`115
`United States Patent
`5,596,604
`[11] Patent Number:
`
` [45] Date of Patent: Jan. 21, 1997
`
`Cioffi et al.
`
`UOAAAA
`
`[54]
`
`[75]
`
`[21]
`
`[22]
`
`[51]
`[52]
`[58]
`
`[56]
`
`MULTICARRIER MODULATION
`TRANSMISSION SYSTEM WITH VARIABLE
`DELAY
`
`Inventors: John M. Cioffi, Cupertino; Po Tong,
`Fremont; James T. Aslanis, Mountain
`View; Antoinette H. Gooch, Palo Alto,
`all of Calif.
`
`Assignee: Amati Communications Corporation,
`San Jose, Calif.
`
`Appl. No.: 107,200
`
`Filed:
`
`Aug. 17, 1993
`
`Tint. Cho occ ccccesssssssssscccscsssssneessssseneccesssssneees H04K 1/10
`
`US. Cl. wees
`. 345/260; 371/43; 371/35
`Field of Search... 375/38, 39; 371/38.1,
`371/39.1, 43, 35
`
`References Cited
`
`OTHER PUBLICATIONS
`
`Bingham,“Multicarrier Modulation for Data Transmission:
`An Idea Whose Time Has Come”, JEEE Communications
`Magazine, pp. 5-(May 1990).
`Chowetal., “A Discrete Multitone Transceiver System for
`HDSL Applications”, IEEE Journal on Selected Area in
`Communications, vol. 9, No. 6, pp. 895-908 (Aug. 1991).
`Chow et al., “Performance Evaluation of a Multichannel
`Transceiver System for ADSL and VHDSLServices”, IEEE
`Journal on Selected Areas in Communications, vol. 9, No. 6,
`pp. 909-919 (Aug. 1991).
`Fleminget al., “ADSL: The On-Rampto the Information
`Highway”, Telephony, pp. 20, 24-26 (Jul. 1993).
`Clark et al., “‘Interleaver Structures for Coded Systems”,
`Error—Correction Coding for Digital Communications, Sec-
`tion 8.3, pp. 347-349.
`Primary Examiner—Stephen Chin
`Assistant Examiner—T. Ghebretinsae
`Attorney, Agent, or Firm—Hickman Beyer & Weaver, LLP
`(57)
`ABSTRACT
`
`U.S. PATENT DOCUMENTS
`
`A transmission system using multicarrier modulation applies
`FECC (forwarderror correcting code) coding and codeword
`7/1983 Currie et al. wees 340/347 DD
`4,394,642
`interleaving differently to input signals from a plurality of
`
`
`4,677,625 6/1987Bettsetab.ovceesecsssecssereetes 371/43
`different data channels to produce encoded data signals
`4,747,104
`5/1988
`. 371/39
`
`havingdifferentreliabilities and different coding delays. Bits
`wen
`4,852,102
`7/1989
`.. 371/40
`of encoded data signals having relatively less delay are
`
`4,980,897
`12/1990 Decker et al.
`.
`.. 375/38
`allocated to carriers that are subject to relatively more
`5,105,442—4/1992 Wei wrecccscscccvssesseressessseeneneenssesees 375/39
`attenuation and/or channel noise, and hence that are allo-
`
`371/37.1
`5,107,504
`4/1992 Nakamuraetal.
`cated fewer bits for transmission in each symbol period, to
`9/1993 Wei 0... eee
`we 371/43
`5,243,629
`
`
`reduce the effects of impulse noise. The data channels can
`.. 375/59
`5,251,236
`10/1993 Brehmeret al.
`
`. 348/473
`5,282,019
`1/1994 Basile et al....
`comprise video, data, and control channels transmitted on an
`2/1994 Parr oe
`we 371/43
`5,287,374
`
`ADSL (asymmetric digital subscriber line) two-wire tele-
`
`4/1994 Calderbank et al.
`we 375/39
`5,305,352
`phoneline.
`
`.........
`.. 375/39
`5,321,725
`6/1994 Paik et al.
`..ccscsesesesneeesees 371/43
`5,392,300
`2/1995 Borth et al.
