant of the initial SCM map. The LPLS arrayed in the +I,+Q quadrant of this complementary SCM map correspond to the LPLS similarly arrayed in the −I,−Q quadrant of the complementary SCM map. The LPLS arrayed in the +I,−Q quadrant of this complementary SCM map correspond to the LPLS similarly arrayed in the −I,+Q quadrant of the initial SCM map. The LPLS arrayed in the −I,−Q quadrant of this complementary SCM map correspond to the LPLS similarly arrayed in the +I,+Q quadrant of the initial SCM map.
The development of a final pair of SCM maps to be used in a method of DCM-COFDM signaling that embodies the invention subsequently continues in the following manner. One of the initial and complementary SCM maps is chosen to be one of the final pair of SCM maps to be used in the method of DCM-COFDM signaling. Each of the LPLs of the other of those the initial and complementary SCM maps is mirrored to develop the other of that final pair of SCM maps. An LPL is “mirrored” by reversing the order its bits as conventionally considered from left to right.
Techniques for developing further sorts of paired SCM maps of square 64NU-QAM symbol constellations to implement methods of DCM-COFDM signaling that embody the invention are described supra, and these techniques are illustrated in
Simply stated, the front-end tuner 80 converts RF COFDM signal received at its input port to digitized samples of baseband COFDM signal supplied from its output port. Typically, the digitized samples of the real component of the baseband COFDM signal are alternated with digitized samples of the imaginary component of that baseband signal for arranging the complex baseband COFDM signal in a single stream of digital samples.
The output port of the front-end tuner 80 is connected for supplying digitized samples of baseband COFDM signal to the respective input ports of a bootstrap signal processor 83 and a cyclic prefix detector 84. The cyclic prefix detector 84 differentially combines the digitized samples of baseband COFDM signal with those samples as delayed by the duration of an effective COFDM symbol. Nulls in the difference signal so generated should occur, marking the guard intervals of the baseband COFDM signal. The nulls are processed to reduce any corruption caused by noise and to generate better-defined indications of the phasing of COFDM symbols. The output port of the cyclic prefix detector 84 is connected to supply these indications to a first of two input ports of timing synchronization apparatus 85.
A first of two output ports of the timing synchronization apparatus 85 is connected for supplying gating control signal to the control input port of a guard-interval-removal unit 86, the signal input port of which is connected for receiving digitized samples of baseband COFDM signal from the output port of the front-end tuner 80. The output port of the guard-interval-removal unit 86 is connected for supplying the input port of discrete-Fourier-transform computer 87 with windowed portions of the baseband COFDM signal that contain effective COFDM samples. A second of the output ports of the timing synchronization apparatus 85 is connected for supplying the DFT computer 87 with synchronizing information concerning the effective COFDM samples.
The indications concerning the phasing of COFDM symbols that the cyclic prefix detector 84 supplies to the timing synchronization apparatus 85 are sufficiently accurate for initial windowing of a baseband COFDM signal that the guard-interval-removal unit 86 supplies to the DFT computer 87. A first output port of the DFT computer 87 is connected for supplying demodulation results for at least all of the pilot carriers in parallel to the input port of a pilot carriers processor 88, and a second output port of the DFT computer 87 is connected for supplying demodulation results for each of the COFDM carriers to the input port of a frequency-domain channel equalizer 89. The processor 88 selects the demodulation results concerning pilot carriers for processing, part of which processing generates weighting coefficients for channel equalization filtering in the frequency domain. A first of four output ports of the processor 88 that are explicitly shown in
A second of the output ports of the pilot carriers processor 88 that are explicitly shown in
A third of the output ports of the pilot carriers processor 88 explicitly shown in
E.g., the complex digital samples from the tail of each half OFDM symbol are multiplied by the conjugates of corresponding digital samples from the cyclic prefix of the half OFDM symbol. The resulting products are summed and low-pass filtered to develop the AFPC signal that the AFPC generator 82 supplies to the front-end tuner 80 for controlling the final local oscillator therein. This method is a variant of a known method to develop AFPC signals in receivers for COFDM signals described in U.S. Pat. No. 5,687,165 titled “Transmission system and receiver for orthogonal frequency-division multiplexing signals, having a frequency-synchronization circuit”, which was granted to Flavio Daffara and Ottavio Adami on 11 Nov. 1997.
The DFT computer 87 is configured so it can demodulate any one of 8K, 16K and 32K options as to the number of OFDM carriers. The correct option is chosen responsive to an instruction from a controller 90 that generates a number of instructions used to configure the COFDM receiver to suit the broadcast standard used transmissions currently received. To keep the drawings from being too cluttered to be easily understood, they do not explicitly illustrate the multitudinous connections from the controller 90 to the elements of the receiver controlled by respective instructions from the controller 90.
As noted supra, the second output port of the DFT computer 87 is connected to supply demodulated complex digital samples of the complex coordinates of NU-QAM symbol constellations in parallel to the input port of the frequency-domain channel equalizer 89. To implement a simple form of frequency-domain channel equalization, the pilot carriers processor 88 measures the amplitudes of the demodulated pilot carriers to determine basic weighting coefficients for various portions of the frequency spectrum. The pilot carriers processor 88 then interpolates among the basic weighting coefficients to generate respective weighting coefficients supplied to the frequency-domain channel equalizer 89 with which to multiply the complex coordinates of NU-QAM symbol constellations supplied from the DFT computer 87. Various alternative types of frequency-domain channel equalizer are also known.
An extractor 91 of COFDM frame preambles selects them from COFDM frames of decoded data supplied from a decoder 106 for BCH coding, which decoder 106 is depicted in
The controller 90 is connected for responding to elements of the bootstrap signal forwarded to a first of its input ports from an output port of the bootstrap signal processor 83. The controller 90 supplies COFDM data frame information to the pilot carriers processor 88, which data frame information can be generated responsive to baseband bootstrap signal that the bootstrap signal processor 88 supplies to the controller 90. Since the bootstrap signal is not always received acceptably free of error, it is good design to provide a source alternative to the bootstrap signal processor 83 for supplying the controller 90 back-up information as to the nature of received DTV signal. Such a source is necessary if bootstrap signal is not transmitted or if the receiver does not include a bootstrap signal processor. Accordingly the response of a decoder 106 for BCH coding, which decoder 106 is depicted in
Responsive to information supplied from the bootstrap signal processor 83 or from the processor 92 of COFDM frame preambles, the controller 90 prescribes the basic sample rate and the size of I-FFT that the controller 90 instructs the DFT computer 87 to use in its operation regarding DTV signal. The controller 90 instructs the channel equalizer 89 and the banks 93 and 65 of parallel-input/serial-output converters to configure themselves to suit the size of DFT that the controller 90 instructs the DFT computer 87 to generate.
