United States Patent [19]
`Jasper et aI.
`
`111111111111111111111111111111111111111111111111111111111111111111111111111
`USOO5381449A
`[11] Patent Number:
`[45] Date of Patent:
`
`5,381,449
`Jan. 10, 1995
`
`[75]
`
`[54] PEAK TO AVERAGE POWER RATIO
`REDUCTION METHODOLOGY FOR QAM
`COMMUNICATIONS SYSTEMS
`Inventors: Steven C. Jasper, Hoffman Estates;
`Mark A. Birchler, Roselle, both of
`Ill.
`[73] Assignee: Motorola, Inc., Schaumburg, TIL
`[21] Appl. No.: 786,681
`[22] Filed:'
`Nov. 1, 1991
`
`Related U.S. Application Data
`[63] Continuation-in-part of Ser. No. 536,825, Jun. 12, 1990.
`Int. CI.6 ............................................. H04L 27/04
`[51]
`[52] U.S. Q •....................................... 375/59; 332/103
`[58] Field of Search ....................... 375/38, 39, 42, 59,
`375/60; 370/9,10,12,18,19,20; 332/t03, 144
`
`[56]
`
`References Cited
`U.S. PATENT DOCUMENTS
`3,955,141 5/1976 Lyon et aI .............................. 375/8
`4,464,767 8/1984 Bremer .................................. 375/67
`4,646,305 2/1987 Tretter et aI .......................... 375/39 .
`4,680,775 7/1987. Exarque et aI ........................ 375/39
`
`Primary Examiner-Stephen Chin
`Attomey, Agent, or Firm-Joseph P. Krause
`
`[57]
`ABSTRACT
`The ratio of peak power level to average power level in
`a power amplifier used in a QAM communication sys(cid:173)
`tem transmitter can be reduced by preselecting magni(cid:173)
`tudes and phase angles of complex-valued pilot symbols
`used in multi-channel, N-Ievel QAM.
`
`20 Claims, 3 Drawing Sheets
`
`SERIAl. BITS
`I
`TO
`101 I M-PARAllEL
`I-'--"~ Cot.flEX
`SYIIIDlS
`I
`
`INFORMATION
`SOURCE
`B BITS/SEC.
`
`LINEAR
`POWER
`
`• • ..
`
`•
`
`~~-------~--~-~-------~
`
`1
`
`APPLE 1014
`
`

`

`u.s. Patent
`
`Jan. 10, 1995
`
`Sheet 1 of 3
`
`5,381,449
`
`(0,0,0,0)
`
`2
`
`FIG.1
`
`-PRIOR ART-
`
`•
`
`•
`
`•
`
`IMAGINARY
`3-
`
`•
`
`•
`
`•
`
`•
`
`.-
`
`ENERGY
`
`FIG.2
`
`IMAGINARY- _
`
`FIG.3-
`
`-PRIOR ART~
`
`----I---+----¥:....-.....t~_I__- REAL
`
`2
`
`

`

`£
`
`...
`I-l
`QO
`~
`...
`UI
`
`~
`I-It
`Q
`N
`
`(I)
`t:r'
`00
`
`(I) ....
`
`UI
`~
`I-l
`.P
`I-l
`p
`
`~
`
`('D =
`"'C = ~
`L: • rIJ. •
`
`~
`
`SYMBOL STREAM
`,j ~ DATA/PILOT
`COMPLEX
`
`0
`
`. I 6PILOT
`.
`IP"= p. 'ej8i,jr(107-110)
`r---:l
`F IG.5B
`
`'-~A __ ...
`I
`lSi •
`
`DATA SYMBOL~
`.-(
`
`STREAM
`
`COMPLEX
`
`I,j
`
`I,j
`
`REAL
`
`FIG.6
`
`•
`
`•
`
`•
`
`•
`
`-.\+
`
`• •
`
`~
`
`-1+
`
`••
`
`. 1+ .. I·
`
`L
`
`e
`
`. IMAGINARY
`
`••
`• 3t
`
`•
`
`•
`
`e.
`
`e
`
`•
`I ~ II I III I I 1.1 I ~ I I I I I I I I I I .. ·1 I ~ I
`D ~ I I I .1
`. c·1 1 ~ 1 1 1 ~ 1 1 1 I 1 ~ 1·1 1 1 1 :·1 1 1 1 1 1 1 ~ 1 1 ···1· I I II
`. B I I ~ I I I ~ I·' I I I II· I , I I ~. I I ·1 I , I 1·1 I I ···1 I ~ I
`
`SUBCHANNEL
`
`3
`
`

