USRE41380E1

Movatterモバイル変換

Info

Publication number: USRE41380E1
Application number: US10/718,507
Authority: US
Inventors: Ronald D. McCallister; Bruce A. Cochran; Bradley P. Badke
Original assignee: Sicom Inc
Current assignee: Intersil Americas LLC
Priority date: 1998-08-28
Filing date: 2003-11-19
Publication date: 2010-06-15
Anticipated expiration: 2018-08-28
Also published as: ATE518347T1; JP4280422B2; WO2000013388A1; EP1108318B1; CA2341439A1; CN1324539A; CN1150730C; AU5334099A; EP1108318A1; JP2002524923A; EP1108318A4; US6104761A; USRE43963E1

Abstract

A constrained-envelope digital-communications transmitter circuit (22) in which a binary data source (32) provides an input signal stream (34), a phase mapper (44) maps the input signal stream (34) into a quadrature phase-point signal stream (50) having a predetermined number of symbols per unit baud interval (64) and defining a phase point (54) in a phase-point constellation (46), a pulse-spreading filter (76) filters the phase-point signal stream (50) into a filtered signal stream (74), a constrained-envelope generator (106) generates a constrained-bandwidth error signal stream (108) from the filtered signal stream (74), a delay element (138) delays the filtered signal stream (74) into a delayed signal stream (140) synchronized with the constrained-bandwidth error signal stream (108), a complex summing circuit (110) sums the delayed signal stream (140) and the constrained-bandwidth error signal stream (108) into a constrained-envelope signal stream (112), and a substantially linear amplifier (146) amplifies the constrained-envelope signal stream (112) and transmits it as a radio-frequency broadcast signal (26).

Description

TECHNICAL FIELD OF THE INVENTION

The present invention relates generally to the field of digital communications. More specifically, the present invention relates to the field of constrained-envelope digital transmitter circuits.

BACKGROUND OF THE INVENTION

A wireless digital communications system should ideally refrain from using any portion of the frequency spectrum beyond that actually required for communications. Such a maximally efficient use of the frequency spectrum would allow the greatest number of communications channels per given spectrum. In the real-world, however, some spectral regrowth (i.e., increase in spectral bandwidth) is inevitable due to imperfect signal amplification.

In wireless communication systems various methodologies have been used to minimize spectral regrowth. Some conventional methodologies utilize complex digital signal processing algorithms to alter a digitally modulated transmission signal in some manner conducive to minimal spectral regrowth. Such complex algorithmic methodologies are well suited to low-throughput applications, i.e., those less than 0.5 Mbps (megabits per second), such as transmission of vocoder or other audio data. This is because the low throughput rate allows sufficient time between symbols for the processor to perform extensive and often repetitive calculations to effect the required signal modification. Unfortunately, high-throughput applications, i.e., those greater than 0.5 Mbps, such as the transmission of high-speed video data, cannot use complex processing algorithms because the processing power required to process the higher data rate is impractical.

A digital signal processing methodology may be used with the transmission of burst signals. With burst transmissions, the interstitial time between bursts may be used to perform the necessary complex computations based upon an entire burst. This methodology is not practical when continuous (as opposed to burst) transmission is used.

A conventional form of post-modulation pulse shaping to minimize spectral bandwidth utilizes some form of Nyquist-type filtration, such as Nyquist, root-Nyquist, raised cosine-rolloff etc. Nyquist-type filters are desirable as they provide a nearly ideal spectrally constrained waveform and negligible inter-symbol interference. This is achieved by spreading the datum for a single constellation phase point over many unit baud intervals in such a manner that the energy from any given phase-point datum does not interfere with the energy from preceding and following phase-point data at the appropriate baud-interval sampling instants.

The use of Nyquist-type filtration in a transmission circuit produces a filtered signal stream containing a pulse waveform with a spectrally constrained waveform. The degree to which a Nyquist-type pulse waveform is constrained in bandwidth is a function of the excess bandwidth factor, α. The smaller the value of α, the more the pulse waveform is constrained in spectral regrowth. It is therefore desirable to have the value of α as small as possible. However, as the value of α is decreased, the ratio of the spectrally constrained waveform magnitude to the spectrally unconstrained waveform magnitude is increased. The spectrally unconstrained waveform is the waveform that would result if no action were taken to reduce spectral regrowth. Typical designs use a α values of 0.15 to 0.5. For an exemplary a α value of 0.2, the magnitude of the spectrally constrained waveform is approximately 1.8 times that of the unconstrained waveform. This means that, for a normalized spectrally unconstrained waveform magnitude power of 1.0, the transmitter output amplifier must actually be able to provide an output power of 3.24 (1.8²) to faithfully transmit the spectrally constrained waveform. This poses several problems.