The frequency-domain channel equalizer 89 is connected for supplying complex coordinates of the NU-QAM symbol constellations from the lower-frequency half COFDM symbol, in parallel and in forward spectral order, to the bank 93 of parallel-to-serial (P/S) converters. One of these P/S converters as selected by the controller 90 supplies the complex coordinates of a first set of NU-QAM symbol constellations extracted from the lower-frequency halves of successive COFDM symbols. The frequency-domain channel equalizer 89 is further connected for supplying complex coordinates of the NU-QAM symbol constellations from the higher-frequency half COFDM symbol, in parallel and in forward spectral order, to the bank 94 of parallel-to-serial (P/S) converters. One of these P/S converters as selected by the controller 90 supplies the complex coordinates of a second set of NU-QAM symbol constellations extracted from the higher-frequency halves of successive COFDM symbols. “Forward spectral order” refers to the complex coordinates of each successive NU-QAM symbol constellation from a half COFDM symbol having been conveyed by the COFDM carrier next higher in frequency than that having conveyed its predecessor NU-QAM symbol. Each of the banks 93 and 65 of P/S converters comprises respective P/S converters that are appropriate for half the number of OFDM carriers that can convey data in a COFDM symbol of prescribed size. The pair of P/S converters selected for current reception is determined by a control signal that the controller 90 supplies in common to each of the banks 93 and 65 of P/S converters.
The first sets of NU-QAM symbol constellations are those that originate from the first mapping procedures in the COFDM transmitter apparatus and are supplied from the output port of the bank 93 of P/S converters to the input port of a bank 95 of demappers for the first sets of NU-QAM symbol constellations, as depicted in
The pairs of NU-QAM demappers in the banks 95 and 96 of demappers could be paired Gray demappers, paired SCM demappers, paired natural demappers, paired anti-Gray demappers, paired “optimal” demappers of various types or some mixture of those types of paired demappers. However, if the demapping results from the antiphase-energy NU-QAM demappers are to be maximal-ratio combined at bit level to improve effective SNR for AWGN reception, it is strongly recommended that NU-QAM symbol constellations be Gray mapped or SCM mapped. SCM mapping is preferred since it better lends itself to reducing PAPR of DCM-COFDM signals using square NU-QAM symbol constellations than does Gray mapping. In one of its aspects, the invention is embodied in a DCM COFDM signal receiver that includes paired SCM demappers for square NU-QAM symbols concurrently conveying similar coded digital data and having labeling diversity designed for keeping PAPR low. It is practical for each of the NU-QAM demappers to constitute a plurality of read-only memories (ROMs), one for each bit of map labeling, addressed by the complex coordinates descriptive of the current NU-QAM symbol. Each ROM is read to provide a “hard” bit followed by a confidence factor indicating how likely that bit is to be correct. Customarily these confidence factors are expressed as logarithm of likelihood ratios (LLRs).
The confidence factors are usually based, at least in substantial part, on judgments of the distance of the complex coordinates descriptive of the current NU-QAM symbol from the edges of the bin containing the “hard” bit. The confidence factors can be further based on whether or not the bin containing the “hard” bit is at an edge of the current NU-QAM symbol constellation and, if so, whether the complex coordinates descriptive of that current NU-QAM symbol closely approach that edge or even pass beyond it. The confidence factor that the “hard” bit is correct is increased if the complex coordinates descriptive of that current NU-QAM symbol closely approach a symbol constellation edge or even pass beyond it. This increase applies to all bits in the map label. This effect obtains if mapping of NU-QAM symbol constellations is Gray mapping or is SCM mapping.
Not all COFDM communication systems will concatenate BCH coding and LDPC coding. Cyclic redundancy check (CRC) coding can be used instead of BCH coding for detecting the successful conclusion of LDPC decoding. In such case, the general structure of COFDM receiver apparatus depicted in
The read-output port of the dual-port RAM 118 is further connected for supplying a posteriori soft demapping results to the minuend-input port of the digital subtractor 102. The subtrahend-input port of the digital subtractor 102 is connected for receiving the bit-interleaved extrinsic error signal from the output port of the interleaver 105 for extrinsic “soft” bits. The difference output port of the digital subtractor 102 connects to the input port of the de-interleaver 103 for bit-interleaved soft bits. The output port of the de-interleaver 103 connects to the input port of the soft-input/soft-output (SISO) decoder 101 for LDPC coding and further connects to the subtrahend input port of the digital subtractor 104. The minuend input port of the subtractor 104 is connected to receive the soft bits of decoding results from the output port of the SISO decoder 101. The subtractor 104 generates soft extrinsic data bits by comparing the soft output bits supplied from the SISO decoder 101 with soft input bits supplied to the SISO decoder 101. The output port of the subtractor 104 is connected to supply these soft extrinsic data bits to the input port of the bit-interleaver 105, which is complementary to the de-interleaver 103. The output port of the bit-interleaver 105 is connected for feeding back bit-interleaved soft extrinsic data bits to the second addend-input port of the digital adder 119, therein to be additively combined with previous a posteriori soft demapping results read from the dual-port RAM 118 to generate updated a priori soft demapping results to write over the previous ones temporarily stored within that memory 118.
More specifically, the RAM 118 is read concurrently with memory within the bit-interleaver 105, and the soft bits read out in LLR form from the memory 118 are supplied to the first input port of the digital adder 119. The adder 119 adds the interleaved soft extrinsic bits fed back via the interleaver 105 to respective ones of the soft bits of a posteriori soft demapping results read from the RAM 118 to generate updated a priori soft demapping results supplied from the sum output port of the adder 119 to the write-input port of the RAM 118 via the write signal multiplexer 117. The soft bits of previous a posteriori demapping results temporarily stored in the RAM 118 are each written over after its being read and before another soft bit is read.
The output port of the bit-interleaver 105 is also further connected for feeding back bit-interleaved soft extrinsic data bits to the subtrahend input port of the subtractor 102. The subtractor 102 differentially combines the bit-interleaved soft extrinsic data bits fed back to it with respective ones of soft bits of the a posteriori demapping results read from the RAM 118, to generate soft extrinsic data bits for the adaptive soft demapper from the difference-output port of the subtractor 102 for application to the input port of the de-interleaver 103. As thus far described, the SISO decoder 101 and the adaptive soft demapper (comprising elements 97, 98 and 117-119) are in a turbo loop connection with each other, and the turbo cycle of demapping NU-QAM constellations and decoding LDPC can be iterated many times to reduce bit errors in the BCH coding that the SISO decoder 101 finally supplies from its output port to the input port of the decoder 106 for BCH coding. Successful correction of BCH codewords can be used for terminating iterative demapping and decoding of LDPC coding after fewer turbo cycles than the maximum number permitted.