`

`~
`...
`t-l
`QO
`(H
`...
`til
`
`~
`~
`~
`
`r ~
`
`~
`
`~ ... Q .. ...
`
`~ a a
`
`rJJ •
`~ •
`
`AMPLIFIER
`
`POWER
`LINEAR
`
`.100 .
`
`FIG.5A
`
`.~ ____ . ___________ ~ ___ ~_J
`, I
`I
`•
`I
`I
`.'
`.
`••
`I
`. I
`•
`I
`I
`I
`. I :
`. I .
`
`., S(t)
`I
`I
`I
`I
`V JOO .
`I
`. I
`107
`-.---------.----,
`
`S1,j
`
`INSERTION
`
`104 PILOT SYMBOL . 112 PULSE SHAPE
`
`e i("'2t)-1J8
`
`I
`
`. ,
`122
`
`128
`
`~~Tm·
`PULSE SHAPE
`
`S1;]
`111
`
`e i("'1 t) -1J6
`
`120
`
`10J PILOT SYMBOL
`
`108 .
`.
`INSERTION
`
`---
`
`-
`
`-
`
`102
`
`I
`r
`I . .
`I
`, .
`'r'--
`
`. 142
`
`e j cv4 t)r
`
`. .
`{
`
`126
`
`. h tx (t)
`FILTER
`
`116 PULSE SHAPE
`
`110..'
`•
`•
`
`. ,j
`... _IN_SE_RT_IO .. N ... · S1'
`PILOT SYMBOL
`106 .----1.--
`
`. .-....... ..
`
`h. bc-(t)
`FILTER
`
`PULSE SHAPE
`
`.
`124
`h tx (t)
`FILTER
`
`105 PILOT SYMBOL
`
`INSERTION
`
`109
`
`SYMBOLS
`. . ·COMPLEX
`
`B BITS/SEC.'.
`SOURCE.
`INFORMATION
`
`101 1M-PARALLEL.
`" SERIA}O' BITS
`.
`,.
`,
`
`4
`
`

