`US 6,292,917 B1
`(10) Patent No.:
`Sep. 18, 2001
`(45) Date of Patent:
`Sinha et al.
`
`US006292917B1
`
`(54) UNEQUAL ERROR PROTECTION FOR
`DIGITAL BROADCASTING USING
`CHANNEL CLASSIFICATION
`
`(75)
`
`Inventors: Deepen Sinha; Carl-Erik Wilhelm
`Sundberg, both of Chatham, NJ (US)
`
`(73) Assignee: Agere Systems Guardian Corp.,
`Orlando, FL (US)
`
`(*) Notice:
`
`Subject to any disclaimer, the term ofthis
`patent is extended or adjusted under 35
`US.C. 154(b) by 0 days.
`
`(21) Appl. No.: 09/163,656
`
`(22)
`
`Filed:
`
`Sep. 30, 1998
`
`TInt C0 eeeeesccccseesnneeececcennnnseeseceneen H03M 13/00
`(SL)
`(52) US. Ch. oe 714/752; 375/299; 714/786;
`714/790
`(58) Field of Search 0... 714/752, 790,
`714/795, 786; 375/299; 382/232; 455/452,
`450; 704/223
`
`(56)
`
`References Cited
`U.S. PATENT DOCUMENTS
`
`9/1993 Wei cocccsccccccsceeesseceeseee 375/299
`5,243,629 *
`10/1995 Dorward etal.
`«. 714/755
`5,463,641
`5,850,482 * 12/1998 Meanyet al.
`..
`« 382/232
`5,887,263 *
`3/1999 Ishii...
`we 455/452
`5,991,717 * 11/1999 Minde et al.
`..
`. 704/223
`6,052,594 *
`4/2000 Chuangetal.
`w. 455/450
`6,108,810 *
`8/2000 Kroegeret al... 714/790
`OTHER PUBLICATIONS
`
`
`
`D. Sinha, J.D. Johnston, S. Dorward and S.R. Quackenbush,
`“The Perceptual Audio Coder,” in Digital Audio, Section 42,
`pp. 42-1 to 42-18, CRC Press, 1998.
`N.S. Jayant and E.Y. Chen, “Audio Compression: Technol-
`ogy and Applications,” AT&T Technical Journal, pp. 23-34,
`vol. 74, No. 2, Mar.—Apr. 1995.
`J. Hagenauer, “Rate Compatible Punctured Convolutional
`Codes (RCPC Codes) and their Applications,” [EEE Trans-
`actions on Communications, vol. 36, No. 4, pp. 389-400,
`Apr. 1988.
`J. Hagenaueret al., “The Performance of Rate-Compatible
`Punctured Convolutional Codes for Digital Mobile Radio,”
`IEEE Transactions on Communications, vol. 38, No. 7, pp.
`966-980, Jul. 1990.
`
`R.V. Cox et al., “Sub-band Speech Coding and Matched
`Convolutional Channel Coding for Mobile Radio Chan-
`nels,” IEEE Transactions on Acoustics, Speech and Signal
`Processing, vol. 39, No. 8, pp 1717-1731, Aug. 1991.
`A.R. Calderbank and N. Seshadri, “Multilevel Codes for
`Unequal Error Protection,” IEEE Transactions on Informa-
`tion Theory, vol. 39, No. 4, pp. 1234-1248, Jul. 1993.
`C-E.W. Sundberg, “Digital Audio Broadcasting in the FM
`Band,” Proceedings of the IEEE Symposium on Industrial
`Electronics, Portugal, Jul. 7-11, 1997.
`C-E.W. Sundberg, “Digital Audio Broadcasting: An Over-
`view of Some Recent Activities in the U-S.,” Proceedings of
`Norsig—97, Norwegian Signal Processing Symposium,
`Tromso, Norway, May 23-24, 1997.