Each of the banks 95 and 96 of demappers of NU-QAM symbols comprises a plurality of read-only memories (ROMs), one ROM for each bit of a particular size of NU-QAM map label, which ROMs each receive as input address thereto the complex coordinates descriptive of a current one of a succession of NU-QAM symbols. Each ROM considers the NU-QAM modulation to range over a square arrangement of square “bins”, each of which bins has a respective map label associated therewith. Each ROM generates a respective “soft” bit, a bit metric composed of the more likely one of the “hard” bits 1 and 0 accompanied by a confidence factor. Customarily, the confidence factor is expressed in digitized numerical form as a logarithm of likelihood ratio (LLR) indicating how likely the accompanying decision as to the “hard” bit is correct. The soft-bit maximal-ratio combiner 971 considers 1 and 0 “hard” bits as sign bits when combining the LLRs of each successive pair of “soft” bits in a signed addition. The sign bit of the resultant sum determines the “hard” bit in the “soft” bit response from the maximal-ratio combiner 971 and the rest of this resultant sum determines the LLR of the correctness of this “hard” bit in the “soft” bit response from the maximal-ratio combiner 971.
Each ROM in a demapper of NU-QAM symbols, which ROM is associated with a particular bit of map labeling, can support soft-bit maximal-ratio combining (SBMRC) in the following manner. When the result from demodulating the NU-QAM modulation addresses the center point of the square bin identified by a particular map label, LLR of the particular bit is a value associated with a high level of confidence that the bit is correct. The LLR of the particular bit is reduced from that value when the result from demodulating NU-QAM modulation addresses a point in that square bin approaching a boundary between that square bin and an adjoining square bin associated with opposite hard-bit value. When such boundary is reached, the level of confidence in the particular bit being correct is reduced to no more than half its level at the center point of the bin. The level of confidence in the particular bit being correct at the center point of a bin increases proportionally to bin size.
Maximal-ratio combining of frequency-diverse NU-QAM signals is superior to other well-known types of diversity combining when those signals are afflicted by AWGN, atmospheric noise, Johnson noise within the receiver, or imperfect filtering of power from an alternating-current power source. However, maximal-ratio combining of frequency-diverse NU-QAM signals performs less satisfactorily when one NU-QAM signal is corrupted by burst noise or in-channel interfering signal and the other is not. These various conditions of unsatisfactory reception will cause errors in the reproduction of soft bits of FEC-coded data from the maximal-ratio combiner 971. The erroneous bits are dispersed by the NU-QAM map label de-interleaver 98 and by a de-interleaver of soft “bits” within the iterative SISO decoder 100 for LDPC coding, which improves the chances for those erroneous bits to be corrected during the decoding of the forward-error-correction (FEC) coding by the decoders 100 and 106.
When dual NU-QAM mapping procedures are applied to a COFDM signal, so its frequency spectrum is as illustrated in
More particularly, the NU-QAM symbols that the DFT computer 87 extracts from the lower-frequency subband of the DCM-COFDM signal are supplied, in parallel and in forward spectral order, directly to the parallel input ports of the selected one of the P/S converters in the bank 93 of them. The output port of that selected P/S converter responds to supply serialized NU-QAM symbols from the lower-frequency subband of the DCM-COFDM signal to the multiplicand input port of the multiplier 891. The parallel input ports of a selected one of the parallel-to-serial (P/S) converters in a bank 931 of them receives, in parallel from the pilot carriers processor 88, the weighting coefficients for frequency-domain channel equalization of the lower-frequency subband of the DCM-COFDM signal. The output port of that selected P/S converter responds to supply serialized weighting coefficients for the lower-frequency subband of the DCM-COFDM signal to the multiplier input port of the complex-number multiplier 891. The multiplier 891 responds to its multiplicand and multiplier input signals to supply from its product output port an equalized first set of NU-QAM symbols, suitable for subsequent demapping.
More particularly, the NU-QAM symbols that the DFT computer 87 extracts from the higher-frequency subband of the DCM-COFDM signal are supplied, in parallel and in forward spectral order, directly to the parallel input ports of the selected one of the P/S converters in the bank 94 of them. The output port of that selected P/S converter responds to supply serialized NU-QAM symbols from the higher-frequency subband of the DCM-COFDM signal to the multiplicand input port of the multiplier 892. The parallel input ports of a selected one of the parallel-to-serial (P/S) converters in a bank 941 of them receives, in parallel from the pilot carriers processor 88, the weighting coefficients for frequency-domain channel equalization of the higher-frequency subband of the DCM-COFDM signal. The output port of that selected P/S converter responds to supply serialized weighting coefficients for the lower-frequency subband of the DCM-COFDM signal to the multiplier input port of the complex-number multiplier 892. The multiplier 892 responds to its multiplicand and multiplier input signals to supply from its product output port an equalized second set of NU-QAM symbols, suitable for subsequent demapping.
A first of the pair 134 of NU-QAM mappers supplies a first stream of complex coordinates of NU-QAM symbols to a serial-input/parallel-output register 135. The SIPO register 135 parses the NU-QAM symbols into effective half COFDM symbols, arranging the NU-QAM symbols therein in a first spectral order following a cyclic prefix. The parallel output ports of the SIPO register 135 are connected to the parallel input ports of a pilot-carrier symbols insertion unit 136, which introduces pilot symbols for the lower- and upper-frequency edges of the complete half COFDM symbol and introduces pilot carrier symbols at suitable intervals between NU-QAM symbols in each effective half COFDM symbol to generate the rest of a respective complete half COFDM symbol. The parallel output ports of the pilot-carrier symbols insertion unit 136 are connected to the parallel input ports of an OFDM modulator 137 for lower-frequency-subband OFDM carriers. The OFDM modulator 137 performs an I-FFT and supplies the results from its output port as amplitude-modulating signal to the modulating-signal input port of a downward amplitude modulator 138, there to modulate radio-frequency carrier supplied from the output port of a radio-frequency oscillator 140 to a principal-carrier input port of the SSB amplitude modulator 138.