`

`PEAK TO AVERAGE POWER RATIO REDUCI'lON
`METHODOLOGY FOR QAM COMMUNICATIONS
`SYSTEMS
`
`This is a continuation-in-part of Ser. No. 07/536,825,
`filed Jun. 12, 1990.
`
`IS·
`
`1
`
`5,381,449
`
`FIELD OF THE INVENTION
`This invention relates to communications systems.
`More particularlY this invention relates to methods for
`improving the peak power to average power ratio in a
`linear modulation communications system particularly
`a QAM communications system.
`
`2
`In QAM communications systems, the ratio of the
`peak power level to average power level of a QAM
`symbol stream will usually continuously vary by virtue
`of the fact that the data represented by the QAM sym-
`5 boIs itself varies randomly. Accordingly, power ampli(cid:173)
`fiers for QAM communications systems must be capable
`of handling a significant peak to average power level
`ratio and, accordingly, any reduction in the peak to
`average power ratio eases the requirements of a QAM
`10 power amplifier.
`Some prior art, single channel QAM systems are
`frequently transmitted on a communications channel,
`such as a radio frequency channel, in conjunction with
`a pilot component. Such pilot components, by construc-
`15 tively or destructively adding with other QAM symbols
`BACKGROUND OF THE INVENTION
`can at times aggravate the the peak to average power
`level ratio requirements of a QAM power amplifier,
`Various communication systems are known in the art.
`thereby further aggravating the requirements of such an
`Pursuant to many such systems, an information signal is
`amplifier.
`modulated on to a carrier signal and transmitted from a
`Any methodology by which the ratio of peak power
`first location to a second location. At the second loca- 20
`amplit11de to average power amplit11de is reduced
`tion, the information signal is demodulated and recov(cid:173)
`would therefore simplify and reduce the amplifier cost
`ered.
`associated with a QAM system and would be an im(cid:173)
`Typically, the communication path used by such a
`provement over the prior art.
`system has various limitations, such as bandwidth. As a 2
`SUMMARY OF THE INVENTION
`result, there are upper practical limitations that restrict 5
`In a multi-subchannel, N-Ievel QAM communication
`the quantity of information that can be supported by the
`communication path over a given period of time. Vari-
`t
`.
`I
`al d
`il t
`b I
`th
`sys em usmg comp ex-v ue p 0 sym 0 s,
`ere
`.
`h
`ha b
`th
`ffi
`~us m~dulation sc .emes
`~e een p~oposed . at e ec-
`provided herein a method of reducing the ratio of peak
`ttvely m~r~e the information handl~g capacity of the 30 to average power by pre-selecting amplitude and/or
`<:ommumc~tlOn pa~ as meas~ed agamst other mo?ula-
`phase angles for the embedded, complex-valued, pilot
`tlon tec!miques. S1Xtee~-pomt quadra~e amplitude
`symbols added to QAM information symbols, so as to
`I?odulatlOn (Q~ p~oV1des a constellation of modula-
`minimize the peak to average power ratio of a compos-
`tion. value~ (dlstingwsh~d ~rom one another by. each
`ite QAM signal that is transmitted on a communications
`havmg a different combmatlon of phase and amphtude) 35 channel. Such pre-selected pilot symbols include com-
`plex-valued symbols that are not part of the well-known
`w~erein ea~h c~nstellation point represents a plurality
`of info~tion bits.. . .
`constellation of values used in an N-Ievel QAM system,
`By vu:ue of therr changmg.amphtude from Q~M-
`such as the 16 constellation points used in a 16 QAM
`symbol time-to:Q~M-symbol time! Q~ symbols m a
`system. In fact, using the method herein, in a multi-
`Q~ ~ommurucation system req~e .lin~ power am- 40 channel, N-Ievel QAM system wherein, over some
`plification to be able. to accurately distmgwsh one QAM
`length of time during which several QAM symbol
`symbol at one amplit11de level and another QAM sym-
`frames can occur in addition to have different valued
`time-coincident pilots in several subchannels, the pilot
`bol at some other power level. In a radio communica-
`tions system, QAM symbols require a very linear ampli-
`values in one or more sub channel can also change over
`fication prior to broadcasting them on an antenna. In 45 this length time. Stated alternatively, pilot values can
`QAM systems, non-linear amplification of QAM sym-
`change both over time and over subchannels to reduce
`bols in a QAM signal, (which QAM signal is typically
`the peak to average power ratio in the composite signal.
`Frequently at least one pilot symbol will be selected to
`considered to be a pulse-shape filtered and frequency
`up-converted stream of QAM symbols), in a radio trans-
`be off the constellation of values in order to maximally
`mitter can make coherent demodulation impossible. 50 reduce the peak to average power level in the compos-
`Another more common problem with using non-linear
`ite signal of a QAM system, which composite signal is
`amplifiers with QAM modulation is the frequency splat-
`comprised of the combination or summation of a plural-
`ter caused by non-linear amplification of a signal. For
`ity of N-Ievel QAM subchannels, which subchannels
`this reason, linear power amplifiers are required in
`are in-turn comprised of complex-valued QAM infor-
`QAM radio transmitters, which power amplifiers in- 55 mation symbols combined with the complex-valued
`crease in cost, size, and complexity as their output
`preselected pilot symbols. In a multi-channel, N-Ievel
`power level and/or linearity increase.
`QAM system, by proper selection of these preselected
`A problem in the design of a linear power amplifier is
`pilot signals, which are combined with the QAM infor-
`providing the ability of an amplifier to accommodate
`mation symbols (which QAM information symbols in-
`widely fluctuating input power levels while producing 60 elude the information of interest to be transmitted) the
`at its output a faithful reproduction of the input signal.
`combined QAM symbols and the preselected pilot can
`While an amplifier can be readily designed to have a
`have a substantially lowered peak power level to aver-
`linear power amplification of a relatively constant-
`age power level ratio, compared to prior art systems
`amplitude input signal, designing an amplifier that can
`that use only one or more QAM constellation points for
`accommodate a peak power level that might, at any 65 pilot symbols.
`In most application of the method herein, and in the
`given time, exceed the average power level by several
`embodiment of a QAM transmitter disclosed herein, at
`decibels (db) can significantly increase the cost and size
`of the amplifier.
`least one pilot symbol that is to be combined with a
`
`5
`
`