`
`22 Claims, 3 Drawing Sheets
`
`SWITCH
`INPUT
`
`DATA
`CHANNELS
`
`
`
`
`
`
`INTERFACE,
`FILTER,
`FILTER,
`-------------
`
`A-D,
`INTERFACE|POTS
`POTS
`FILTER
`
`
`
`20
`L
`
`
`ADD PREFIX D-A,
`
`
`REMOVE
`PREFIX
`
`:
`
` SWITCH
`OUTPUT
`
`DATA
`CHANNELS
`
`62
`64
`
`IPR2019-00958
`Apple Inc. EX1006 Page 1
`
`IPR2019-00958
`Apple Inc. EX1006 Page 1
`
`
`
`U.S. Patent
`
`Jan. 21, 1997
`
`Sheet 1 of 3
`
`5,596,604
`
`TOHLNOD
`
`OAGIA
`
`TOYHLNOO
`
`NQSI
`
`Nas!
`
`NaSsI
`
`TOYULNOD
`
`IPR2019-00958
`Apple Inc. EX1006 Page 2
`
`IPR2019-00958
`Apple Inc. EX1006 Page 2
`
`
`
`U.S. Patent
`
`Jan. 21, 1997
`
`Sheet 2 of 3
`
`YA4d4ne
`
`XlaaudGdV
`
`viva
`
`HO.LIMS
`
`5,596,604
`
`S1Od|3OV4YSLNIHalls|siod
`
`iMdinoSau.HALHOS
`STANNVHO
`‘WAIav
`Wa‘aNd
`
`v9c909bScs
`
`cr9bEc&0€
`HOLIMS HALHOS
`
`02
`
`7)
`
`@‘ble
`
`‘FOVAHALNI
`
`JAOWSH
`
`
`
`~Xidaud~
`
`
`
`H344N¢‘O34
`
`STANNVHO
`
`LONI
`
`viva
`
`IPR2019-00958
`Apple Inc. EX1006 Page 3
`
`IPR2019-00958
`Apple Inc. EX1006 Page 3
`
`
`
`
`
`
`U.S. Patent
`
`Jan. 21, 1997
`
`Sheet 3 of 3
`
`5,596,604
`
`
`
`
`CODER
`
`
`MICRO-
`PROCESSOR
`
`CONTROL
`UNIT
`
` TRELLIS
`
`Fig. 3
`
`TRELLIS
`
`CODER
`
`IPR2019-00958
`Apple Inc. EX1006 Page 4
`
`IPR2019-00958
`Apple Inc. EX1006 Page 4
`
`
`
`5,596,604
`
`1
`MULTICARRIER MODULATION
`TRANSMISSION SYSTEM WITH VARIABLE
`DELAY
`
`This invention relates to transmission systems using
`multicarrier modulation, also known as discrete multitone
`(DMT) modulation where, as is desirable, the modulationis
`effected using a discrete Fourier Transform.
`
`BACKGROUND OF THE INVENTION
`
`The principles of multicarrier modulation are described
`for example in “Multicarrier Modulation For Data Trans-
`mission: An Idea Whose Time Has Come” by John A. C.
`Bingham, IEEE Communications Magazine, Vol. 28, No. 5,
`pages 5-14, May 1990. As is known, in a transmission
`system using multicarrier modulation, FDM (frequency
`division multiplex) carriers spaced within a usable fre-
`quency band of a transmission channel, forming a set of
`subchannels, are modulated at a block or symbol transmis-
`sion rate of the system. The bits of input data for transmis-
`sion within each block or symbol period are allocated to the
`carriers or subchannels so that the bit error rates of the
`subchannels as monitored at the receiver, are substantially
`equal. As a result, the different subchannels carry different
`numbersofbits in each symbol period. With an appropriate
`allocation of bits and transmit powers to the carriers or
`subchannels, such a system provides a desirable perfor-
`mance.