A second of the pair 134 of NU-QAM mappers supplies a second stream of complex coordinates of NU-QAM symbols to a serial-input/parallel-output register 145. The SIPO register 145 parses the NU-QAM symbols into effective half COFDM symbols, arranging the NU-QAM symbols therein in a second spectral order following a cyclic prefix. The parallel output ports of the SIPO register 145 are connected to the parallel input ports of a pilot-carrier symbols insertion unit 146, which introduces pilot symbols for the lower- and upper-frequency edges of the complete half COFDM symbol and introduces pilot carrier symbols at suitable intervals between NU-QAM symbols in each effective half COFDM symbol to generate the rest of a respective complete half COFDM symbol. The parallel output ports of the pilot insertion unit 146 are connected to the parallel input ports of an OFDM modulator 147 for higher-frequency-subband OFDM carriers. The OFDM modulator 147 performs an I-FFT and supplies the results from its output port as amplitude-modulating signal to the modulating-signal input port of an upward amplitude modulator 148, there to modulate radio-frequency carrier supplied from the output port of the RF oscillator 140 to a principal-carrier input port of the SSB amplitude modulator 148.
The pilot-carrier symbols insertion units 136 and 146 combine with the SIPO registers 135 and 145 so as to constitute a COFDM symbol generator for supplying respective halves of COFDM symbols to the OFDM modulators 137 and 147, which halves of COFDM symbols are respectively responsive to first and second sets of NU-QAM symbols supplied from respective ones of the pair 134 of NU-QAM mappers. First and second input ports of a radio-frequency signal combiner 150 are respectively connected for receiving the lower-frequency SSB amplitude-modulated RF signal from the output port of the amplitude modulator 138 and for receiving the upper-frequency SSB amplitude-modulated RF signal from the output port of the amplitude modulator 148. The RF oscillator 140, SSB amplitude modulator 138, SSB amplitude modulator 148 and RF signal combiner 150 combine to constitute a generator of DCM-COFDM radio-frequency signal. Owing to arrangements of first and second sets of successive NU-QAM symbols in the frequency spectrum carried out by the preceding generator of COFDM symbols, the lower-frequency subband of this RF signal conveys the first set of successive NU-QAM symbols and the upper-frequency subband of this RF signal conveys a second set of successive NU-QAM symbols.” The output port of the RF signal combiner 150 is connected for supplying ISB signal to the input port of the linear power amplifier 67, which may be of Doherty type but need not be. The output port of the linear power amplifier 67 is connected for driving RF analog COFDM signal power to the transmission antenna 68. The effective COFDM symbols are caused to have spectral response as shown in
U.S. Pat. No. 10,171,280 titled “Double-sideband COFDM signal receivers that demodulate unfolded frequency spectrum” issued 1 Jan. 2019 to A. L. R. Limberg based on an application filed 3 Jul. 2017. This patent alludes to its inventor's previous design for a COFDM signal receiver, which employed a beat-frequency oscillator (BFO) supplying in-phase (I) and quadrature-phase (Q) beat-frequency oscillations to the respective carrier input ports of analog mixers via a direct connection and via a −90° phase-shifter, respectively. As pointed out in the patent, this previous design is problematic in the following two respects. It is difficult to realize a phase-shifter with analog circuitry, which phase-shifter provides exact −90° phase shift despite change in BFO frequency. Also, maintaining the amplitudes of the beat-frequency oscillations to the respective carrier input ports of the two analog mixers the same is rather difficult.
The latter of these difficulties is avoided by mixers 201 and 202 being of switching type receiving I and Q square waves at their respective carrier input ports. Fundamental-frequency components of the I and Q square waves that are at quite exactly at 0° and −90° respective phasings, despite change in frequency, are supplied from a 2-phase divide-by-4 frequency divider 203 in response to rising edges of pulses from a clock oscillator 204. The frequency divider 203 can be constructed from two gated D flip flop-flops (or data latches) suitably connected as depicted in
An analog-to-digital converter 205 performs analog-to-digital conversion of baseband signal supplied from the output port of the mixer 201. The sampling of the mixer 201 output signal by the A-to-D converter 205 is timed by a first set of alternate clock pulses received from the clock oscillator 204. An analog-to-digital converter 206 performs analog-to-digital conversion of the baseband signal supplied from the output port of the mixer 202. The sampling of the mixer 202 output signal by the A-to-D converter 206 is timed by a second set of alternate clock pulses received from the clock oscillator 204. The digitized in-phase baseband signal supplied from the output port of the A-to-D converter 205 is supplied to the input port of a digital lowpass filter 207. The digitized quadrature-phase baseband signal supplied from the output port of the A-to-D converter 206 is supplied to the input port of a digital lowpass filter 208. The digital lowpass filters 207 and 208 are of similar design, each to supply a response to a respective subband which response is free of components of image signal remnant from the synchrodyning procedures. Preferably, that is, the design of the digital lowpass filters 207 and 208 provides a rapid roll-off of their higher-frequency responses, so as to suppress adjacent-channel interference (ACI).
The response of the digital lowpass filter 208 to quadrature-phase baseband signal is supplied to the input port of a finite-impulse-response digital filter 209 for Hilbert transformation. The response of the digital lowpass filter 207 to in-phase baseband signal is supplied to the input port of a clocked digital delay line 210 that affords delay to compensate for the latent delay through the FIR filter 209. The Hilbert transform response of the FIR filter 209 and the response of the digital delay line 210 are supplied to respective addend input ports of a digital adder 211 operative to recover, at baseband, the lower-frequency subband of the DCM-COFDM signal at its sum output port. The Hilbert transform response of the FIR filter 209 and the response of the digital delay line 210 are supplied respectively to the minuend input port and the subtrahend input port of a digital subtractor 212 operative to recover, at baseband, the higher-frequency subband of the DCM-COFDM signal at its difference output port.