`

`5,381,449
`
`3
`QAM symbol stream, will not lie on the constellation of
`the QAM symbol. Such a non-constellation-based pilot
`symbol is considered to be any complex-valued symbol
`that does not have a phase angle and/or amplitude
`within pre-determined limits, or ranges, circumscribing
`the mathematical points on the rectilinear complex
`plane upon which QAM symbols are identified.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`FIG. 1 shows a 16 QAM constellation map;
`FIG. 2 shows a plot of energy versus frequency for a
`four subchannel QAM system;
`FIG. 3 shows a plot of a prior art QAM pilot symbol;
`FIG. 4 shows a graphical representation of the place(cid:173)
`ment of pilot symbols and QAM information symbols in
`various symbol times, in various time frames in a four
`subchannel QAM system;
`FIG. SA shows a simplified block diagram of an
`improved four channel, 16 QAM communications
`transmitter that provides for improved peak power 20
`ratio to average power ratio in a composite S(t)
`achieved by non-constellation-based pilot symbols;
`FIG. SB shows a simplified representation of the
`improved pilot and its insertion into the QAM symbol
`stream.
`FIG. 6 shows a graphical representation of a 16-point
`constellation for a 16 QAM system, and shows examples
`of both, non-constellation-based pilot symbols and con(cid:173)
`stellation-based pilot symbols.
`
`4
`symbols are transmitted with amplitudes and phase
`angles that correspond to the magnitudes and phase
`angles of the vectors used to represent the various pat(cid:173)
`terns, such as those shown for the vectors 10 and 14
`5 depicted in FIG. 1.
`FIG. SA shows a simplified block diagram of a four(cid:173)
`channel 16 level QAM (I6-QAM) transmitter (100).
`Though depicted in block diagram format for the con(cid:173)
`venience of explanation and understanding it should be
`10 understood that the invention shown in FIG. SA, can be
`practiced in a variety of embodiments; but in particular
`most of the functions (300) of the preferred embodiment
`are performed in a digital signal processor such as the
`Motorola DSP56000 or DSP96000 families. Further-
`15 more, although the embodiment described below is in
`the context of a 16 QAM amplification it should also be
`understood that the teachings herein are also applicable
`to other, multi-subchannel, n-QAM systems.
`Referring to FIG. SA, a processing unit (102) re(cid:173)
`ceives an original information signal (101) from an in(cid:173)
`formation source. In this particular embodiment, this
`information signal constitutes a serial bit stream having
`an effective baud rate of 53.2 kilobits per second. This
`bit stream can represent, for example, true data, digi-
`25 tized voice, or other appropriate signals.
`The processing unit (102) functions to convert groups
`of 16 serial bits of the original information signal into
`four 16 QAM complex signal points (symbols). For
`example, FIG. 1 depicts a 16 QAM complex signal
`30 symbol constellation (2). Each symbol from the pro(cid:173)
`DETAILED DESCRIPTION OF THE
`cessing unit is a complex quantity, substantially within
`PREFERRED EMBODIMENT
`the constellation and represents a different combination
`FIG. 1 shows a constellation for a 16 QAM communi-
`of four serial bits from the information signal (101). For
`example, a first one of these symbols (201) represents
`cation system that is a map of 16 points on the complex
`plane defmed by a horizontal axis representing the real 35 the bits "0001." A second symbol (202), on the other
`hand, represents the bits "0100," all in accordance with