`
`The characteristics and performance of one such system,
`for communicating data at a rate of 1.6 Mb/s over a twisted
`pair channel using 256 subchannels,
`is described in “A
`Discrete Multitone Transceiver System For HDSL Applica-
`tions”by J. S. Chow et al., IEEE Journal On Selected Areas
`In Communications, Vol. 9, No. 6, pages 895-908, August
`1991. A companion paper by P. S. Chow et al. entitled
`“Performance Evaluation Of A Multichannel Transceiver
`System For ADSL and VHDSLServices”, at pages 909-919
`of the same publication, addresses a similar system applied
`to an asymmetric digital subscriber line (ADSL).
`An article by S. Fleming et al. entitled “ADSL: The
`on-ramp to the information highway”, Telephony, Jul. 12,
`1993, pages 20-26 describes one example of an ADSL
`arrangementapplied to a two-wire telephone subscriberline,
`in which four asymmetric 1.5 Mb/s channels are provided
`for transmission in a downstream direction from a telephone
`CO (central office) to a subscriber, in addition to various data
`channels and POTS (plain old telephone service) carried
`symmetrically (i.e. bidirectionally) on the line. The data
`channels for example comprise an ISDN (integrated services
`digital network) HO channel at 384 kb/s or an ISDN basic
`access channel at 144 kb/s, and a control channel for
`example at a bit rate of 16 kb/s. The four asymmetric
`channels provide a total bandwidth of 6 Mb/s that can be
`used for digital video signals.
`A well knownproblem in the art of transmission systems
`is that of impulse noise, which can producebursts of errors
`on transmission channels. In order to address this problem,
`it is knownto apply forwarderror correction coding (FECC)
`and interleaving techniques in which a block of input data to
`be transmitted is augmented with parity data that enables
`one or moreerrors in the block to be detected and corrected,
`the input data and parity data constituting a codeword, and
`over time parts of different codewords are interleaved for
`transmission to reducetheeffect of error bursts on individual
`codewords. For example, Currie et al. U.S. Pat. No. 4,394,
`
`10
`
`20
`
`30
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`2
`642 issued Jul. 19, 1983 and entitled “Apparatus For Inter-
`leaving and De-Interleaving Data” describes one such
`arrangement. The interleaving can instead be convolutional
`interleaving, for example as described in Section 8.3.1.2 of
`“Error-Correction Coding for Digital Communications” by
`George C. Clark, Jr. and J. Bibb Cain, Plenum Press, pages
`347-349.
`
`The use of FECC increases the bit rate required of the
`actual transmission system in dependence uponthe parity
`overhead, i.e.
`the size of the parity data relative to the
`codeword size, and increases complexity. The interleaving
`process increases immunity to error bursts due to impulse
`noise, but adds transmission delay. Longer periods over
`which the interleavingis effected result in greater immunity
`to impulse noise, but greater transmission delays. Thusthere
`is a trade-off between high reliability (requiring effective
`error correction and immunity to impulse noise) and short
`transmission delay.
`Different types of signals, which may be required to be
`transmitted via a single transmission system, may have
`different requirements forreliability and transmission delay.
`For example, digital video signals that are highly com-
`pressed require a highreliability for their transmission, and
`ISDN voice signals must meet strict transmission delay
`requirements. In known transmission systems, it has been
`necessary to provide a compromise between highreliability
`and transmission delay for different types of signals.
`It is also knownto apply trellis code modulation (TCM)
`techniques to a system using multicarrier modulation in
`order to improve the performance of the system through the
`coding gain provided by the trellis coding. For example,
`Deckeret al. U.S. Pat. No. 4,980,897 issued Dec. 25, 1990
`and entitled “Multi-Channel Trellis Encoder/Decoder” dis-
`closes such a system in which the encoding and decoding
`processes operate on all of the subchannels sequentially in
`order to reduce delay. As is well known, TCM results in
`sequences of transmitted signal constellation points that
`have a greatly increased separation, hence the improved
`performance, and necessitates more complex decoding
`involving maximumlikelihood sequence estimation, usually
`implemented using the Viterbi algorithm.
`The performance improvements that are provided by
`TCM and by FECCasdescribed aboveare largely indepen-
`dent of one another, so that TCM and FECC can be used
`together in a transmission system.