The sum output port of the digital adder 211 connects to the input port of a guard interval remover 861. The output port of the guard interval remover 861 is connected for supplying the input port of a discrete-Fourier-transform (DFT) computer 871 with windowed portions of the baseband digitized lower-frequency subband of the DCM-COFDM signal that span respective COFDM symbol intervals. The complex coordinates of NU-QAM symbols the DFT computer 871 extracts from lower-frequency subband carriers in each COFDM symbol sampling interval that convey coded data are supplied as parallel input signal to a frequency-domain channel equalizer 893 for just those NU-QAM symbols connected
for supplying equalized NU-QAM symbols to the parallel inputs of the P/S converter 93 in the
Subsequent to the recovery of the digitized higher-frequency subband of the DCM-COFDM signal at baseband by phase shift method, it is supplied from the difference output port of the digital subtractor 212 to the input port of a guard interval remover 862. The output port of the guard interval remover 862 is connected for supplying the input port of a DFT computer 872 with windowed portions of the baseband digitized higher-frequency subband of the DCM-COFDM signal that span respective COFDM symbol intervals. The complex coordinates of NU-QAM symbols the DFT computer 872 extracts from higher-frequency subband carriers in each COFDM symbol sampling interval that convey coded data are supplied as parallel input signal to a frequency-domain channel equalizer 894 for just those NU-QAM symbols. Parallel output ports of the channel equalizer 894 are connected for supplying equalized NU-QAM symbols to the parallel inputs of the P/S converter 94 in the
The DFT computers 871 and 872 are similar in construction, each configured so it can demodulate any one of 4K, 8K or 16K options as to half the nominal number of OFDM carriers. The correct option is chosen responsive to an instruction from a controller 90 that generates a number of instructions used to configure the COFDM receiver to suit the broadcast standard used transmissions currently received. The bootstrap signal processor 83, the controller 90, the extractor 91 of FEC frame preambles, and the processor 92 of COFDM frame preambles are not explicitly depicted in any of the
The guard interval removers 861 and 862 are each constructed similarly to the guard interval remover 86 in the
the digital subtractor 212. Alternatively, the input port of the cyclic prefix detector 84 can instead be connected for detecting the occurrences of cyclic prefixes in the digitized lower-frequency subband of the DCM-COFDM signal supplied at baseband from the output port of the digital adder 211. The cyclic prefix detector 84 differentially combines the digitized samples of baseband COFDM signal with those samples as delayed by the duration of an effective COFDM symbol. Nulls in the difference signal so generated should occur, marking the guard intervals of the baseband COFDM signal.
The nulls are processed to reduce any corruption caused by noise and to generate better-defined indications of the phasing of COFDM symbols. The output port of the cyclic prefix detector 84 is connected to supply these indications to a first of two input ports of timing synchronization apparatus 285. First and second output ports of the timing synchronization apparatus 285 are connected for supplying similar gating control signals to the control input ports of the guard interval removers 861 and 862. Third and fourth output ports of the timing synchronization apparatus 285 are connected for supplying indications of the phasing of COFDM symbols to the DFT computers 871 and 872 respectively.
The complex coordinates of NU-QAM symbols extracted from pilot carriers in each COFDM symbol sampling interval are supplied as parallel input signal to a pilot carriers processor 288. The pilot carriers processor 288 responds to complex coordinates of NU-QAM symbols extracted from lower-frequency-subband pilot carriers to generate weighting coefficients for the frequency-domain channel equalizer 893 to apply to NU-QAM symbols extracted from the higher-frequency subband of the
DCM-COFDM signal. A first of five output ports of the processor 288 that are explicitly shown in
A third of the output ports of the pilot carriers processor 288 that are explicitly shown in
A fifth of the output ports of the pilot carriers processor 288 explicitly shown in
In
The PLL frequency synthesizer 301 includes a programmable frequency divider, a clocked counter that counts the first local oscillations supplied to its counter input connection from the VCO 303. When the count reaches a selected large positive integer, the counter resets to zero count and generates a carry pulse supplied to an AFPC detector within the PLL frequency synthesizer 301. Responsive to carry pulses from the counter, that AFPC detector samples the 1 MHz oscillations from the crystal oscillator and integrates the response of such sampling to generate the AFPC voltage applied to the VCO 303. The crystal oscillator 300 is designed for supplying 1 MHz reference oscillations since it is the largest common submultiple of the central carrier frequencies of all the allocated TV broadcast channels in the U.S.A.
The PLL frequency synthesizer 302 includes a fixed frequency divider, a clocked counter that counts the second local oscillations supplied to its counter input connection from the VCO 304. When the count reaches a prescribed large positive integer, the counter resets to zero count and generates a carry pulse supplied to an AFPC detector within the PLL frequency synthesizer 302. Responsive to carry pulses from the counter, that AFPC detector samples the 1 MHz oscillations from the crystal oscillator and integrates the response of such sampling to generate the AFPC voltage applied to the VCO 304. Choosing the prescribed large positive integer at which the counter in the PLL frequency synthesizer 302 resets to zero count is preferably done so as to position the central carrier frequency of the second-IF DCM-COFDM signal at 11 MHz. This frequency is low enough that analog-to-digital conversion of the second-IF DCM-COFDM signal is practical. Also, the fourth harmonic of the central carrier frequency of the second-IF signal is at 44 MHZ, which is at the center of the 41-47 megahertz final IF signals commonly used in prior-art television receivers. Since these frequencies are not allocated for high-power RF transmissions, this reduces the possibility of strong interference with operation of the clock oscillator 204 depicted in
The input port of a pre-filter 305 is connected for receiving radio-frequency (RF) COFDM signal supplied by an antenna or a cable distribution system. (The pre-filter 305 is typically constructed either as a group of fixed frequency band pass filters, or as a tracking type of filter.) The pre-filter 305 reduces the bandwidth of the signal entering the subsequent radio-frequency amplifier 306, which RF amplifier 306 is subject to automatic gain control (AGC). The pre-filter 305 reduces the number of channels amplified by the AGC'd RF amplifier 306, thereby reducing the intermodulation interference generated by the amplifier 306 and subsequent circuits. In a pre-filter 305 comprising a group of fixed-frequency bandpass filters, the proper band is selected according to channel selection information supplied from a controller not explicitly depicted in
The RF output of the pre-filter 305 is amplified or attenuated to a desired level by the AGC'd RF amplifier 306 and then supplied to a first mixer 307, there to be mixed with first local oscillations from the VCO 303. The signal at the output port of first mixer 307, resulting from the desired TV channel signal being multiplied by the VCO 303 oscillations, is defined as the first intermediate frequency signal. The frequency of this first-IF signal is the difference between the frequency of the VCO 303 first local oscillations and the frequency of the DCM-COFDM signal to be received. Since the mixer 307 shifts the spectrum of the desired TV channel to a frequency higher than the TV broadcast frequency, this operation is referred to as an up-conversion. The first-IF is chosen to be above all of the spectrum used by terrestrial or cable distribution TV broadcasting in the particular environment in which the tuner operates in. By this choice, the image frequency (the frequency which is the numerical sum of the VCO 303 signal and the first-IF frequency) generated in the up-conversion process can be rejected by the pre-filter 305. This choice of first intermediate frequencies also requires the frequency of the VCO 304 to be above the spectrum used by TV broadcasting, thereby avoiding other possible interference.