`portions, and a vertical axis representing imaginary
`portions, of a complex number. Transmitted QAM in-
`well understood prior art methodology.
`formation symbols on a communications channel, (and
`For each serially received 16 original information
`the pilot symbols as well) are discrete, packets of a
`bits, the processing unit (102) outputs, in parallel, on
`carrier signal modulated to convey information using 40 each of 4 signal paths (103 cc: 106), an appropriate repre-
`both the amplitude and phase-angle displacement of the
`sentative multibit symbol as described above. A pilot
`carrier from some reference. QAM information symbols
`insertion unit (107-110), located in each signal path
`are represented on the constellation of FIG. 1 as com-
`(103-106), inserts a predetermined symbol following
`receipt of 7 serially received information symbols from
`plex quantities represented as vectors having both mag-
`nitude (represented as length) and phase angles (which 45 the processing unit (102) pursuant to one embodiment of
`angles are measured with respect to one of the axes). In
`a communication methodology in accordance with the
`a 16 QAM system, having 16 different magnitude and
`invention. (Other embodiments of the invention would
`phase angle combinations that correspond to 16 differ-
`of course include pilot insertion more or less frequently
`ent possible bit patterns of four bindery digits, (which
`than once every 7 information symbols.) For each seri-
`bits are from a serial stream of bits from an information 50 ally received 16 original information bits, (from the
`source), each of the 16 points on the constellation is
`information signa1101) the processing unit (102) out-
`identified as representing one combination of four bits.
`puts, in parallel on each of the four signal paths
`A vector (10) (expressed in rectangular coordinates
`(103-106), an appropriate representative multibit sym-
`as 3+3j and having a length=(32+32)178 and a phase
`bol as described above.
`angle (12) equal to the arctan of 3/3 or forty five de- 55 A reduction in the ratio of peak output power level to
`grees with respect to the real axis), points to the point
`average power level in the composite output signal s(t)
`{3,3j} on the constellation, which point is shown in
`(500) can be achieved by pre-selecting, in advance, at
`FIG. 1 as representing the series of four binary digits,
`least the magnitude of, phase angle of, or both, for each
`(0,0,0,0). A second QAM symbol (14) points to yet
`of a plurality of the pilot symbols inserted by the pilot
`another point (1, -lj) in this constellation and repre- 60 insertion units (107, 108, 109, and 110). When these
`sents four other digital symbols.
`preselected pilots are added to form QAM subchannel
`From the foregoing, it can be seen that eight bits of
`symbols streams, (111, 112, 114, and 116) and they are
`information can be represented by two, 16-QAM sym-
`combined by the adder (400), (after they are pulse-shape
`boIs. When a digital information stream is converted to
`fIltered 120, 122, 124, and 126 and mixed 128, 130, 132,
`16 QAM, four-bit blocks of the data are mapped to the 65 and 134, with an appropriate injection signal 136, 138,
`various vectors that correspond to the bit pattern em-
`140, and 142 of the form e(2'fr/oJjKt), wherein j is the
`square root of negative one, t is time, and foj}kcomprises
`bodied in the four bits. When the QAM symbols that
`represent the digital information are transmitted, the
`an offset frequency corresponding to the kth composite
`
`6
`
`