When the transmitter output amplifier is biased so that the maximum spectrally unconstrained waveform (1.0 normalized) is at or near the top of the amplifier's linear region, all “overpower” will be clipped as the amplifier saturates. Such clipping causes a marked increase in spectral regrowth, obviating the use of Nyquist-type filtration.

When the transmitter output amplifier is biased so that the maximum spectrally constrained waveform (1.8 normalized) is at or near the top of the amplifier's linear region, the spectrally unconstrained waveform is at only 56 percent (i.e., 1/1.8) of the amplifiers peak linear power. This results in an inefficient use of the output amplifier.

Also, the biasing of the transmitter output amplifier so that the spectrally constrained waveform is at or near the top of the amplifier's linear region requires that the output amplifier be of significantly higher power than that required for the transmission of a spectrally unconstrained waveform. Such a higher-power amplifier is inherently more costly than its lower-power counterparts.

SUMMARY OF THE INVENTION

It is an advantage of the present invention that a circuitry and a methodology are provided that allow a transmitter output amplifier to be biased so that the spectrally unconstrained waveform is at or near the top of the amplifier's linear region without incurring clipping of a spectrally constrained waveform.

It is another advantage of the present invention that a circuitry and methodology are provided that allow a spectrally constrained waveform to have approximately the same magnitude as the spectrally unconstrained waveform without effecting a significant increase in spectral regrowth.

It is another advantage of the present invention that a circuitry and methodology are provided which allow a spectrally constrained waveform to be utilized with a continuous transmission scheme.

It is another advantage of the present invention that a circuitry and methodology are provided which allow efficient use of a transmitter output amplifier, thus allowing higher power output for a given output amplifier and a given bandwidth constraint than was previously feasible.

It is another advantage of the present invention that a circuitry and methodology are provided which allow efficient use of a transmitter output amplifier, which allows allowing a lower-power amplifier to be used for achieving given bandwidth constraints than was previously feasible, thus effecting a significant saving in the cost thereof.

These and other advantages are realized in one form by a constrained-envelope digital communications transmitter circuit. The transmitter circuit has a pulse-spreading filter configured to receive a quadrature phase-point signal stream of digitized quadrature phase points and produce a filtered signal stream, which filtered signal stream exhibits energy corresponding to each phase point spread throughout a plurality of baud intervals. The transmitter circuit also has a constrained-envelope generator coupled to the pulse-spreading filter and configured to produce a constrained-bandwidth bandwidth error signal stream. The transmitter circuit also has a combining circuit coupled to the pulse-spreading filter and to the constrained-envelope generator, which combining circuit is configured to combine the filtered signal stream and the constrained-bandwidth error signal stream to produce a constrained-envelope signal stream. The transmitter circuit also has a substantially linear amplifier with an input coupled to the combining circuit.

These and other advantages are realized in another form by a method for the transmission of a constrained-envelope communications signal in a digital communications system. The transmission method includes the step of filtering a quadrature phase-point signal stream to produce a filtered signal stream, which filtering step spreads energy from each phase point over a plurality of baud intervals. The transmission method also includes the step of generating a constrained-band error signal stream from the filtered signal stream and a threshold signal. The transmission method also includes the step of combining the filtered signal stream and the constrained-bandwidth error signal stream to produce a constrained-envelope signal stream. The transmission method also includes the step of linearly amplifying the constrained-envelope signal stream to produce the constrained-envelope communications signal. The transmission method also includes the step of transmitting the constrained-envelope communications signal.

BRIEF DESCRIPTION OF THE DRAWINGS

A more complete understanding of the present invention may be derived by referring to the detailed description and claims when considered in connection with the Figures, wherein like reference numbers refer to similar items throughout the Figures, and:

FIG. 1 depicts a simplified block diagram of a digital communications system in accordance with a preferred embodiment of the present invention;

FIG. 2 depicts a block diagram of a constrained-envelope digital communications transmitter circuit in accordance with a preferred embodiment of the present invention;

FIG. 3 depicts a 16-P-APSK constellation illustrating a locus of a quadrature phase-point signal stream over twelve exemplary consecutively mapped phase points in accordance with a preferred embodiment of the present invention;

FIG. 4 depicts a plurality of signal streams in accordance with a preferred embodiment of the present invention;

FIG. 5 depicts the phase-point constellation ofFIG. 3 illustrating an exemplary locus of a filtered signal stream over the twelve consecutively mapped phase points ofFIG. 3 in accordance with a preferred embodiment of the present invention;

FIG. 6 depicts a pair of Nyquist-type data bursts in accordance with a preferred embodiment of the present invention; and

FIG. 7 depicts a noise-influenced constellation illustrating constrained-envelope phase-point probabilities of the phase points of the constellation ofFIG. 3 in accordance with a preferred embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

FIG. 1 depicts a simplified block diagram of adigital communications system20 andFIG. 2 depicts a block diagram of a constrained-envelope digitalcommunications transmitter circuit22 in accordance with a preferred embodiment of the present invention. The following discussion refers toFIGS. 1 and 2.