The first-IF output signal supplied from the mixer 307 is amplified by a narrow-band amplifier 308 and then supplied to a first-IF bandpass filter 309 such as a dielectric resonance filter, a strip-line filter or a SAW filter. The characteristics of the first-IF BPF 309 are designed, with consideration to the characteristics of subsequent digital filtering that will be used to suppress ACI (adjacent-channel interference). I.e., the bandwidth of the first-IF BPF 309 is no less than that of a single digital TV channel, and the passband group delay response is sufficiently linear so as not to cause adverse effects on subsequent demodulation of a second-intermediate-frequency (second-IF) DCM-COFDM signal. Furthermore, the first-IF BPF 309 is designed to have sufficient out-of-band attenuation at the image frequency range of the subsequent down-conversion process by a second mixer 310 so as not to introduce excessive image frequency interference to degrade the performance of the subsequent demodulation of the second-IF DCM-COFDM signal. (In alternative front-end tuner designs the positions of the first-IF amplifier 308 and the first-IF BPF 309 within their cascade connection are interchanged.)
The output signal from the first-IF BPF 309 principally consists of just the desired TV channel signal as up-converted, possibly accompanied by small amounts of up-converted adjacent-channel signals that have not been completely attenuated owing to the band-edge roll-off characteristics of BPF 309. This signal is supplied to a second mixer 310 to be mixed with second local oscillations, which are supplied from the VCO 304. The signal supplied from the output port of the mixer 310, resulting from the first-IF DCM-COFDM signal being multiplied by second local oscillations from the VCO 304, is defined as the second-intermediate-frequency (second-IF) DCM-COFDM signal. The frequency of this second-IF DCM-COFDM signal is the numerical difference between the frequency of second local oscillations from the VCO 304 and the somewhat lower frequencies of the first-IF DCM-COFDM signal. The second-IF DCM-COFDM signal supplied from the output port of the mixer 310 is amplified by a second IF amplifier 311 of such design as to suppress image signals that have frequencies almost twice that of the frequency of the second local oscillations above the UHF TV band. Since the mixer 310 shifts the first-IF signal to a lower frequency, this operation is referred to as a down-conversion.
The amplified second-IF DCM-COFDM signal supplied from the output port of the second IF amplifier 311 is applied to the input port of pseudo-RMS detection circuitry 312. The output port of the pseudo-RMS detection circuitry 312 is connected for supplying an approximation of the RMS (root-mean-square) voltage of the response from the second IF amplifier 311 to a first input port of circuitry 313 for generating respective automatic gain control (AGC) signals for the RF amplifier 306 and for the first-IF amplifier 308. The peak-to-average ratio (PAPR) of COFDM signals is very high, and occasional peak clipping of them is better design. Detecting the peak voltage of the response from the second-IF amplifier 311 would not provide a good basis from which to develop AGC signals.
A second port of the circuitry 313 for generating AGC signals is connected for receiving pilot carrier amplitude information from the pilot carriers processor 288 depicted in
Designs of circuitry for generating AGC signals in double-conversion radio receivers are known in the prior art. The circuitry 313 generates delayed AGC signal for the RF amplifier 306, avoiding reduction of the RF amplifier 306 gain as long as RF signal strength is not so strong that RF amplifier 306 response consistently drives the first mixer 307 outside its range of acceptably linear response. During the reception of such weaker strength RF signals, the circuitry 313 generates AGC signal for the first-IF amplifier 308 that regulates its gain control to maintain desired value of the approximate RMS value of the second IF amplifier 311 response. This maintains the second mixer 310 within its range of acceptably linear response. The circuitry 313 generates the delayed AGC signal for the RF amplifier 306 so as to exhibit slower response to second IF amplifier 311 output signal than the AGC signal for the first-IF amplifier 308. This accommodates clipping of occasional extraordinarily large peaks of received COFDM signal in the first mixer 307 and the RF amplifier 306. The AGC signal for the first-IF amplifier 308 that circuitry 313 generates no longer reduces the gain of the first-IF amplifier 308 when circuitry 313 supplies delayed-AGC signal to the RF amplifier 306 for reducing its gain.
In a front-end tuner 280 configuration as used in
The amplified second-IF DCM-COFDM signal supplied from the output port of the second IF amplifier 311 is suitable to provide the intermediate-frequency DCM-COFDM output signal for a front-end tuner 180 configuration. In such front-end tuner 180 configuration the A-to-D converter 314 and the digital bandpass filter 315 are unnecessary and can be omitted.
The leading I square wave that the frequency divider 203 supplies to the control input port of the +1, (−1) multiplier 213 conditions the +1, (−1) multiplier 213 to perform a 2-to-1 decimation of the 0°, 90°, 1800 and 2700 digital samples of DCM-COFDM signal supplied to its input port, selecting the 0° digital samples for multiplication by +1 responsive to positive half cycles of I square wave, and selecting the 1800 digital samples for multiplication by −1 responsive to negative half cycles of I square wave. The output port of the +1, (−1) multiplier 213 is connected for supplying the in-phase synchrodyne results to the input port of a digital lowpass filter 207. The lowpass filter 207 responds to the baseband portion of the in-phase synchrodyne results, but not to image signal.
The lagging Q square wave that the frequency divider 203 supplies to the control input port of the +1, (−1) multiplier 214 conditions the +1, (−1) multiplier 214 to perform a 2-to-1 decimation of the 0°, 90°, 1800 and 2700 digital samples of DCM-COFDM signal supplied to its input port, selecting the 900 digital samples for multiplication by −1 responsive to negative half cycles of Q square wave, and selecting the 2700 digital samples for multiplication by +1 responsive to positive half cycles of Q square wave. The output port of the +1, (−1) multiplier 214 is connected for supplying quadrature-phase synchrodyne results to the input port of to the input port of a digital lowpass filter 208. The lowpass filter 208 responds to the baseband portion of the quadrature-phase synchrodyne results, but not to image signal.
If the front-end tuner 280 contains digital lowpass filtering of the digitized IF DCM-COFDM signal with rapid roll-off to suppress ACI, there is no reason for the digital lowpass filters 207 and 208 necessarily having to have sharp roll-offs of higher frequencies to suppress AC. The Hilbert transform response of the FIR filter 209 and the response from digital delay line 210 are utilized in the subsequent portions of the
As with the
As with the
Subsequent portions of the
The digital lowpass filter 207 is connected for supplying digitized samples of baseband folded DCM-COFDM signal to the input port of the cyclic prefix detector 84. (Alternatively, the digital lowpass filter 208 is connected for supplying digitized samples of baseband folded DCM-COFDM signal to the input port of the cyclic prefix detector 84 instead). The cyclic prefix detector 84 differentially combines the digitized samples of baseband folded DCM-COFDM signal with those samples as delayed by the duration of an effective COFDM symbol. Nulls in the difference signal so generated should occur, marking the guard intervals of the baseband folded DCM-COFDM signal. The nulls are processed to reduce any corruption caused by noise and to generate better-defined indications of the phasing of COFDM symbols. The output port of the cyclic prefix detector 84 is connected to supply these indications to the first of two input ports of the timing synchronization apparatus 285.