`
`* cited by examiner
`
`Primary Examiner—Albert Decady
`Assistant Examiner—David Ton
`(74) Attorney, Agent, or Firm—Ryan, Mason & Lewis, LLP
`
`(57)
`
`ABSTRACT
`
`The invention provides methods and apparatus for process-
`ing information,e.g., audio, video or image information, for
`transmission in a communication system. In an illustrative
`embodiment, interference characteristics are determined for
`a set of n channels to be used to transmit audio information
`bits, where n is greater than or equal to two. The audio
`information bits are separated into n classes based on error
`sensitivity, for example, the impact of errors in particular
`audio data bits on perceived quality of an audio signal
`reconstructed from the transmission. The classesofbits are
`then assigned to the n channels such that the classes of bits
`having the greatest error sensitivity are transmitted over the
`channels which are the least susceptible to interference. The
`interference characteristics associated with the n channels
`
`can be determined by, for example, measuring interference
`levels for one or more of the channels, or obtaining infor-
`mation regarding knowninterference levels for one or more
`of the channels. The channels may correspond to different
`frequency bands, timeslots, code division slots or any other
`type of channels. The invention can provide UEP for dif-
`ferent classes of audio information bits even in cases in
`
`which the same convolutional code, or the same comple-
`mentary punctured pair convolutional (CPPC) codepair, is
`used to encode the classes. The assignmentof the classes of
`bits to the channels, as well as the characteristics of the
`classes and the channels, may be fixed or dynamic.
`
`32 Claims, 2 Drawing Sheets
`
`210:
`242
`
`
`won | Xr
`
`
`|
`BAHUER
`CLASSIFIER
`TRANSMITTER
`202.
`
`
`
`|
`
`
`
`
`
`200
`
`
`
`Oy
`|
`‘RECEIVER
`224
`m5
`ALDI
`CHANNEL
`
`DECODER L— DECLASSIFIER ESTECOUERS
`
`:
`ia
`
`|
`|
`
`IPR2018-1556
`HTC EX1022, Page 1
`
`IPR2018-1556
`HTC EX1022, Page 1
`
`
`
`U.S. Patent
`
`Sep. 18, 2001
`
`Sheet 1 of 2
`
`US 6,292,917 B1
`
`FIG.
`
`1
`2/5(4/5+4/5)
`
`2/5(4/5+4/5)
`
`IPR2018-1556
`HTC EX1022, Page 2
`
`IPR2018-1556
`HTC EX1022, Page 2
`
`
`
`U.S. Patent
`
`Sep. 18, 2001
`
`Sheet 2 of 2
`
`US 6,292,917 B1
`
`|
`AUDIO
`CHANNEL|2
`
`
`
`
`
`CLASSIFIER Fe CODERS c= INTERLEAVERS
`
`—~7)et6
`
`| TRANSMITTERTT
`mt
`
`
`
`
`
`
`as _|
`RECEIVER
`
`904 2
`crDEINTERLEAVERS
`DECLASSIFIEA ADECODERS
`OUTPUT
`
`
`
`AUDIO
`
`
`
`
`CHANNEL jae
`
`
`
`
`IPR2018-1556
`HTC EX1022, Page 3
`
`IPR2018-1556
`HTC EX1022, Page 3
`
`
`
`US 6,292,917 B1
`
`1
`UNEQUAL ERROR PROTECTION FOR
`DIGITAL BROADCASTING USING
`CHANNEL CLASSIFICATION
`
`FIELD OF THE INVENTION
`
`The present invention relates generally to digital audio
`broadcasting (DAB) and other techniques for transmitting
`information, and moreparticularly to techniques for provid-
`ing unequal error protection (UEP) for different classes of
`audio, video, image or other information bits encoded in a
`source coding device.
`
`BACKGROUND OF THE INVENTION
`
`Mostsource coded bit streams exhibit unequal sensitivity
`to bit errors. For example, certain source bits can be much
`more sensitive to transmission errors than others. Moreover,
`errorsin certain bits, such as control bits, may lead to severe
`error propagation and a corresponding degradation in recon-
`structed signal quality. Such error propagation can occur, for
`example, in the output audio bits of an audio coderdueto the
`use of control bits for codebook information, frame size
`information, synchronization information and so on. The
`perceptual audio coder (PAC) described in D. Sinha, J.D.