`In Betts et al. U.S. Pat. No. 4,677,625 issued Jun. 30, 1987
`and entitled “Distributed Trellis Encoder” there is described
`an FECC arrangement in which interleaving is effected by
`switching among a plurality of trellis coders with delay
`units, with corresponding switching among a plurality of
`trellis decoders at a receiver. The use of a plurality of
`encoders and decoders in such a manner undesirably addsto
`the costs and complexity of the arrangement. The Bettsetal.
`patentrelates to a transmission system using QAM (quadra-
`ture amplitude modulation) of a single carrier, and not to a
`transmission system using multicarrier modulation.
`An object of this invention is to provide an improved
`transmission system using multicarrier modulation.
`
`SUMMARY OF THE INVENTION
`
`According to one aspect of this invention there is pro-
`vided a data transmission system using multicarrier modu-
`lation, comprising: FECC (forward error correction code)
`coding and codeword interleaving apparatus arranged for
`differently encoding a plurality of data signals to provide a
`
`IPR2019-00958
`Apple Inc. EX1006 Page 5
`
`IPR2019-00958
`Apple Inc. EX1006 Page 5
`
`
`
`5,596,604
`
`3
`plurality of encoded data signals with different delays
`through the coding and interleaving apparatus; and a modu-
`lator arranged to modulate bits of the encoded data signals
`onto multiple carriers of the transmission system, different
`numbersof bits in each transmission symbol period being
`allocated to different carriers.
`
`The different delays through the coding and interleaving
`apparatus correspond to different levels of coding and/or
`interleaving and hence to different degrees of reliability of
`the transmitted signals, especially with respect
`to their
`immunity to impulse noise, the least delayed signals being
`the most vulnerable to impulse noise. Thus each signal to be
`transmitted can be communicated with an individually
`selected compromise between high reliability and short
`transmission delay.
`In order to compensate for the greater vulnerability of the
`least delayed (i.e. least interleaved) signals to impulse noise,
`preferably the modulator is arranged to allocate bits of
`encoded data signals having relatively less interleaving to
`carriers carrying relatively fewerbits in each symbolperiod.
`The greater transmission attenuation of such carriers, which
`for signal transmission are compensatedfor by the allocation
`of fewer bits in each symbolperiod, acts to advantage in also
`attenuating the impulse noise.
`The coding and interleaving apparatus can comprise a
`plurality of data paths providing the encoded data signals
`with different delays, and a switch for switching different
`data signals to different data paths.
`In an embodiment of the system described below, the
`coding and interleaving apparatus comprisesa first store for
`storing the plurality of data signals, a second store, an FECC
`coder coupled between an output of the first store and an
`input of the second store, and a control unit for controlling
`the supply of data signals from thefirst store to the coder and
`for controlling storage of FECC codewords from the coder
`into the second store to provide codeword interleaving. The
`coder can be a programmable FECC coder arranged to be
`programmed by the control unit to provide different coding
`for different data signals, and the contro] unit can provide
`different depths of interleaving for the FECC codewords of
`different data signals.
`In this case the modulator preferably comprises a table for
`providing indices of the carriers in order of the number of
`bits in each transmission symbol period allocated to the
`carriers, means for reading the respective numberofbits for
`each carrier from the second store in order of the different
`delays through the coding and interleaving apparatus, and
`sorting means for supplying the respective bits for modula-
`tion onto the respective carriers.
`The multicarrier modulation is preferably implemented
`using a discrete Fourier Transform and preferably incorpo-
`rates trellis coding. Accordingly the modulator preferably
`comprisesa trellis coder, apparatus for providing an Inverse
`Fast Fourier Transform (IFFT), a table for providing indices
`of the carriers in order of the number of bits in each
`transmission symbol period allocated to the carriers, means
`for reading the respective number of bits for each carrier
`from the second store to the trellis coder in order of the
`different delays through the coding and interleaving appa-
`ratus, and a sorter for supplying signal amplitudes supplied
`by the trellis coder to storage locations of the IFFT apparatus
`identified by the respective carrier indices.