The signal input port of a guard interval remover 863 is connected for receiving digitized samples of an in-phase baseband COFDM signal from the output port of the digital lowpass filter 207. The output port of the guard interval remover 863 is connected for supplying the input port of a discrete-Fourier-transform (DFT) computer 873 with windowed portions of the quadrature-phase baseband signal that span respective COFDM symbol intervals. The signal input port of the guard interval remover 864 is connected for receiving digitized samples of a quadrature-phase baseband COFDM signal from the output port of the digital lowpass filter 208. The output port of the guard interval remover 864 is connected for supplying the input port of a discrete-Fourier-transform (DFT) computer 874 with windowed portions of the in-phase baseband signal that span respective COFDM symbol intervals. The DFT computers 873 and 874 are similar in construction, each having the capability of transforming a respective half of the COFDM carriers nominally 4K, 8K or 16K in number to the complex coordinates of respective NU-QAM symbols. The DFT computers 873 and 874 perform bandpass filtering of individual OFDM carriers which bandpass filtering should be unresponsive to frequencies outside baseband. This bandpass filtering may allow digital lowpass filters 207 and 208 to be replaced by respective direct connections in modified
The timing synchronization apparatus 285 is connected for supplying gating control signals to respective control input ports of the guard interval removers 863 and 864. The timing synchronization apparatus 285 is further connected for supplying COFDM symbol timing information to the DFT computers 873 and 874. The indications concerning the phasing of COFDM symbols that the cyclic prefix detector 84 supplies to the timing synchronization apparatus 285 are sufficiently accurate for (a) initial windowing of the in-phase baseband folded COFDM signal that the guard interval remover 863 supplies to the DFT computer 873 and (b) initial windowing of the quadrature-phase baseband folded COFDM signal that the guard interval remover 862 supplies to the DFT computer 874.
The output port of the DFT computer 874 is connected via Hilbert transformation connections 875 for supplying complex coordinates of NU-QAM symbols conveyed by respective ones of the received COFDM carriers to first addend input ports of a parallel array 876 of digital complex-number adders and to minuend input ports of a parallel array 877 of digital complex-number subtractors. These connections 875 are such as to perform Hilbert transform of the complex coordinates of NU-QAM symbols, which procedure is explained in greater detail in the remaining portion of this paragraph. The real coordinates of the complex coordinates of NU-QAM symbols are applied as imaginary components of input signals to the first addend input ports of the parallel array 876 of digital adders and to the minuend input ports of the parallel array 877 of digital subtractors. The imaginary coordinates of the complex coordinates of NU-QAM symbols are applied as real components of input signals to the first addend input ports of the parallel array 876 of digital adders and to the minuend input ports of the parallel array 877 of digital subtractors. There is essentially no delay in this Hilbert transformation procedure, and it takes up little (if any) extra area on the silicon die in a monolithic integrated circuit construction. The output port of the DFT computer 873 is connected for supplying complex coordinates of NU-QAM symbols conveyed by respective ones of the received COFDM carriers to second addend input ports of the parallel array 876 of digital complex-number adders and to subtrahend input ports of the parallel array 877 of digital complex-number subtractors.
The parallel array 876 of digital adders additively combines the complex coordinates of NU-QAM symbols the DFT computer 873 generates, as transformed by the Hilbert transformation connections 875, with the complex coordinates of corresponding NU-QAM symbols the DFT computer 874 generates. The sum output ports of the parallel array 876 of digital adders recover at baseband the complex coordinates of NU-QAM symbols from the lower-frequency subband of the DCM-COFDM signal. The complex coordinates of NU-QAM symbols extracted from pilot carriers in each COFDM symbol sampling interval are supplied as parallel input signal to the pilot carriers processor 288. The complex coordinates of NU-QAM symbols extracted from carriers in each COFDM symbol sampling interval that convey coded data are supplied as parallel input signal to the frequency-domain channel equalizer 893 for NU-QAM symbols extracted from the lower-frequency subband of the DCM-COFDM signal.
The parallel array 877 of digital subtractors differentially combines the complex coordinates of NU-QAM symbols the DFT computer 874 generates, as transformed by the Hilbert transformation connections 875, with the complex coordinates of corresponding NU-QAM symbols the DFT computer 873 generates. The difference output ports of the parallel array 877 of digital subtractors recover at baseband the complex coordinates of NU-QAM symbols from the higher-frequency subband of the DCM-COFDM signal. The complex coordinates of NU-QAM symbols extracted from pilot carriers in each COFDM symbol sampling interval are supplied as parallel input signal to the pilot carriers processor 288. The complex coordinates of NU-QAM symbols extracted from carriers in each COFDM-symbol sampling interval that convey coded data are supplied as parallel input signal to the frequency-domain channel equalizer 894 for NU-QAM symbols extracted from the higher-frequency subband of the DCM-COFDM signal.
More particularly, the NU-QAM symbols from the lower-frequency subband of the DCM-COFDM signal that convey data are supplied by respective ones of the parallel array 876 of digital adders directly to respective ones of the parallel input ports of the selected one of the P/S converters in
the bank 93 of them. The output port of that selected P/S converter responds to supply serialized NU-QAM symbols from the lower-frequency subband of the DCM-COFDM signal to the multiplicand input port of the multiplier 891. The parallel input ports of a selected one of the parallel-to-serial (P/S) converters in a bank 293 of them receives, in parallel from the pilot carriers processor 288, the weighting coefficients for frequency-domain channel equalization of the lower-frequency subband of the DCM-COFDM signal. The output port of that selected P/S converter responds to supply serialized weighting coefficients for the lower-frequency subband of the DCM-COFDM signal to the multiplier input port of the complex-number multiplier 891. The multiplier 891 responds to its multiplicand and multiplier input signals to supply from its product output port an equalized first set of NU-QAM symbols, suitable for subsequent demapping.
More particularly, the NU-QAM symbols from the higher-frequency subband of the DCM-COFDM signal that convey data are supplied by respective ones of the parallel array 877 of digital subtractors directly to the parallel input ports of the selected one of the P/S converters in the bank 94 of them. The output port of that selected P/S converter responds to supply serialized NU-QAM symbols from the higher-frequency subband of the DCM-COFDM signal to the multiplicand input port of the multiplier 892. The parallel input ports of a selected one of the parallel-to-serial (P/S) converters in a bank 294 of them receives, in parallel from the pilot carriers processor 288, the weighting coefficients for frequency-domain channel equalization of the higher-frequency subband of the DCM-COFDM signal. The output port of that selected P/S converter responds to supply serialized weighting coefficients for the lower-frequency subband of the DCM-COFDM signal to the multiplier input port of the complex-number multiplier 891. The multiplier 891 responds to its multiplicand and multiplier input signals to supply from its product output port an equalized second set of NU-QAM symbols, suitable for subsequent demapping.