Digital communications system

20, as depicted inFIG. 1, includes atransmitter circuit22 and atransmitter antenna24 together configured to modulate and transmit a radio-frequency (RF)broadcast signal26 to areceiver antenna28 and areceiver circuit30, together configured to receive and demodulateRF broadcast signal26. Those skilled in the art will appreciate that the embodiment ofsystem20 depicted is a simplistic one for purposes of discussion only. In normal use,system20 would likely be a complex system consisting of many more components and broadcast signals. It will be appreciated that the use of such a complex communications system forsystem20 in no way departs from the spirit of the present invention or the scope of the appended claims.

Transmitter circuit

22 has abinary data source32 providing a binaryinput signal stream34.Binary data source32 may be any circuitry, device, or combination thereof producinginput signal stream34.Input signal stream34 is made up of binary data that may be pre-encoded in any desired manner. That is,input signal stream34 may be made up of data that has no encoding, concatenated encoding, Reed-Solomon block encoding, or any other form of encoding desired for or required of the communications scheme in use.

In the preferred embodiment,input signal stream34 is a steam of continuous data (as contrasted with burst data) passing to an input of aconvolutional encoder36.Convolutional encoder36 convolutionally encodes (e.g., Viterbi encodes)input signal stream34 into an encodedsignal stream38. The use ofconvolutional encoder36 intransmitter circuit22 and a like convolutional decoder (not shown) inreceiver circuit30 significantly reduces the error rate of the overall signal in a manner well understood by those skilled in the art. However,convolutional encoder36 may be omitted.

Interleaver

40 temporally decorrelates encodedsignal stream38 to produce an interleavedsignal stream42. That is, the symbols making up the binary signal stream are temporally decorrelated (i.e., separated) intransmitter circuit22 and temporally correlated inreceiver circuit30. This is done so that correlated errors produced by downstream transmitter components, discussed hereinbelow, will then be decorrelated through a complimentary de-interleaver located inreceiver circuit30 before convolutional decoding inreceiver circuit30.

In the preferred embodiment, interleavedsignal stream42 passes to an input of aphase mapper44. Those skilled in the art will appreciate thatinterleaver40 is not desired in all embodiments oftransmitter circuit22, for example whenconvolutional encoder36 is omitted. When interleaver40 is omitted, encodedsignal stream38 is passed directly to the input ofphase mapper44. When bothconvolutional encoder36 andinterleaver40 are omitted, binary input signal stream passes directly to the input ofphase mapper44.

FIG. 3 depicts a sixteen phase-point polar amplitude and phase shift keying (16-P-APSK)constellation46 illustrating alocus48 of a quadrature phase-point signal stream50 (FIG. 2) over twelve exemplary sequential phase points52 in accordance with a preferred embodiment of the present invention. The following discussion refers toFIGS. 2 through 3.

Phase mapper

44 maps symbols (i.e., binary data units) present in interleavedsignal stream42, encodedsignal stream38, orinput signal stream34, into phase points54 in phase-point constellation46. Whileconstellation46 is depicted inFIG. 3 as a 16-P-APSK constellation, those skilled in the art will appreciate that the circuitry and methodology of the present invention may be applied to all forms of constellations. The present invention is especially beneficial when used with constellations having rings of different magnitudes, i.e., amplitude and phase-shift keying (APSK) constellations. This is true because APSK constellations, requiring amplitude modulation of the signal, desirably use linear amplifiers to reproduce that amplitude modulation.

Eachphase point54 inconstellation46 represents a plurality, in this example four, of symbols. The values of the symbols in a givenphase point54 determine the location of thatphase point54 withinconstellation46 in a manner well known to those skilled in the art.

Eachquadrature phase point54 may be thought of as having a vector value expressed as I,Q in the Cartesian coordinate system, where I is the in-phase (abscissa) value and Q is the quadrature (ordinate) value of the vector, or expressed as M,φ in the polar coordinate system, where M is the magnitude and φ is the phase angle of the vector. In this discussion, the M,φ designation will be used throughout, as the vector magnitude is the most discussed vector component.

In the exemplary 16-P-APSK constellation46 ofFIG. 3, eachphase point54 resides upon anouter ring56 or aninner ring58. Phase-points54 residing uponouter ring56 are outer-ring or maximum-magnitude phase points60. That is, outer-ring phase points60 have a maximum magnitude (maximum value of M) as represented by the radius ofouter ring56. For purposes of discussion, the magnitudes of outer-ring phase points60 are normalized to 1.00.