`According to another aspect, the invention provides a data
`transmission system using multicarrier modulation, com-
`prising: apparatus for applying FECC (forward error cor-
`rection code) coding and codewordinterleaving differently
`
`15
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`4
`to different data signals to produce encoded data signals with
`different delays; and meansfor allocating the encoded data
`signals to the multiple carriers in dependence upon said
`delays and in accordance with a predeterminedcriterion.
`Preferably the predetermined criterion comprises trans-
`mission attenuation of and/or noise on the carriers, bits of
`the encoded data signals having relatively less delay being
`allocated to carriers that are subject
`to relatively more
`attenuation and/or noise. Alteratively,
`the predetermined
`criterion may comprise the signal-to-noise ratios of the
`carriers, or another characteristic of the carriers or the
`transmitted signals that can be monitored by the transmis-
`sion system, and may take into accounta desired distribution
`of error rates amongthe carriers.
`The invention also provides a method of modulating
`multiple carriers with signals of a plurality of data channels,
`comprising the steps of: applying FECC (forward error
`correcting code) coding and codeword interleaving differ-
`ently to signals of different data channels to produce
`encoded data signals having different delays; and modulat-
`ing different numbersofbits of the encoded data signals onto
`different carriers.
`
`Preferably the modulating step includes the step of allo-
`cating bits of the encoded data signals havingrelatively less
`delay to carriers carrying relatively fewerbits.
`Preferably the step of applying FECC coding and code-
`word interleaving comprises the stepsof: storing signals of
`the different data channels; sequentially FECC coding the
`stored signals to produce FECC codewords; and storing the
`FECC codewordsin an interleaved manner,the interleaving
`being different for the codewords of the different data
`channels. The modulating step preferably also includes the
`step oftrellis coding the bits of the encoded data signals.
`The invention further extends to a method of modulating
`multiple carriers with signals of a plurality of data channels,
`comprising the steps of: applying FECC (forward error
`correction code) coding and codeword interleaving differ-
`ently to signals of different data channels to produce
`encoded data signals having different coding delays; and
`allocating the encoded data signals to the carriers in depen-
`dence upon the coding delays and in accordance with a
`predetermined criterion. Conveniently the predetermined
`criterion comprises transmission attenuation of and/or noise
`on the carriers, bits of the encoded data signals having
`relatively less delay being allocated to carriers that are
`subject to relatively more attenuation and/or noise.
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`The invention will be further understood from the fol-
`lowing description with reference to the accompanying
`drawings, in which:
`FIG.1 illustrates a block diagram of an ADSL transmis-
`sion system using multicarrier modulation in accordance
`with an embodimentof this invention;
`FIG.2 illustrates a functional block diagram of a down-
`stream transmitter, provided at a telephone CO, and a
`downstream receiver, provided at a subscriber’s premises, of
`the system of FIG.1;
`FIG.3 illustrates a block diagram of an embodiment of
`parts of the downstream transmitter; and
`FIG.4 is a diagram illustrating the operation of parts of
`the downstream transmitter.
`
`DETAILED DESCRIPTION
`
`Referring to FIG. 1, an ADSL transmission system com-
`prises a telephone central office (CO) transceiver 10 and
`
`IPR2019-00958
`Apple Inc. EX1006 Page 6
`
`IPR2019-00958
`Apple Inc. EX1006 Page 6
`
`
`
`5,596,604
`
`10
`
`20
`
`30
`
`5
`remote terminal (RT) transceiver 12, linked via a channel 14
`providing a downstream transmission path 16, from a down-
`stream transmitter (Tx) 18 in the CO to a downstream
`receiver (Rx) 20 in the RT, and an upstream transmission
`path 22 from an upstream transmitter 24 in the RT to an
`upstream receiver 26 in the CO.
`Signals transmitted in the downstream direction are indi-
`cated by way of example as comprising digital video, ISDN,
`control, and POTS signals, and signals transmitted in the
`upstream direction are indicated by way of example as
`comprising ISDN,control, and POTS signals; other types of
`data may be transmitted in a similar manner. The system is
`asymmetric in that the video signals, which require a large
`bandwidth of for example 6 Mb/sas already discussed, are
`transmitted only in the downstream direction. The other
`signals,
`transmitted in both directions, require a much
`smaller bandwidth. In addition to these signals, an opera-
`tions control channel (OPC) provides for transmission in
`both directions on the channel 14 between the transceivers
`10 and 12 of control signals required for operation of the
`transceivers as described below.