`

`5,381,449
`
`5
`6
`the subchannel signals will have a single, or fIrst value
`signal) the composite signal (500) has a reduced peak to
`and these might be considered fIxed-pilot sub channel
`average powerl ratio.
`signals. In at least one other channel, channel B for
`The form of the pilot in the embodiment shown in
`example, using the method taught herein, during this
`FIG. SA is substantially represented by the quantity
`Pij= pi,;ei6i,j, where (Jij is the phase of the i'th subchan- 5 fInite time period (which period might be as short as one
`symbol time), will have a second value.
`nel pilot at symbol time j and in the preferred embodi-
`In many instances, (depending for example on the
`ment is empirically determined using a computer pro-
`gram, the source code of which is appended below. The
`number of symbol times in a frame and/or other factors)
`determination of the optimum value for the pilot sym-
`during a fInite length of time, the pilot symbols to be
`boIs in the preferred embodiment contemplated a fIxed 10 added to a fIrst subchannel at the pilot symbol times
`QAM symbols frame, which frame is graphically de-
`therefor, (sub channel A for example) can be selected
`picted in FIG. 4 as the time period of seven consecutive
`from a fIrst set of pilot values, and the pilots for a second
`QAM symbol times (including the pilot symbol time 1)
`sub channel signal be selected from a second set of pilot
`at a nominal carrier frequency (foin FIG. 1
`symbol values. The pilots from such a fIrst and second
`In the preferred embodiment, for a given magnitude 15 set of values might change in either their order of inser-
`of pilot, the optimum phase of the injected pilot symbols
`tion, their values, or both, to optimally reduce the peak
`to average ratio for the system. A key feature of the
`are calculated. (In a sense, both the magnitude and
`phase angle are "calculated" by the program. An initial,
`method herein is that in any subchannel, during some
`desired magnitude is supplied to the program by the
`finite period of time, pilot symbol values might vary
`user and, using this user-supplied quantity, the program 20 during successive pilot symbol times, in order to reduce
`the peak to average ratio, of the system, during such
`calculates optimum phase-angle values, assuming that
`length of time.
`all other symbols transmitted are zero. Accordingly, in
`the embodiment of the program below, to select differ-
`Referring to FIG. SA, since the QAM information
`symbols on lines 103-106 are complex valued quantities,
`ent phase angles and magnitudes, a user must select a
`different starting magnitude and re-calculate an opti- 25 and which represent data, (which data is substantially
`random over time) the combination of these complex
`mum phase angle value for the new magnitude). Alter-
`nate embodiments of the program contemplate includ-
`valued quantities by the summation circuit (400) (ignor-
`ing in the calculation of the phase angles and/or magni-
`ing momentarily the effect of the inserted pilot symbols)
`tudes of the pilots, the effects of the random nature of
`will have a varying peak power level to average power
`the QAM information streams (103, 104, 105, 106) upon 30 level. This might be appreciated by referring to the
`constellation map again depicted on FIG. 1. At any
`the peak to average power ratio in the composite signal
`particular instant, the output or anyone of the QAM
`s(t) (500).
`Referring to FIG. 4, there is graphically depicted, the
`information streams might have a QAM symbol that is
`placement and spacing of complex valued pilot symbols
`either identical with or differs from other QAM sym-
`that are combined with the complex valued QAM infor- 35 boIs on the other channels. Upon their combination at
`the summing circuit (400), over time they will have
`mation symbols (103-106 in FIG. SA) at predetermined
`symbol times, (A symbol time is typically the time dura-
`randomly varying peak power level.
`tion of one QAM symbol.) to produce a QAM subchan-
`In this invention by appropriate manipulation of the
`nel signal that is comprised of the complex valued QAM
`complex valued pilot symbols inserted by the pilot in-
`information symbols (103-106) and the complex valued 40 sertion units (107 and 110), the peak power level to
`pilots added by the pilot insertion units (107, 108, 109,
`average power level ratio in the composite signal s(t),
`and 110).
`(500) can be substantially reduced, when various proba-
`In the embedded pilot sequence shown in FIG. 4,
`bilistic factors of the data with which they are com-
`pilot symbols are shown being added in subchannels A
`bined with are considered. Improvements in the peak to
`and D, in frame 1, at symbol time 1. Pilot symbols are 45 average power ratio of 1.5 dB have been realized in at
`added to the subchannels B and C at symbol time 3 in
`least one embodiment of the invention.
`frame number 1. (pilots inserted into subchannels dur-
`Referring to FIG. SB a simplifJed graphical represen-
`ing the same symbol time are considered to be time
`tation of the pilot insertion unit (107-110 in FIG. SA
`coincident.) In time frame number 2, which depicts an
`and which is implemented in a digital signal processor).
`alternate implementation of the pilot symbol insertion, 50 The improved pilot insertion unit produces pilot sym-
`boIs that are added to the QAM symbol stream and
`time coincident pilot symbols might just as well be
`inserted during a single symbol time 1 wherein insertion
`which can vary substantially anywhere on or off the 16
`QAM constellation map shown in FIG. 1. In addition to
`of the pilots are all time coincident with respect to each
`other. Frame 3 shows yet another QAM frame, having
`having the time-coincident pilots of multiple subchan-
`more than seven symbol times and having time coinci- 55 nels being different, the magnitude and/or phase of the
`pilot symbols in each subchannel (A-D for example),
`dent pilots added to at least two channels, every seven
`symbol times. For purposes of this invention it is prefer-
`can vary from pilot symbol time to pilot symbol time in
`able that at least two subchannels have so called time
`that subchannel.)
`In this invention, while legitimate QAM information
`coincident pilots and it is yet even more preferable that
`all sub channels have time coincident pilots but alternate 60 symbols will of necessity have to be mapped to one of
`embodiments would contemplate using time coincident
`the 16 constellation points, it is expected using the meth-
`pilots in two or more subchannels to further manipulate
`odology of this invention that at least one of the pilot
`the peak to average power ratio.
`symbols that are combined with the QAM information
`During a flnite time period, which period might even
`streams (103-106) will not fall on a valid constellation.
`be considered the sum of the frame times for frames 1, 65 point. Instead, the invention, (which includes the imple-
`2, and 3, at least one pilot symbol, in at least one sub-
`mentation of the apparatus shown in FIG. SA) contem-
`channel, will have a value different from the values of
`plates pilots such as those shown in FIG. 6 identifled
`the other pilots. It should be expected that a plurality of
`and depicted by reference numerals 84, 86 and 88 which
`
`7
`
`