`Johnston, S. Dorward and S.R. Quackenbush, “The Percep-
`tual Audio Coder,” in Digital Audio, Section 42, pp. 42-1 to
`42-18, CRC Press, 1998, whichis incorporated by reference
`herein, attempts to minimize the bit rate requirements for the
`storage and/or transmission of digital audio data by the
`application of sophisticated hearing models and signal pro-
`cessing techniques. In the absence of channel errors, a PAC
`is able to achieve near stereo compact disk (CD) audio
`quality at a rate of approximately 128 kbps. At a lowerbit
`rate of 96 kbps, the resulting quality is still fairly close to
`that of CD audio for many important types of audio material.
`The rate of 96 kbpsis particularly attractive for FM band
`transmission applications such as in-band digital audio
`broadcasting (DAB) systems, which are also known as
`hybrid in-band on-channel (HIBOC), all-digital IBOC and
`in-band adjacent channel (IBAC)/in-band reserve channel
`(IBRC) DAB systems. There is also a similar effort under-
`wayto provide digital audio broadcasting at lower audio bit
`rates in the AM band. For these AM systems,audio bit rates
`of about 32 to 48 kbps are being considered for daytime
`transmission and about 16 kbps for nighttime transmission.
`Higheraudio bit rates, greater than about 128 kbps,are being
`used in multiple channel DAB systems. The transmission
`channels in the above-noted DAB systems tend to be
`severely bandlimited and noise limited at the edge of a
`coverage area. For mobile receivers, fading is also a severe
`problem. It is therefore particularly important in these and
`other applications to design an error protection technique
`that is closely matchedto the error sensitivity of the various
`bits in the compressed audio bit stream.
`PACsandother audio coding devices incorporating simi-
`lar compression techniques are inherently packet-oriented,
`ie., audio information for a fixed interval (frame) oftimeis
`represented by a variable bit length packet. Each packet
`includescertain control information followed by a quantized
`spectral/subband description of the audio frame. For stereo
`signals, the packet may contain the spectral description of
`two or more audio channels separately or differentially, as a
`center channel and side channels (e.g., a left channel and a
`right channel). Different portions of a given packet can
`therefore exhibit varying sensitivity to transmission errors.
`For example, corrupted control information leads to loss of
`synchronization and possible propagation of errors. On the
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`2
`other hand, the spectral components contain certain inter-
`frame and/or
`interchannel
`redundancy which can be
`exploited in an error mitigation algorithm incorporated in a
`PAC codec. Even in the absence of such redundancy, the
`transmission errors in different audio components have
`varying perceptual implications. For example, loss of stereo
`separation is far less annoying to a listener than spectral
`distortion in the mid-frequency range in the center channel.
`Unequal error protection (UEP) techniques are designed
`to match error protection capability with sensitivity to
`transmission errors, such that the most important bits are
`provided with the highest level of error protection, while less
`important bits are provided with a lesser level or levels of
`error protection. A conventional two-level UEP technique
`for use in DAB applications is described in N.S. Jayant and
`E.Y. Chen, “Audio Compression: Technology and
`Applications,” AT&T Technical Journal, pp. 23-34, Vol. 74,
`No. 2, March-April 1995. In this technique, which is based
`on a Reed-Solomon (RS) code, the control information is
`protected more robustly since it is not possible to use error
`mitigation on the non-redundantcontrol information. In fact,
`the proper operation of the error mitigation algorithm used
`in a PAC codec is itself dependent upon reliable control
`information. All of the non-control spectral information in
`this technique is protected using a uniform level of error
`protection. U.S. patent application Ser. No. 09/022,114,
`which wasfiled Feb. 11, 1998 in the name of inventors
`Deepen Sinha and Carl-Erik W. Sundberg, and which is
`incorporated by reference herein, discloses techniques for
`providing UEPof a PACbitstream by classifying the bits in
`different categories of error sensitivity. These classes were
`then matched to a suitable level of error protection to
`minimize the overall impactof errors,1.e., the most sensitive
`bits are more protected than the others. Certain of the UEP
`techniques described in the above-cited application gener-
`ally provide improvements without regard to the type of
`channel, and the channel noise is typically assumed to be
`averaged over time and frequency by interleaving in both
`time and frequency for each channel codeclass. Thus, a UEP
`technique with a more powerful channel code properly
`matched to the most sensitive source bits always outper-
`forms the corresponding equal error protection (EEP) tech-
`nique. However, determining the channel codes for such
`UEP scenarios is often a nontrivial problem,particularly in
`the case of determining single sideband complementary
`punctured-pair convolutional codes (CPPC) codes for
`HIBOC applications. Therefore, although the techniques in
`the above-cited application provide considerable improve-
`ment over prior approaches to UEPfor digital audio, further
`improvements are needed for certain implementations, such
`as the above-noted HIBOC systems and other similar sys-
`tems.