`
`6
`The number of bits carried by each subchannel in each
`symbol period can be zero (i.e. the subchannelis not being
`used) or can vary from a minimum number, for example 1
`or 2 bits, to a maximum number, for example in a range from
`10 to 16 bits. Because the subchannels carry variable num-
`bers ofbits, the total transmission rate of the system is not
`fixed but can be increased or decreased to meet particular
`requirements.
`Signals from a plurality of input data channels, for
`example information and control channels as discussed
`above, are supplied to inputs of the switch 30, which
`switches these signals to one of a plurality of outputs each
`coupled to a respective oneof the plurality of FECC coders
`32. Although two FECCcoders 32 are shown in FIG. 2, a
`different number of such coders may alternatively be pro-
`vided. Each FECC coder 32 provides a respective degree of
`FECCcoding and interleaving, and hencereliability of the
`data signals that it encodes, and a corresponding transmis-
`sion delay. The switch 30 switches the data signals to the
`respective outputs and hence FECC coders in accordance
`with latency(i.e. transmission delay) and reliability require-
`ments for the respective signals. As already indicated, such
`The channel 14 is for example a two-wire telephone
`requirements can vary for different types of signals, and for
`subscriber line on which the POTSsignals are transmitted in
`signals on a single channel at different times.
`a low frequency band below about 10 kHz and on which the
`The particular characteristics of the individual FECC
`other signals are transmitted by multicarrier modulation at
`coders 32 are not important to the invention, but the different
`higher frequencies. The downstream and upstream signals
`coders have different degrees of interleaving so that the
`can be separated by frequency division multiplexing or
`transmitted data signals have different- susceptibilities to
`using echo cancellation techniques.
`impulse noise as discussed further below. For example, one
`Referring to FIG. 2, the downstream transmitter 18 func-
`of the FECC coders 32 may provide coding with relatively
`tionally comprises a switch 30, one or more FECC (forward
`short term or no interleaving, or even may be omitted
`error correcting code) coders 32 two of whichareillustrated,
`entirely, thus propagating data signals with little or no delay
`data buffers 34, a trellis coder 36 having an associated bit
`but with a relatively high susceptibility to impulse noise, and
`and energy allocation table 38 and a carrier index table 40,
`the other of the coders 32 may provide coding with inter-
`a sorter 42, an Inverse Fast Fourier Transform (IFFT)
`leaving over relatively long periods, thereby propagating
`apparatus 44 including a bufferat its input and a prefix adder
`data signals with a high immunity to impulse noise but with
`at its output, and a digital-to-analog converter, analogfilter,
`a relatively long delay.
`and line interface block 46. Conversely, the downstream
`The data signal codewords output from the FECC coders
`receiver 20 comprises a line interface, analog filter, and
`32 are buffered in the buffers 34, which are represented in
`analog-to-digital converter block 48 that is assumed also to
`FIG.2 as being divided into relatively fast, i.e. short delay,
`include a digital filter, an FFT apparatus 50 including a
`buffers and relatively slow, or long delay, buffers, corre-
`prefix removerat its input and frequency domain equalizers
`sponding to the respective interleaving periods associated
`(FEQ)andabufferatits output, a sorter 52, a trellis decoder
`with the respective FECCs 32. In practice, as described
`54, operating in accordance with the Viterbi algorithm to
`below,the storage in the buffers 34 is used in implementing
`perform maximum likelihood sequenceestimation in known
`the FECCinterleaving. The units 36 to 42 serve to read bits
`manner, having an associated bit allocation table 56 and
`of the codewords from the buffers 34,
`implementtrellis
`carrier index table 58, data buffers 60, one or more FECC
`coding, and supply the resulting amplitudes to the IFFT
`decoders 62 complementary to the coders 32, and a switch
`64,
`apparatus 44 in each transmission symbol period for trans-
`mission via the multicarrier subchannels. The operation of
`The components 44 to 50 form a multicarrier modulation
`the units 36 to 42 is described further below.