`

`5,381,449
`
`7
`pilots do not fallon valid constellation points, but
`which when combined with the QAM information
`streams (103-106) to form piloted QAM subchannel
`symbol streams (123, 125, 127, and 129), can when com(cid:173)
`bined with the other complex values on the other sub- 5
`channels to form a composite signal (500) substantially
`reduce the peak to average power ratio of the compos(cid:173)
`ite signal (500). (in addition to using the non-constella(cid:173)
`tion-based pilots 84, 86, and 88, constellation based
`pilots, such as pilot 90, when appropriately identified, 10
`can at appropriate times, also be used to reduce peak to
`average power ratio.) Both the constellation-based pilot
`symbols and the non-constellation based pilot symbols
`are considered to be pre-selected and pre-determined
`symbols. By virtue of the fact that the pilots occur at 15
`discrete instants of time, i.e. one or more QAM symbol
`times they are considered time domain pilots. In so far
`as the pilots are spread across the frequencies of the
`different subchannels, each subchannel of which is cen(cid:173)
`tered about a different frequency, the pilots are also 20
`considered to have a frequency domain characteristics.
`In this sense, the pilots can be considered to be both
`time and frequency domain, and both non-constellation
`and constellation based, predetermined, complex- 25
`valued pilots.
`In the preferred embodiment, the phase angle selec(cid:173)
`tion for the pilots is accomplished by means of a com(cid:173)
`puter program which is depicted in the attached appen(cid:173)
`dix. It should be appreciated that in other embodiments, 30
`a plurality of the pilots might have either their ampli(cid:173)
`tudes and/or their phase angles selected such that when
`combined with a probablistic determination of permissi(cid:173)
`ble QAM information symbols minimizes the peak to
`average power ratio of the composite signal over some 35
`fInite length of time. In the embodiment of the appara(cid:173)
`tus shown in FIG. SA, at least one of the complex val(cid:173)
`ued pilot symbols that are added to the QAM informa(cid:173)
`tion symbols by the pilot insertion units (107, 108, 109,
`
`8
`and 110) does not lie on the permissible constellation
`points depicted in either FIGS. 1 or 5.
`An exemplary selection process to select these prede(cid:173)
`termined pilot symbols is described below. The embodi(cid:173)
`ment of program below assumes:
`1) a four-subchannel QAM system, having 4, time
`coincident pilot symbols (220), such as those shown
`in frame 2 of FIG. 4.
`2) that the DSP, in which the apparatus (300) shown
`in FIG. SA is embodied, has a simulation sampling
`rate, Fs, (i.e. samples the composite signal s(t» at 36
`times the composite symbol rate.
`3)
`that the subchannel center frequencies are
`Ct)2 = 2'lT( -IFsl64);
`Ct)1 =2'lT( -3Fsl64);
`Ct)3=2'lT(1Fsl64); Ct)4=2'lT(3Fsl64).
`The user supplies the program: a slot format me con(cid:173)
`sistent with assumption 1 above; a fInite impulse re(cid:173)
`sponse mter me, which defInes the coefficients of the
`sub channel pulse shape mters 120, 122, 124, 126; the
`symbol time position of the pilot in a slot for which to
`carry out the pilot-phase set search; the desired magni(cid:173)
`tude (squared) of the pilot symbol, (referred to in the
`program as sync symbols); the number of steps around
`the unit circle at which to calculate s(t) for each of the
`four time coincident pilots for which the search is being
`carried out.
`The programs' output is a set of phase angles for (i.e.
`the e's shown in FIG. 5B) at which the peak transmitter
`output power is smallest, after calculating peak trans(cid:173)
`mitter output power levels for all the possible combina(cid:173)
`tions of e, in the step sizes specmed by the user.
`While this embodiment of the invention contemplates
`a particular pilot configuration, i.e. as shown in time
`frame 2 of FIG. 4, other embodiments contemplate
`other slot/pilot confIgurations, such as that shown in
`frame 1 of FIG. 4 for example. A similar program could
`well be written by those skilled in the art for any other
`slot/pilot confIguration.
`
`fdefine
`
`EXTERN
`
`finclude <stdio.h>
`
`finclude <math.h>
`
`finclude "readlib.h"
`
`finclude "defcmath.h"
`
`finclude "defdsp.h"
`
`main ()
`..
`
`{
`
`/*
`* Transmitter variables and pointers
`*/
`
`8
`
`

`

`double
`
`(sps)
`
`double
`
`9
`fsymb:
`*/ ....
`
`tsymb;
`
`(sec per symb) */
`
`5,381,449
`
`10
`/* QAM sub-channel symbol rate
`
`/* QAM sub-channel symbol period
`
`double
`
`ftx:
`
`/* sub-channel TX pulse shape filter
`
`sampling
`
`*/
`
`/*
`
`rate (ksps)
`
`*/
`
`int
`
`nzst; .
`
`/* zero stuff ratio in sub-channel·
`
`*1"
`
`double
`
`ttx;
`
`/* l/ftx (seconds per sample)
`
`*/
`
`int
`
`nsub;
`
`/* number of QAl-1 sub-channels in system
`
`*/
`
`double
`
`fsc;
`
`/* sub-channel frequency spacing (Hz)
`
`double
`
`*/
`wcent[4];
`
`(nautral)
`
`* /
`
`1* sub-channel center frequencies
`
`double
`for sub-c':l * /
`tX.J'hase[4];
`double
`
`*/
`
`double
`
`pS.J'ow:
`impulse resp.
`
`*/
`
`double
`
`output
`
`COMPLEX
`
`tx.J'ow;
`
`*/
`
`*/
`
`1* phase step in one sample period
`
`/* phase of sub-channel
`
`1* power in p~l~e shape filter
`
`1* average power at trans~~tter
`
`/* transmitter output
`
`COMPLEX
`
`qam_symb;
`
`1* complex QAM symbol variable
`
`COl-'.PLEX
`
`*/
`COMPLEX
`
`*/
`sub_ch[4]:
`
`/* sub-channel data vector
`
`ps_out[4] [144];
`
`1* sub-channel data vector
`
`*/
`
`int
`
`outs:
`
`.1* .number of filter outputs
`*1
`r .
`/* pointer to pulse shaping TX filter
`IDFIR *t~.J'ulse(4J:
`
`strcts
`
`*/
`
`...
`
`double
`
`9
`
`