The modified phase shift method of ISB demodulation as described in connection with
An analog-to-digital converter 205 performs analog-to-digital conversion of the in-phase and quadrature-phase components of the baseband signal supplied from the output port of the mixer 201. An analog-to-digital converter 206 performs analog-to-digital conversion of the in-phase and quadrature-phase components of the baseband signal supplied from the output port of the mixer 202. The digitized in-phase baseband signal supplied from the output port of the A-to-D converter 205 is supplied to the input port of a digital lowpass filter 207. The digitized quadrature-phase baseband signal supplied from the output port of the A-to-D converter 206 is supplied to the input port of a digital lowpass filter 208. Preferably, the design of the digital lowpass filters 207 and 208 provides a rapid roll-off in frequency response, so as to suppress adjacent-channel interference (ACI). The DFT computers 871 and 872 perform bandpass filtering of individual OFDM carriers which bandpass filtering should be unresponsive to frequencies outside baseband. This bandpass filtering may allow digital lowpass filters 207 and 208 to be replaced by respective direct connections in modified
The output port of the lowpass filter 207 and the output port of the lowpass filter 208 are connected to respective addend input ports of the digital adder 211, which is operative to recover at baseband the lower-frequency subband of the DCM-COFDM signal at its sum output port. The output ports of the lowpass filters 207 and 208 are respectively connected to the subtrahend input port and the minuend input port of the digital subtractor 212, which is operative to recover at baseband the higher-frequency subband of the DCM-COFDM signal at its difference output port. The responses from the sum output port of the digital adder 211 and from the difference output port of the digital subtractor 212 are utilized in the subsequent portions of the
The leading I square wave that the frequency divider 203 supplies to the control input port of the +1, (−1) multiplier 213 conditions the +1, (−1) multiplier 213 to select the 0° digital samples of the in-phase second-IF DCM-COFDM signal for multiplication by +1 responsive to positive half cycles of I square wave, and selecting the 1800 digital samples of the in-phase second-IF DCM-COFDM signal for multiplication by −1 responsive to negative half cycles of I square wave. The output port of the +1, (−1) multiplier 213 is connected for supplying the in-phase synchrodyne results to the input port of a digital lowpass filter 207. The lowpass filter 207 responds to the baseband portion of the in-phase synchrodyne results, but not to image signal.
The lagging Q square wave that the frequency divider 203 supplies to the control input port of the +1, (−1) multiplier 214 conditions the +1, (−1) multiplier 214 to select the −90° digital samples of the quadrature-phase second-IF DCM-COFDM signal for multiplication by +1 responsive to positive half cycles of Q square wave, and selecting the 900 digital samples of the quadrature-phase second-IF DCM-COFDM signal for multiplication by −1 responsive to negative half cycles of Q square wave. The output port of the +1, (−1) multiplier 214 is connected for supplying quadrature-phase synchrodyne results to the input port of to the input port of a digital lowpass filter 208. The lowpass filter 208 responds to the baseband portion of the quadrature-phase synchrodyne results, but not to image signal.
If the front-end tuner 480 contains digital lowpass filtering of the digitized IF COFDM DCM signal with rapid roll-off in frequency response for suppressing ACI, there is no reason for the digital lowpass filters 207 and 208 necessarily having to have rapid roll-offs in frequency response to suppress AC. The output port of the lowpass filter 207 and the output port of the lowpass filter 208 are connected to respective addend input ports of the digital adder 211, which is operative to recover at baseband the lower-frequency subband of the DCM-COFDM signal at its sum output port. The output ports of the lowpass filters 207 and 208 are respectively connected to the minuend input port and the subtrahend input port of the digital subtractor 212, which is operative to recover at baseband the higher-frequency subband of the DCM-COFDM signal at its difference output port. The responses from the sum output port of of the digital adder 211 and from the difference output port of the digital subtractor 212 are utilized in the subsequent portions of the
The single second mixer 310 of the
The output port of the switching mixer 316 connects to the input port of a lowpass filter 322 that suppresses image signal in the response supplied from its output port to the input port of an amplifier 323 of the in-phase (“I”) second-IF signal. The output port of the “I” second-IF amplifier 323 is connected to supply analog amplified in-phase second-IF signal that is suitable for an output signal from the
The output port of the switching mixer 317 connects to the input port of a lowpass filter 325 that suppresses image signal in the response supplied from its output port to the input port of an amplifier 326 of the quadrature-phase (“Q”) second-IF signal. The output port of the “Q” second-IF amplifier 326 is connected to supply analog amplified quadrature-phase second-IF signal that is suitable for an output signal from the
Each of the
The structures depicted in
Rather than operating two DFT computers in parallel in the in-phase and quadrature-phase branches of the receiver apparatus shown in
Various other modifications and variations can be made in the specifically described apparatuses without departing from the spirit or scope of the invention in certain broader ones of its aspects. For example, in variations of the structures depicted in
Persons skilled in the art of designing OFDM communications systems and acquainted with this disclosure are apt to discern that various modifications and variations can be made in the specifically described apparatuses without departing from the spirit or scope of the invention in certain broader ones of its aspects. (For example, the invention can be usefully employed in electronic apparatus used in wireless telephonic communication systems.) Accordingly, it is intended that such modifications and variations be considered to result in further embodiments of the invention, to be included within the scope of the appended claims and their equivalents in accordance with the doctrine of equivalents.
In the appended claims, the word “said” rather than the word “the” is used to indicate the existence of an antecedent basis for a term being provided earlier in the claims. The word “the” is used for purposes other than to indicate the existence of an antecedent basis for a term appearing earlier in the claims, the usage of the word “the” for other purposes being consistent with customary grammar in the American English language.
This is a continuation-in-part of U.S. patent application Ser. No. 17/237,045 filed on 24 Apr. 2021. The invention relates to communication systems, such as a digital television (DTV) broadcasting system, that employ dual-carrier modulation (DCM) coded orthogonal frequency-division multiplexed (COFDM) signal. The invention relates more particularly to applying labeling diversity to DCM-COFDM signals employed in such communication systems, which labeling diversity improves transmission and reception of DCM-COFDM signals communicated via a channel afflicted with additive white Gaussian noise (AWGN) or other continuous noise.