`
`SUMMARYOF THE INVENTION
`
`The present invention provides methods and apparatus for
`implementing UEPfor a source coded bit stream such asthat
`generated by a perceptual audio coder (PAC). In an illus-
`trative embodiment, interference characteristics are deter-
`mined for a set of n channels to be used to transmit audio
`
`information bits, where n is greater than or equal to two. The
`audio information bits are separated into n classes based on
`error sensitivity, for example, the impact of errors in par-
`ticular audio data bits on perceived quality of an audio signal
`reconstructed from the transmission. The classesofbits are
`
`then assigned to the n channels such that the classes of bits
`having the greatest error sensitivity are transmitted over the
`channels which are the least susceptible to interference. The
`
`IPR2018-1556
`HTC EX1022, Page 4
`
`IPR2018-1556
`HTC EX1022, Page 4
`
`
`
`US 6,292,917 B1
`
`3
`interference characteristics associated with the n channels
`can be determined by, for example, measuring interference
`levels at different times and locations for one or more of the
`
`channels, or obtaining information regarding known inter-
`ference levels for one or more of the channels. The channels
`
`may correspond to different frequency bands, time slots,
`code division slots or any other type of channels. The
`channel properties may also change with factors such as
`time and location within a coverage area.
`In accordance with another aspect of the invention, the
`assignmentof the classes of bits to the channels, as well as
`the characteristics of the classes and the channels, may be
`fixed or dynamic. For example, in applications in which the
`interference characteristics associated with one or more of
`the channels vary as a function of time, position within a
`coverage area, or other factors, the assignmentof the classes
`of bits to the channels can be varied so as to ensure that the
`
`classes of bits having the greatest error sensitivity continue
`to be transmitted over the channels which are least suscep-
`tible to interference. As another example, amounts of chan-
`nel resources used for particular classes of audio information
`bits can vary as a function of time.
`The invention can provide UEP for different classes of
`information bits even in cases in which the same convolu-
`tional code, or the same CPPC codepair, is used to encode
`the classes, although different channel codes could also be
`used to encode the classes. The invention can be applied to
`other types of digital information, including, for example,
`video and image information. Moreover, the invention is
`applicable not only to perceptual coders but also to other
`types of source encoders using other compression tech-
`niques operating over a wide range of bit rates, and can be
`used with transmission channels other than radio broadcast-
`ing channels.
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`FIG. 1 illustrates a two-class frequency division unequal
`error protection (UEP) technique is accordance with the
`invention as applied to an exemplary hybrid in-band
`on-channel (HIBOC)digital audio broadcasting (DAB) sys-
`tem.
`
`FIGS. 2 through 4 illustrate a number of possible alter-
`native implementations of the two-class UEP technique of
`FIG. 1.
`
`FIG. 5 is a block diagram of a communication system in
`which an n-class frequency division UEP techniqueis imple-
`mented in accordance with an illustrative embodimentof the
`invention.