`

`double
`
`double
`
`11
`
`maxyow;
`
`5,381,449
`
`12
`
`int
`
`/* slot length in sub-channel symbols
`
`*/
`FILE *slot_def;
`
`/* slot definition file pointer
`
`*/
`slot_symb[lOO];
`
`int
`
`*/
`
`/* slot symbol definition array
`
`double
`double
`
`ampydf[600];
`pdf_cnt;
`
`int
`
`pdf_index;
`
`long cnt;
`
`long sync_length;
`
`/* number of samples simulated per sync
`
`pattern
`
`*/
`
`long steps;
`
`variable
`
`*/
`
`/* number of phase ~t~ps for each
`
`phase_step; /* 2pi/steps
`
`*/
`
`phil;
`
`phi2;
`
`thetal;
`
`theta2;
`
`expyl;
`
`expy2;
`( .
`exp_tl;
`
`exp_t2;
`
`mc_exp_tl;
`
`int
`
`double
`
`*/
`
`double
`
`double
`
`double
`
`double
`double ,
`double
`COMPLEX
`
`COMPI.EX
`
`COMPLEX
`
`COMPLEX
`
`COMPLEX
`
`COMPLEX
`
`COMPLEX
`
`double
`
`best [100] [6] ;
`
`10
`
`

`

`5,381,449
`
`14
`
`13
`
`int
`
`numb-pilot;
`
`int .. numb_data;
`
`int
`
`int
`
`int
`
`numb_dUIn-P;
`
`numb_dUIn_d;
`
`slot_cnt;
`
`double
`
`inv_np;
`
`COMPLEX
`FILTER
`
`sync[lOOl];
`*mf[lOO];
`
`COMPLEX
`
`double
`
`double
`
`max [2] ;
`
`buf[3];
`
`int
`
`pilot[lOO];
`
`/*
`
`* general purpose variables and pointers
`
`*/
`
`char strl[60];
`char str2[60];
`..
`char str3[60];
`
`double
`
`double
`.. double
`
`double
`
`double
`
`/* general purpose string buffer
`
`/* general purpose string buffer
`
`*1
`*/
`
`/* general purpose string buffer
`*/
`1* general purpose temporary variable
`
`temp;
`*/ '
`
`r .
`temp2;
`
`temp3;
`pi2;
`
`pi~.
`
`1* 2.0*pi
`
`*/ .
`
`long
`
`i,j,k,l,m,n;
`
`/* general purpose counters
`
`*/
`
`COMPLEX
`
`c_temp;
`
`variable
`
`*/
`
`COMPLEX
`
`COMPLEX
`
`COMPLEX
`
`c_templ:
`
`c_temp2;
`
`c_temp3;
`
`1* complex general purpose temporary
`
`11
`
`

`

`COMPLEX
`
`15
`cone;
`
`*/
`
`5,381,449
`
`16
`
`/* 1 + jO (complex one)
`
`COMPLEX
`
`c zero;
`
`/* 0 + jO (complex zero)
`
`*/
`FILE *fr, *fr1, *fr2, *fw, *fwl, *fw2;
`c one. real
`1.0;
`
`c_one.imag
`
`0.0;
`
`c zero. real = 0.0;
`
`c_zero.imag = 0.0;
`
`pi2 = 8.0*atan(1.0);
`
`pi = 4.0*atan(1.0);
`
`nsub
`
`4;
`
`fsymb
`
`4000.0;
`
`tsymb
`
`1.0/fsymb;
`
`ftx
`
`144.0;
`
`ttx
`
`1.0/(ftx*1000.0)i
`
`nzst = 36;
`..
`
`fsc
`
`ftx*1000.0/32.0;
`
`temp = -(nsub -.1)*(fsc/2.0)i
`for ( i = 0; i < nsub; i++ )
`
`wcent[i] = pi2*tempi
`
`txyh_st [~.~ = wcent [i] *ttx;
`
`temp += fsc;
`
`again22:
`
`printf("\nEnter·the slot definition filename.\nlt);
`
`scanf("%s", str3);
`
`slot_def = fopen(str3, "r")j
`
`if
`
`slot def == NULL )
`
`printf("\nFILE ACCESS ERROR, TRY AGAIN.\n"};
`
`gata again22i
`
`12
`
`

`

`17
`
`5,381,449
`
`18
`
`f.scanf (slot_def, " %d", &slot_len);
`slot len > 90
`
`if
`
`printf("\nMAXIMUM SLOT LENGTH OF 90 SYMBOLS EXCEEDED, TRY
`
`AGAIN. In") ;
`
`goto ag