`or DMTtransmission system of generally known form, to
`Conversely, in the receiver 20 amplitudes produced by the
`which system the components 36, 38, 54, and 56 addtrellis
`FFT apparatus 50 are converted by the units 52 and 54 to
`coding and decoding in generally known manner. In this
`codeword bits that are supplied to the buffers 60,
`the
`system datasignals, constituted by the video, ISDN,control,
`codewords being de-interleaved and decoded by the FECC
`and OPC signals represented in FIG. 1, are frequency
`decoders 62, each of which operates in a complementary
`division multiplexed at frequencies above about 10 kHz,i.e.
`mannerto a respective one of the FECC coders 32. From the
`above the frequencies of POTS telephone signals that are
`FECCdecoders 62 the decoded data signals are supplied to
`supplied to and derived from the blocks 46 and 48 and are
`respective output data channels via the switch 64.
`separated by the analogfilters within these blocks.
`Although details of the units 44 to 50 are not significant
`By way of example, the DMT system may have 256
`to the present invention,it is noted here that the prefix added
`carriers with a frequency spacing of 4 kHz, the discrete
`at the output of the IFFT apparatus 44 consists of a repetition
`Fourier transform accordingly having a length of N=512,
`of information from the end of a data block in order to
`with a symbol period of 250 ps and about 1700 bits per
`provide a guard space to eliminate interference from one
`symbol to provide a total transmission rate of about 6.8
`transmitted symbolto the next. The digital filter in the block
`Mb/s, with each carrier or subchannel carrying a number of
`48 is a time domain equalizerin the form ofa finite impulse
`bits in each symbol period that depends on the characteris-
`responsefilter that limits such interference to less than the
`tics (e.g. signal to noise ratio, or SNR) of the subchannel.
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`IPR2019-00958
`Apple Inc. EX1006 Page 7
`
`IPR2019-00958
`Apple Inc. EX1006 Page 7
`
`
`
`5,596,604
`
`7
`the prefix then being removed or
`length of the prefix,
`subtracted at the input of the FFT apparatus 50 thereby
`eliminating the interference. The frequency domain equal-
`ization (FEQ)at the output of the FFT apparatus 50 operates
`individually on each subchannel and adaptively adjusts for
`the attenuation and delay of each carrier.
`In addition it is observed that, while preferred, the use of
`trellis coded modulation is not essential to the invention.
`Accordingly the trellis coder 36, which converts data bits
`from the buffers 34 into scaled amplitudes for the IFFT
`apparatus 36 as well as implementing the trellis coding, may
`be replaced by simpler apparatus for converting data bits
`into scaled amplitudes without any trellis coding, with a
`complementary replacementof the trellis decoder 54.
`As is well known and discussed above, in each symbol
`period of a system using multicarrier modulation different
`subchannels carry different numbers of bits, in accordance
`with the SNR of the respective subchannels. Typically, it
`may be desired for all of the subchannels to have substan-
`tially the same SNR as monitored at the receiver, and this is
`assumed by way of example in the remainder of this
`description, but other distributions of SNR among subchan-
`nels may be desirable and can alteratively be provided. The
`SNR of each subchannel is dependent upon the attenuation
`and the noise level of the subchannel. Accordingly, subchan-
`nels with a relatively high attenuation or noise level are
`allocated relatively fewer bits, and hence a greater signal
`point spacing at the transmitter, than subchannels with a
`relatively low attenuation or noise level. Due to the different
`subchannel attenuations,
`the signal point spacing at the
`receiver(and the bit error rate) is approximately the same for
`all of the subchannels. Thus the characteristics of the sub-
`channels are substantially compensated for by an appropri-
`ate distribution of the number of bits carried by each
`subchannel.
`
`impulse noise can be
`The invention recognizes that
`coupled onto the transmission path at any point along its
`length, and from its coupling point to the receiver 20 is
`subjected to the same frequency-dependent attenuation as
`the data signals. (Impulse noise may also be introduced in
`the transmitter 10 as a result of clipping in digital logic or at
`the digital-to-analog converter,
`this possibly being intro-
`duced deliberately in order to reduce implementation costs.)