Inner-ring phase points62, i.e., those phase points54 residing uponinner ring58, have a lesser magnitude as represented by the radius ofinner ring58. For the exemplary 16-P-APSK constellation46 depicted inFIG. 3, the magnitudes of inner-ring phase points62 may desirably be approximately 0.63 when outer-ring phase point60 magnitudes are normalized to 1.00.

FIG. 4 depicts a plurality of signal streams, in accordance with a preferred embodiment of the present invention. The following discussion refers toFIGS. 2 through 4.

The output ofphase mapper44 is phase-point signal stream50.Phase mapper44 processes onephase point54 perunit baud interval64. That is, phase-point signal stream50 consists of a series of consecutive phase-point pulses66, each of which represents onephase point54, whose leading edges are oneunit baud interval64 apart. Those skilled in the art will appreciate that other embodiments of phase-point signal stream50 are equally valid, that the embodiment utilized is dependent upon the circuitry producing and processing phase-point signal stream50, and that the use of other embodiments of this or any other signal stream does not depart from the spirit of the present invent nor the scope of the appended claims.

FIGS. 3 and 4 illustrate a series of twelve exemplary sequential phase points52, representative of a random data stream processed by transmitter circuit22 (FIG.2). These twelve exemplary phase points52 reside at temporally consecutive locations labeled t₀, t₁, t₂, t₃, t₄, t₅, t₆, t₇, t₈, t₉, t₁₀, and t₁₁. These labels represent sequential integral times atunit baud intervals64, i.e., integral-baud times, and indicate the leading-edge times of phase-point pulses66. For purposes of simplification within this discussion, any occurrence at time t_Nshall be referred to as “occurrence t_N”. For example, anexemplary phase point52 occurring at time t₂shall be referred to as phase point t₂, and the associated phase-point pulse66 whose leading edge occurs at time t₂shall be referred to as phase-point-signal pulse t₂. In other words, at time t₂, phase point t₂is clocked and phase-point-signal pulse t₂begins. Oneunit baud interval64 later, at time t₃, phase point t₃is clocked and phase-point pulse t₃begins. This process continues indefinitely, with twelve exemplary phase points t₀through t₁₁depicted in FIG.3 and twelve corresponding phase-point-signal pulses t₀through t₁₁depicted in phase-point signal stream50 of FIG.4.

Table 1 below illustrates the magnitudes for phase-point-signal pulses to through t₁₁.

TABLE 1

Phase-Point Pulse Magnitudes

	Phase-Point-Signal
	Pulse	Magnitude

	t₀	Outer-Ring 68
	t₁	Inner-Ring 70
	t₂	Outer-Ring 68
	t₃	Outer-Ring 68
	t₄	Inner-Ring 70
	t₅	Outer-Ring 68
	t₆	Outer-Ring 68
	t₇	Outer-Ring 68
	t₈	Outer-Ring 68
	t₉	Inner-Ring 70
	t₁₀	Outer-Ring 68
	t₁₁	Inner-Ring 70

Phase point t₀is an outer-ring phase point60. Phase-point-signal pulse to therefore has an outer-ring magnitude68. In like manner, phase point t₁is an inner-ring phase point62 and phase-point-signal pulse t₁has an inner-ring magnitude70.

Phase-point signal stream50

effects locus

48 throughconstellation46.Locus48 coincides with the location of each exemplary phase point t₀through t₁₁in turn atunit baud intervals64. InFIG. 3,locus48 is depicted as effecting a minimum distance (straight line) path between adjacent exemplary phase points52. Those skilled in the art will appreciate thatlocus48 is so depicted solely for the sake of simplicity, and that in actual practice,locus48 instantly jumps or snaps between exemplary phase points52 in a discontinuous manner.

FIG. 5 depicts an expanded phase-point constellation46′ illustrating alocus72 of a filtered signal stream74 (FIG. 2) over twelve exemplary sequential phase points52 in accordance with a preferred embodiment of the present invention. The following discussion refers toFIGS. 2 through 5.

In the preferred embodiment, phase-point signal stream50 passes to the input of a pulse-spreadingfilter76, preferably realized as a Nyquist-type filter, such as a Nyquist, root-Nyquist, raised cosine-rolloff, etc., filter. Pulse-spreadingfilter76 filters phase-point signal stream50 into filteredsignal stream74, depicted in FIG.5. In orthogonal frequency division multiplex (OFDM) systems, also known as multitone modulation (MTM) systems, pulse-spreadingfilter76 may be implemented using a transmultiplexer or equivalent circuitry.

In accordance with Shannon's theory, well known to those skilled in the art, pulse-spreadingfilter76 produces at least two (only two in the preferred embodiment) output filtered-signal pulses78, i.e., complex samples offiltered signal stream74, for each input phase-point pulse66 received. This is demonstrated inFIG. 4 where filteredsignal stream74 possesses two filtered-signal pulses78 perunit baud interval64. In the preferred embodiment, filtered-signal pulses78 consist of alternating on-time pulses80, i.e., samples of filtered signal stream at integralunit baud intervals64, and off-time pulses82, i.e., samples offiltered signal stream74 between integral unit baud intervals. In effect, filteredsignal stream74 is made up of two interleaved data streams, an on-time signal stream84 and an off-time signal stream86.