`
`DETAILED DESCRIPTION OF THE
`INVENTION
`
`The invention will be described below in conjunction with
`exemplary unequal error protection (UEP) techniques for
`use in the transmission of audio informationbits, e.g., audio
`bits generated by an audio coder such as the perceptual audio
`coder (PAC) described in D. Sinha, J-D. Johnston, S. Dor-
`ward and S.R. Quackenbush, “The Perceptual Audio
`Coder,” in Digital Audio, Section 42, pp. 42-1 to 42-18,
`CRCPress, 1998. It should be understood, however, that the
`UEPtechniques of the invention may be applied to many
`other types of information, e.g., video or image information,
`and other types of coding devices. In addition, the invention
`may be utilized with a wide variety of different types of
`communication applications,
`including communications
`over the Internet and other computer networks, and over
`cellular multimedia, satellite, wireless cable, wireless local
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`4
`loop, high-speed wireless access and other types of com-
`munication systems. Although illustrated at least in part
`using frequency bands as channels, the invention may also
`be applied to many other types of channels, such as, for
`example, time slots, code division multiple access (CDMA)
`slots, and virtual connections in asynchronoustransfer mode
`(ATM) or other packet-based transmission systems. The
`term “channel” as used herein should be understood to
`
`include any identifiable portion or portions of a communi-
`cation medium whichis used to transmit one or more signals
`and has an interference characteristic associated therewith,
`and is thus intended to include, for example, a sub-channel,
`segment or other portion of a larger channel.
`FIG. 1 illustrates channel classification UEP in accor-
`dance with an illustrative embodiment of the invention. In
`this embodiment, which is particularly well-suited for use in
`HIBOC DAB applications, the channels correspond gener-
`ally to frequency bands, and the UEPtechniqueis therefore
`referred to as frequency division UEP. Unlike certain of the
`approaches described in the above-cited U.S. patent appli-
`cation Ser. No. 09/022,114, which can generally be charac-
`terized as time division UEP in which enhanced error
`protection may be provided for a certain class or classes of
`audio bits transmitted using a numberof different channels,
`frequency division UEP in accordance with the invention
`provides enhanced error protection for a given class of bits
`by assigning that class of bits to a particular channel for
`transmission.
`
`In the embodiment of FIG. 1, a portion of a frequency
`spectrum in an exemplary HIBOC DAB system is shown,
`including an analog host FM signal 100 with associated
`lower sidebands 102L, 104L and corresponding upper side-
`bands 102U, 104U. The sidebands represent portions of the
`frequency spectrum used to transmit digital audio
`information, and the sets of sidebands 102L, 102U and
`104L, 104U correspond generally to frequency channels
`102, 104, respectively, used to transmit the digital audio
`information. In accordance with the invention, a determina-
`tion is madeas to the interference characteristics associated
`
`with each of the frequency channels 102 and 104. This
`determination may be based, for example, on actual mea-
`surements of average signal-to-interference ratios within the
`channels, on known orestimated interference levels, or on
`any other information which provides an indication of
`relative or absolute interference levels for the channels. For
`
`example, it has been estimated based on previous experience
`with HIBOC systems that the portion of the spectrum of
`FIG. 1 at the highest and lowest frequencies is typically
`more susceptible to interference than the portion closest to
`the analog host FM signal 100. It will therefore be assumed
`that one of the channels, 1.e., channel 102 in this example,
`has been determined to be less susceptible to interference
`than channel 104.
`
`Theillustrative embodimentof the invention, after deter-
`miningthe relative or absolute interference levels associated
`with n channels, where n=2, to be used for transmission of
`digital audio information, separates the audio information
`into n classes of bits based on error sensitivity, and then
`assigns the n classes of bits to the n channels such that the
`bits most sensitive to errors are transmitted in the channels
`which are least susceptible to interference. In the FIG. 1
`example, the audio information bits are separated into two
`classes, designated class I and class II, with class I including
`the bits most sensitive to errors. The determination of error
`
`sensitivity may be based on perceptual audio coding con-
`siderations such as those described in the above-cited U.S.