`Consequently, impulse noise that appears on therelatively
`more attenuating subchannels, which carry relatively fewer
`bits in each symbol period, is relatively more attenuated at
`the receiver. It follows that subchannels carrying relatively
`fewer bits in each symbol period are less susceptible to
`impulse noise (because of the greater attenuation) than
`subchannels carrying relatively more bits in each symbol
`period.
`In order to provide the best possible overall performance,
`correspondingto all of the data signals having substantially
`the same immunity to impulse noise, the invention matches
`those data signals that are the most susceptible to impulse
`noise, by virtue of interleaving of the FECC codewords over
`relatively shorter periods, with the subchannels that have the
`mostattenuation (and hence carry the fewest numberofbits
`in each symbol period) and hence on which the impulse
`noise is most attenuated so that it has the least effect at the
`receiver.
`
`To this end, in the transmitter 18, the index table 40 is
`arranged to store the index, or number, i of each subchannel
`sorted in order of increasing number ofbits b, (from the
`minimum number to the maximum number as discussed
`above) per symbol period allocated to the subchannels. In
`
`10
`
`15
`
`20
`
`30
`
`35
`
`45
`
`50
`
`55
`
`60
`
`65
`
`8
`each symbolperiod, the sorted indices i are read in tum from
`the index table 40 and used to addressthe table 38, and are
`supplied to the sorter 42. The allocated number ofbits to be
`used for the respective subchannel, and an energy scale
`factor for the subchannel, are accordingly read from the
`table 38 and suppliedto the trellis coder 36, which reads the
`allocated numberofbits from the buffers 34 starting with the
`fastest (least interleaving and delay) buffer and progressing
`gradually through all of the bits to be read and transmitted
`in the symbol period, ending with the slowest (most inter-
`leaving and delay) buffer. For each subchannel index i the
`trellis coder 36 produces amplitudes, representing a signal
`point in a constellation of 2” signal points and scaled in
`accordance with the energy scale factor for the subchannel,
`which are written by the sorter 42 into a position i in the
`buffer at the input of the IFFT apparatus 44.
`In the receiver 20, the index table 58 is the same as the
`index table 40, andthebit allocation table 56 stores the same
`allocations of numberofbits for each subchannelas thetable
`
`38. As in the transmitter, in each symbol period the sorted
`indices i are read in turn from the index table 58, are
`supplied to the sorter 52, and are used to address the table
`56. The allocated number of bits used for the respective
`subchannel i are read from the table 56 and supplied to the
`trellis decoder 54, which reads the amplitudes from the
`position i in the output buffer of the FFT apparatus 50 as
`identified by the sorter 52 and supplies the relevant number
`of bits to the buffers 60, starting in each symbol period with
`the fastest (least interleaving and delay) buffer and progress-
`ing gradually to the slowest (most interleaving and delay)
`buffer. The operation of the FECC decoders 62 and the
`switch 64 is the inverse of the coders 32 and switch 30.
`The various units in the transceivers 10 and 12 can be
`implemented in various ways, using arrangements of hard-
`ware and/or software. FIG, 3 illustrates by way of example
`a block diagram of one embodimentof the units 30 to 42,
`and the buffer at the input of the IFFT apparatus 44, of the
`downstream transmitter 18.
`
`Referring to FIG. 3, a microprocessor contro! unit 68,
`which may comprise one or more microprocessors such as
`digital signal processors for carrying out respective func-
`tions,
`is coupled via various control paths to a RAM
`(random access memory) 70 that operates as the switch 30
`in FIG. 2, a programmable FECC coder 72 that constitutes
`the plurality of FECC coders 32 in FIG. 2, a RAM 74 that
`correspondsto the buffers 34,a trellis coder 76 correspond-
`ing to the trellis coder 36, and a RAM 78that constitutes the
`buffer at the input of the IFFT apparatus 44 in FIG. 2. The
`RAM 70, shown with multiple inputs for simplicity,
`is
`supplied at its input with the input data channels and OPC
`data from the control unit 68 via a path 80, and the control
`unit 68 is supplied with information from the control chan-
`nel (e.g. for selecting between high reliability and short
`delay for t