On-time signal stream84 is substantially a version of phase-point signal stream50, wherein each phase-point pulse66 has been reduced in duration from oneunit baud interval64 to a half-unit baud interval88 to become on-time pulse80 while maintaining substantially the same relative leading-edge time. That is, filtered-signal pulse to has substantially the same magnitude and substantially the same leading edge time as phase-point pulse to with approximately one-half the duration. Of course, those skilled in the art will appreciate that signal streams74 and84 may be delayed fromsignal stream50 by a delay imposed byfilter76.

The generation of both on-time pulses80 and off-time pulses82 by pulse-spreadingfilter76 effectively populates expandedconstellation46′ (FIG. 5) with on-time phase-points90 (circles) and off-time phase points92 (squares). The original phase points54 of constellation46 (FIG.3), i.e., the phase points carrying the intelligence to be communicated bytransmitter circuit22, are on-time phase points90 of expandedconstellation46′.

Added to expandedconstellation46′ are off-time phase points92, with each off-time phase-point92 occurring approximately midway in time between consecutive on-time phase points90. Therefore, exemplary sequential phase points52 become exemplary filtered phase points94. Exemplary filtered phase points94 are made up of alternating exemplary on-time filtered phase points96 and exemplary off-time filtered phase points98, and reside at temporally consecutive locations labeled t₀, t_0.5, t₁, t_1.5, t₂, t_2.5, t4, t_3.5, t₄, t_4.5, t₅, t_5.5, t₆, t_6.5, t₇, t_7.5, t₈, t_8.5, t₉, t_9.5, t₁₀, t_10.5, and t₁₁, InFIG. 5, exemplary on-time filtered phase points96 are located at integral-baud times (t₀, t₁, t₂, etc.), whereas exemplary off-time filtered phase points98 are located at fractional-baud (non-integral-baud) times (t_0.5, t_1.5, t_2.5, etc.).

The generation of off-time phase points92 approximately midway in time between consecutive on-time phase points90 causes filteredsignal locus72 to effect excursions havinglocal peak magnitudes99 greater than outer-ring magnitude68. Such excursions occur because the immediate position oflocus72 at any given instant in time is not only a result of those phase points54 proximate that position, but of a plurality of phase points54 both preceding and following that instant in time. That is, in the preferred embodiment, the determination of the position oflocus72 at time t_2.5(i.e., coincident with off-time phase point t_2.5) is determined not only by the positions of phase points t₂and t₃, but by the positions of numerous phase points54 preceding phase point t_2.5(i.e., phase points t₂, t_1.5, t₁, t_0.5, etc.) and the positions of numerous phase points54 following phase point t_2.5(i.e., phase points t₃, t_3.5, t₄, t_4.5, etc.)

This phenomenon is illustrated inFIG. 6, which depicts a pair of Nyquist-type datum bursts100 in accordance with a preferred embodiment of the present invention. The following discussion refers toFIGS. 2,4,5, and6.

In the preferred embodiment, pulse-spreadingfilter76 is realized as a Nyquist-type filter. Therefore, when a single phase-point pulse66 is filtered by pulse-spreadingfilter76, thatsingle pulse66 is transformed into a Nyquist-type datum burst100 extending over a plurality ofunit baud intervals64. It is a property of Nyquist-type filters that datum burst100 attains a datum-burst peak value102 (i.e., a local peak magnitude) at the primary sampling time of the specific phase-point pulse66 (i.e., at time t₂for phase-point pulse t₂), and attains a zero datum-burst value104 (i.e., is equal to zero) at integralunit baud intervals64 preceding and following peak datum-burst value102 (i.e., at times . . . , t₋₁, t₀, t₁, and t₃, t₄, t₅, . . . , for phase point pulse t₂). In this manner, the energy of eachpulse78 is spread over a plurality ofbaud intervals64 preceding and following the clocking instant (time t₂).

FIG. 6 illustrates Nyquist-type datum bursts100 for phase-point pulses t₂and t₃, with datum burst t₂depicted as a solid line and datum burst t₃depicted as a dashed line. As an example, it may be seen fromFIG. 6 that at time t₂the value of datum burst t₂is peak datum-burst value102. At every other time separated from time t₂by an integral number ofunit baud intervals64, the value of datum burst t₂is zero. An analogous condition occurs for datum burst t₃.