`
`patent application Ser. No. 09/022,114. For example, class I
`
`IPR2018-1556
`HTC EX1022, Page 5
`
`IPR2018-1556
`HTC EX1022, Page 5
`
`
`
`US 6,292,917 B1
`
`5
`may include the audio control bits as well as certain audio
`data bits corresponding to frequency bands which are per-
`ceptually important
`in reconstructing the encoded audio
`signal. These and othererror sensitivity classification tech-
`niques are described in greater detail in Application Ser. No.
`09/022,114, and will not be further described herein.
`In the FIG. 1 example, the most sensitive audio informa-
`tion bits, i.e., class I, are transmitted in channel 102, i.c., the
`channel determined to be less susceptible to interference.
`This provides an increased robustness for the class I bits
`against the higher interference levels in channel 104. The
`two-class frequency division UEP approach illustrated in
`FIG. 1 will provide improvements over a conventional EEP
`approach. In one possible implementation of the FIG. 1
`approach, the same channel code may be used for both the
`class I and II bits, but with a separate interleaving in time
`and frequency. It should be noted that the above-described
`frequency division UEP approach generally provides no
`improvement for channels which have a uniform interfer-
`ence level as a function of frequency. However, by taking
`into account the different interference characteristics of the
`channels, it can provide UEP for different classes of bits
`using the same code.
`FIG. 2 illustrates another possible implementation of a
`two-class frequency division UEP approach in accordance
`with the invention. This example uses complementary
`punctured-pair convolutional (CPPC) codes, such as those
`described in greater detail in U.S. patent application Ser. No.
`09/006,570, which was filed Jan. 13, 1998 in the name of
`inventors Brian Chen and Carl-Erik W. Sundberg, and is
`incorporated by reference herein. In this example, thebits in
`classes I and II are each separately coded using a rate-2/5
`code which is formed as a combination of a pair of rate-4/5
`CPPC codes. These rate-4/5 codes are referred to as half-
`bandwidth codes, and combine to form a rate-2/5 error
`correction code referred to as a full-bandwidth code. As is
`
`described in U.S. patent application Ser. No. 09/006,570, a
`rate-1/3 mother code can be punctured to meet these exem-
`plary HIBOC code requirements. The rate-1/3 mother code
`may bea rate-1/3 convolutional code having a constraint
`length K=7 as described in J. Hagenauer, “Rate-compatible
`punctured convolutional codes (RCPC codes) and their
`applications,” IEEE Transactions on Communications, Vol.
`36, No. 7, pp. 389-400, April 1988.
`The coderate is the ratio of input bits to output bits for the
`convolutional encoder,
`i.e., a rate-1/3 encoder generates
`three output bits for each input bit. A group of three coded
`output bits is referred to as a symbol. The value of K refers
`to the number of uncoded input bits which are processed to
`generate each output symbols. For example, a rate-1/3
`convolutional encoder with K=7 generally includes a seven-
`bit shift register and three modulo-two adders. The inputs of
`the each of the adders are connected to a different subset of
`
`the bits of the shift register. These connections are specified
`by the “generators” of the encoder. Because a given output
`symbolin this example is generated using the latest input bit
`as well as the previous six input bits stored in the shift
`register, the K=7 encoderis said to have a “memory”ofsix.
`The rate-1/3, K=7 code used in this example has the fol-
`lowing three generators:
`o=1011011
`
`g,=1111001
`
`go=1100101
`
`Each of the generators may be viewed as specifying the
`connections betweenbits of the seven-bit shift register and
`
`6
`inputs of one of the modulo-2 adders. For example, the adder
`corresponding to generator g, generates the first bit of each
`output symbol as the modulo-2 sum of the bits in thefirst,
`third, fourth, sixth and seventh bit positions in the shift-
`register, with the first bit position containing the latest input
`bit. Similarly, the generators g, and g, generate the second
`and third bits,
`respectively, of each output symbol as
`modulo-2 sumsofthe bits in the positions designated by the
`respective generator values. The free Hamming distance d,
`of the rate-1/3, K=7 code with the above-noted generators is
`14, and its information error weight c,/P
`is one. Whenthis
`code is punctured in a rate-compatible mannerto rates of
`4/11, 4/10, 4/9 and 1/2,the resulting rate-1/2 codeis also the
`best rate-1/2, K=7 convolutional code. Additional details
`regarding specific CPPC codes suitable for use in embodi-
`mentsof the invention, as well as bit placementstrategies for
`arranging the bits within the upper and lower sideband
`portions of the channels, can be found in U.S. patent
`application Ser. No. 09/006,570.