The value oflocus72 is, at each moment in time, the sum of all datum bursts100 at that moment. In the simplified two-datum-burst example ofFIG. 6,locus72, depicted by a dotted line, is the sum of datum burst t₂and datum burst t₃. Since datum bursts t₂and t₃are zero at each integral time t_Nexcept times t₂and t₃, the value oflocus72 is also zero except at times t₂and t₃, where it assumes the peak values of datum bursts t₂and t₃, respectively.

The value oflocus72 at any instant in time between integral-baud times is the sum of the values of all datum bursts100 at that instant. For example, inFIG. 6 where only two datum bursts100 are considered,locus72 has a value at time t_2.5that is the sum of the values of datum bursts t₂and t₃at time t_2.5. Since datum bursts t₂and t₃both have significant positive values at time t_2.5,locus72 has a value significantly greater than the maximum values of either datum burst t₂or datum burst t₃.

Sincelocus72 describes the sum of all datum bursts100,locus72 is a function of the shape of the curves (FIG. 6) describing those datum bursts100. That is,locus72 is a function of a filtered-signal peak magnitude component of a filtered-signal complex digital value at any given point. The shape of the datum-burst curve is a function of the excess bandwidth factor, α, a design property of pulse-spreadingfilter76. The smaller the value of α, themore locus72 may rise above the peak datum burstvalues102 of adjacent datum bursts100. Typical designs of pulse-spreadingfilters76 use α values of 0.15 to 0.5. For like-valued adjacent phase points54 and an α value of 0.2, a maximum excursion magnitude105 (i.e., the potentiallocal peak magnitude99 of locus72) is approximately 1.8 times the value of the maximum phase-point magnitude. That is, the magnitude of the constrained envelope is approximately 1.8 times that of the unconstrained envelope. In the preferred embodiment depicted inFIGS. 3,4, and6, on-time phase points t₂and t₃are both outer-ring phase points60 having a normalized outer-ring magnitude68 of 1.00. Therefore, off-time phase point t_2.5may have a normalizedmaximum excursion magnitude105 of 1.8. This implies thattransmitter circuit22, to faithfully transmit phase point t_2.5without excessive distortion, and without the benefit of the present invention, would require an output power of 3.24 (1.8²) times the power required to transmit phase point t₂or t₃, which are representative to the highest magnitude intelligence-carrying phase points54. This represents an inefficient use of available power.

The following discussion refers toFIGS. 2,4, and5.

Off-time signal stream86, a portion offiltered signal stream74, passes from an output of pulse-spreadingfilter76 to an input of an off-time constrained-envelope generator106. It is the task of off-time constrained-envelope generator106 to produce an off-time constrained-bandwidtherror signal stream108 from off-time signal stream86. A complex summing or combiningcircuit110 combines off-time constrained-bandwidtherror signal stream108 with a delayed version of filtered signal stream74 (discussed below) to produce a constrained-envelope signal stream112. Constrained-envelope signal stream112 is effectively filteredsignal stream74 with compensation for excursions oflocus72 with magnitudes greater than outer-ring magnitude68.

Aquadrature threshold generator118 generates aquadrature threshold signal120. In the preferred embodiment,threshold signal120 is a steady-state, constant signal having a value approximately equal to outer-ring magnitude68.Threshold signal120 is used to establish a reference with which off-time signal stream86 is compared. Those skilled in the art will appreciate thatthreshold signal120 may assume many forms and values in keeping with the methodology and circuitry incorporated in the comparison. The use of other forms and/or other values does not depart from the spirit of the present invention nor from the scope of the appended claims.

Threshold signal

120 and off-time signal stream86 are combined in an off-time complex summing or combiningcircuit122 to produce an off-timedifference signal stream124. Off-timedifference signal stream124 is made up of a series of off-time difference pulses126 whose values are the difference between the values of equivalent off-time pulses82 and the value ofthreshold signal120. Since any given off-time pulse82 may have a value greater than, equal to, or less than the value ofthreshold signal120, off-timedifference signal stream124 would normally be made up of a combination of off-time difference pulses126 having positive, zero, and negative values.

Off-timedifference signal stream124 is passed to the input of an off-time discriminator128 to produce an off-timeerror signal stream130. In the preferred embodiment, off-timeerror signal stream130 is a variation of off-timedifference signal stream124 in which all off-time difference pulses126 having positive values are passed unchanged as off-time error pulses132 while all other off-time difference pulses126 are passed as zero-value pulses (i.e., eliminated). In other words, off-timeerror signal stream130 is formed from pulses, the timing of which coincide with excursions oflocus72 beyond the outer-ring magnitude68 and the magnitudes of which correspond to the degree to whichlocus72 passes beyond outer-ring magnitude68.

Off-timeerror signal stream130 is then passed to the input of an off-time pulse-spreadingfilter134. Off-time pulse-spreadingfilter134 is substantially identical to first pulse-spreadingfilter76. That is, in the preferred embodiment, both

pulse spreading filters

76 and134 are realized as Nyquist-type filters with substantially identical transfer characteristics. Off-time pulse-spreadingfilter134 produces off-time constrained-bandwidtherror signal stream108 and completes the action of off-time constrained-envelope generator106.

Within off-time constrained-envelope generator106, off-time pulse-spreadingfilter134 receives one off-time error pulse132 from off-time discriminator128 perunit baud interval64. Off-time pulse-spreadingfilter134 then transforms each off-time error pulse132 into a Nyquist-type error burst (not shown) extending over a plurality of unit baud intervals. Since off-time pulse-spreadingfilter134 is a Nyquist-type filter, each error burst attains an error-burst peak value (not shown) at the primary sampling time of the specific off-time error pulse132 (i.e., at time t_2.5for error pulse t_2.5), and attains a zero error-burst value (not shown) at integralunit baud intervals64 preceding and following the peak error-burst value (i.e., at times . . . , t_−1.5,t_0.5, t_1.5, and t_3.5, t_4.5, t_5.5, . . . , for error pulse t_2.5). In this manner, the energy of each off-time constrained-envelope error pulse136 is spread over a plurality ofbaud intervals64 preceding and following the clocking instant (time t_2.5). This results in the conversion of off-timeerror signal stream130 into off-time constrained-bandwidtherror signal stream108. Off-time constrained-bandwidtherror signal stream108 is made up of off-time constrained-envelope error pulses136. This operation is essentially the same as the operation of pulse-spreadingfilter76 in the conversion of phase-point signal stream50 into filteredsignal stream74 described hereinabove.

Since off-time constrained-envelope error pulses136 are derived from off-time pulses82, the error-burst peak and zero values occur approximately midway between integral baud times, i.e., at baud times t_0.5, t_1.5, t_2.5, etc., hence between datum-burst peak and zero

values

102 and104 of filteredsignal stream74.

The production of off-time constrained-bandwidtherror signal stream108 completes the operation of off-time constrainedenvelope generator106.

Filteredsignal stream74 is also passed to the input of adelay element138.Delay element138 produces delayedsignal stream140, which is effectively filteredsignal stream74 delayed sufficiently to compensate for the propagation and other delays encountered in off-time constrained-envelope generator106, and particularly in off-time pulse-spreadingfilter134. In other words, delayedsignal stream140 is filteredsignal stream74 brought into synchronization with off-time constrained-bandwidtherror signal stream108.

Combiningcircuit110 combines filteredsignal stream74, in the form ofdelayed signal stream140, and off-time constrained-bandwidtherror signal stream108 to reduce peak magnitude components offiltered signal stream74. A resultant constrained-envelope signal stream112 is made up of a series ofdigital pulses142 whose values are the difference between the values of corresponding filtered-signal pulses78 and off-time constrained-envelope error pulses136. The result is a series ofdigital pulses142 whose values do not appreciably exceed outer-ring magnitude68 of expandedconstellation46′.

In some embodiments of the present invention, certain of outer-ring phase points60 may have magnitudes greater than outer-ring magnitude68, i.e., may be located beyondouter ring56. This condition may occur as a result of pulse-spreadingfilter76 executing certain Nyquist-type functions well known to those skilled in the art. In such an embodiment,transmitter circuit22 contains an on-time constrainedenvelope generator106′ in addition to off-time constrained-envelope generator106 discussed above.

On-time signal stream84, also a portion offiltered signal stream74, passes from an output of pulse-spreadingfilter76 to an input of on-time constrained-envelope generator106′. It is the task of on-time constrained-envelope generator106′ to produce an on-time constrained-bandwidtherror signal stream108′ from on-time signal stream84. Combiningcircuit110 combines both off-time and on-time constrained-bandwidth error signal streams108 and108′ with the delayed version of filtered signal stream74 (discussed below) to produce constrained-envelope signal stream112.

On-time constrained-envelope generator106′ operates in a manner analogous with the operation of off-time constrained-envelope generator106.Threshold signal120 and on-time signal stream84 are combined in an on-time complex summing or combiningcircuit122′ to produce an on-timedifference signal stream124′. On-timedifference signal stream124′ is passed to the input of an on-time discriminator128′ to produce an on-timeerror signal stream130′. On-timeerror signal stream130′ is then passed to the input of an on-time pulse-spreadingfilter134′, which produces on-time constrained bandwidtherror signal stream108′. Like off-time pulse-spreadingfilter134, on-time pulse-spreadingfilter134′, is substantially identical to first pulse-spreadingfilter76.

Since on-time constrained-envelope error pulses (not shown) are derived from on-time pulses80, the error-burst peak and zero values occur at integral baud times, i.e., at baud times t₁, t₂, t₃, etc., hence between datum-burst peak and zero

values

102 and104 of filteredsignal stream74.