`FIGS. 3 and 4 illustrate other embodiments of the inven-
`tion in which a dynamic boundary betweenclass I and class
`II bits is used. In each of these embodiments, the boundary
`between class I andclass II is as indicated by the dashed line
`110. The portion of the frequency spectrum shown in FIGS.
`3 and 4 includes the analog host FM signal 100, along with
`a lower sideband 106 and an upper sideband 108. As in the
`examples of FIGS. 1 and 2, the upper and lower sidebands
`are used to transmit digital audio information. In the FIG. 3
`embodiment,
`the channels do not correspond directly to
`specific portions of the upper and lower sidebands. Instead,
`a first channel is defined by a portion of both the upper and
`lower sideband to one side of the dashed line 110, while a
`second channel is defined by the portion of the upper and
`lowersidebandto the otherside of the dashed line 110. Each
`of the upper and lower sidebands 106 and 108 uses,e.g, the
`same rate-2/5 code, as indicated. The use of a dynamic
`boundary allows a channel occupying a greater portion of
`the available frequency spectrum to be used to transmit class
`I bits. FIG. 4 shows another possible implementation using
`a dynamic boundary 110. A control channelor other suitable
`mechanism may be used to inform the receiver in a particu-
`lar geographical area which configuration, e.g., the configu-
`ration of FIG. 3, the configuration of FIG. 4, or another type
`of configuration, is being used at the transmitter. The con-
`figuration may vary as a function of factors such as time or
`position within a coverage area.
`It should be noted that in the embodiments of FIGS. 1
`
`through 4, the same code, e.g., the same CPPC codepair,
`may be used for both classes I and II, or different codes may
`be used for each of the classes. In addition, as previously
`noted, the techniques can be readily extendedin a straight-
`forward manner to n channels and classes, where n=2.
`Other possible variations include, for example, separate or
`joint interleaving, soft combining or equal gain combining,
`fixed or variable bit assignments, and use of other types of
`codes such as block codes.
`
`FIG. 5 is a block diagram of an exemplary communication
`system 200 which implements the above-described fre-
`quency division UEP in accordance with the invention. The
`system 200 includes a transmitter 202 and a receiver 204
`which communicate over an n-channel
`transmission
`medium 206. The transmitter 202 includes an audio encoder
`210, e.g., a PAC encoder, for generating a sequence of audio
`packets from an analog audio input signal. Although this
`embodimentuses audio packets, such as those generated by
`a PAC encoder, the invention is more generally applicable to
`digital audio information in any form and generated by any
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`IPR2018-1556
`HTC EX1022, Page 6
`
`IPR2018-1556
`HTC EX1022, Page 6
`
`
`
`US 6,292,917 B1
`
`7
`type of audio compression technique. The audio packets
`from encoder 210 are applied to a classifier 212 which
`converts the packets into separate bit streams corresponding
`to n different classes of audio informationbits. The classifier
`212 is also responsible in this embodiment for assigning
`each of the classes of bits to one of the available channels
`such that the classes of bits most sensitive to errors are
`
`transmitted in the channels which are least susceptible to
`interference, as previously described. The separate bit
`streams from the classifier 212 are applied to a set of channel
`coders 214. The symbol outputs of the channel coders 214
`are supplied to a set of interleavers 215 which provide
`interleaving of the symbols within each channel over mul-
`tiple audio packets. The interleaved symbols are then sup-
`plied to a set of orthogonal frequency division multiplexed
`(OFDM) modulators 216 for modulation in accordance with
`conventional OFDM techniques. The OFDM modulators
`may provide, for example, single-carrier modulation in each
`of the channels. Of course, other types of modulation may be
`used in alternative embodiments.
`The transmitter 202 may include additional processing
`elements, such as a multiplexer, an upconverter andthe like,
`which are not shown in FIG. 5 for simplicity of illustration.
`In addition, the arrangement of elements may be varied in
`alternative embodiments. For example, other types of modu-
`lators may be used in place of the OFDM modulators 216,
`such as modulators suitable for generating signals for trans-
`mission over a telephone line or other network connection,
`and separate interleaving and coding need not be applied to
`each of the channels.
`
`The receiver 204 receives the transmitted OFDMsignals
`from the transmission channels 206, and processes them in
`OFDM demodulators 219 to recoverthe interleaved symbols
`for each of the channels. The symbols are deinterleaved in
`a set of deinterleavers 220, and then applied to a set of
`channel decoders 222. The bit streams at the output of each
`of the decoders in the set of decoders 222 correspond to the
`different classes of audio information bits. These bit streams
`
`are then processed in a declassifier 224 which reconstructs
`audio packets from the bit streams. The resulting sequence
`of audio packets are then decoded in an audio decoder 226
`to reconstruct the original analog audio signal.
`Like the transmitter 202, the receiver 204 may include
`additional processing elements which are not shownin FIG.
`5. It should also be noted that various elements of the system
`200, such as the interleavers 215 and the deinterleavers 220,
`may be eliminated in alternative embodiments. Moreover,
`various elements of the system 200, such as the audio
`encoder 210 and decoder 226, the channel coders 214 and
`decoders 222, and the classifier 212 and declassifier 224,
`may be implemented using an application-specific inte-
`grated circuit, microprocessor or any other type of digital
`data processor, as well as portions or combinations of such
`devices. Various aspects of the invention may also be
`implemented in the form of one or more software programs
`executed by a central processing unit (CPU)orthe like in the
`digital data processor.
`Simulation results for an exemplary frequency division
`UEP (FD-UEP)system suchas that described in conjunction
`with FIGS. 1-5 are shown in TABLE 1 below. In the
`simulations, a channel was assumedto include two disjoint
`segments, designated segment I and segmentII. Such seg-
`ments are also referred to herein as sub-channels, and it
`should be noted that each segmentis itself consideredto fall
`within the general definition of “channel” given above. In
`other words, each segment may be considered a channel.
`With a suitable interleaver depth, the channel quality may be
`
`8
`assumed to be constant over a particular segment. The two
`segments can thus be parameterized by an interference
`characteristic such as,
`for example,
`the corresponding
`signal-to-noise level measured in terms of E,/N,. Gaussian
`channel conditions are assumed in the simulations.
`In an EEPtransmission system operating over segments I
`andII, it is reasonable to assume half of the channel coded
`bits encounter a channel condition which exists in segment
`I and another half encounter conditions existing in segment
`II. For the FD-UEP system,it is assumed that audio infor-
`mation bits are separated into a class I which includes
`control bits and a first portion of the audio data bits, and a
`class II which includes a second portion of the audio data
`bits. These classes I and II may correspond, for example, to
`classes 1* and 2*, respectively, as described in application
`Ser. No. 09/022,114.
`In accordance with the present
`invention, the class I and II bits may be interleaved and
`transmitted independently over segments I and II, respec-
`tively. Therefore, class I bits are exposed to the channel
`condition in segment I and class II bits face the channel
`condition in segment
`II.
`In each of the simulations, a
`convolutional channel code with a rate of 2/5 was used, as
`described above, and the same outer cyclic redundancy
`codes (CRCs) were also used.
`
`TABLE1
`
`Sim-
`i
`
`Channel Condition
`E,/No)
`in dB
`
`10
`
`15
`
`20
`
`25
`
`30
`
`lation
`No.
`
`Segment I
`
`Segment
`Ir
`
`EEP Quality
`
`FD-UEP Quality
`
`1.
`2.
`
`3.
`
`-0.5
`-0.5
`
`-0.5
`
`-0.5
`-2.5
`
`-3.0
`
`Slight di