Combiningcircuit110 combines filteredsignal stream74, in the form ofdelayed signal stream140, with both off-time and on-time constrained-bandwidth

error signal stream

108 and108′ to reduce peak magnitude components offiltered signal stream74.

A side effect of this methodology is thatlocus72 at integralunit baud intervals64 adds a signal-dependent, baud-limited noise factor to the positions of phase points54 in constellation46 (FIG.3). This results intransmitter circuit22 transmitting a “noise-influenced” phase-point constellation46″. InFIG. 7, noise-influencedconstellation46″ is depicted illustrating constrained-envelope phase-point probabilities144 of phase points54 in accordance with a preferred embodiment of the present invention. The following discussion refers toFIGS. 2,3,5 and7.

Phase-point probabilities144 reside in noise-influencedconstellation46″ exactly as phase points54 reside inconstellation46, i.e., in the same configuration with centers at the same locations. The actual location of a given transmittedphase point145 within a given phase-point probability144 is a function of a plurality of variable conditions and, although somewhat correlated, except in certain specialized cases, cannot readily be predicted. In effect, for a givenphase point54, the resultant transmittedphase point145 may be located anywhere within phase-point probability144, i.e., within an indeterminate area having a center coincident with the location of theoriginal phase point54. The probability of transmittedphase point145 being located at any specific position within that indeterminate area varies as an inverse function of the distance of that specific position from the location of theoriginal phase point54.

For any givenphase point54, the transmittedphase point145 may be said to be proximate its idealized position within noise-influencedconstellation46″. That is, a locus (not shown) of constrained-envelope signal stream112 passes proximate the idealized positions of exemplary phase points t₀, t₁, t₂, etc., at the clocking instants in time.

The original phase points54 ofconstellation46, as produced byphase mapper44, are on-time phase points90 (circles) of expandedconstellation46′. It is these on-time phase points90 that carry the intelligence ofRF broadcast signal26 as ultimately transmitted. Off-time phase points92 (squares) are by-products of pulse-spreadingfilter76, required to constrain spectral regrowth, and carry no intelligence. Phase-point probabilities144 of noise-influencedconstellation46″ represent the resultant areas of probable locations of transmittedphase points145 as derived from on-time phase points90. The centers of phase-point probabilities144 occupy the same normalized locations within noise-influencedconstellation46″ as do on-time phase points90 within expandedconstellation46′.

The positional aberrations of transmittedphase points145 relative to the corresponding on-time phase points90 represent a degree of positional error. This positional error degrades the bit error rate and effects a detriment to transmission. The absence of off-time phase points92 with a magnitude significantly greater than outer-ring magnitude68 (FIG. 4) in constrained-envelope signal stream112, however, allows an increase in power output for a given bandwidth and power amplifier that more than compensates for the position error of transmitted phase points145. A net improvement in performance results.

Referring back toFIG. 2, the output of combiningcircuit110, constrained-envelope signal stream112, is passed to an input of a substantiallylinear amplifier146. Substantiallylinear amplifier146 producesRF broadcast signal26, which is then broadcast viatransmitter antenna24. In the preferred embodiment, substantiallylinear amplifier146 is made up of adigital linearizer148, a digital-to-analog converter150, and a radio-frequency (RF) amplifyingcircuit152. Those skilled in the art will appreciate that substantiallylinear amplifier146 may be realized in any of a plurality of different embodiments other than that described here, and that utilization of any of these different embodiment does not depart from the intent of the present invention nor the scope of the appended claims.

Within substantiallylinear amplifier146,digital linearizer148 alters constrained-envelope signal stream144 into a pre-distorteddigital signal stream154. Pre-distorteddigital signal stream154 is made non-linear in just the right manner to compensate for non-linearities within digital-to-analog converter150 andRF amplifying circuit152, hence linearizing substantiallylinear amplifier146.

Digital-to-analog converter150 then converts pre-distorteddigital signal stream154 into ananalog baseband signal156. Analog baseband signal156 is then amplified byRF amplifying circuit152 intoRF broadcast signal26 and transmitted viatransmitter antenna24.

In summary, the present invention teaches a methodology and circuitry by which a transmitter circuit utilizing Nyquist-type filtration may produce a constrained envelope having a magnitude at or near the approximate unconstrained envelope magnitude of the desired constellation. This enables the transmitter output amplifier to be biased so that the maximum unconstrained envelope magnitude is at or near the top of the amplifier's linear region without incurring clipping of the constrained envelope transmissions. This in turn produces a more efficient output amplifier and effects an increase in the power output of a given output amplifier. Conversely, a lower power amplifier may be used to provide the same output power that was previously output. This effects a significant savings in output amplifier cost.

Although the preferred embodiments of the invention have been illustrated and described in detail, it will be readily apparent to those skilled in the art that various modifications may be made therein without departing from the spirit of the invention or from the scope of the appended claims.