MXPA97010065A - Radio with efficient modulation of peak power and ba amplacement - Google Patents

Abstract

The present invention relates to a method for modulating a digital information, comprising: generating the digital information, projecting the digital information in a constellation diagram to produce data symbols each having a beginning, processing the data symbols to a speed that separates them at their respective start through a range of symbols, represent the data symbols in signal components I and Q, and graduate the I and Q signal components to reduce the peak power while maintaining the power promed

Description

RADIO WITH EFFICIENT MODULATION OF PEAK POWER AND BAND AMPLITUDE Technical Field of the Invention This invention relates in general to communication devices and more particularly to communication devices with efficient modulation. BACKGROUND OF THE INVENTION In a digital dimensional communication system the transmitted waveform is formed by the addition of time-shifted versions of a basic pulse form. The amplitude of this pulse is adjusted according to the data that is sent (for example, manipulated by binary phase shift). In multi-dimensional digital communication systems (for example, Modulated by Quadrature Amplitude), multiple pulse currents are generated according to the data. To minimize the bandwidth of the transmitted waveform and thereby ensure that the transmitted waveform does not interfere with other systems operating in a near channel (frequency), the pulse shape used must have a time span that covers several symbol intervals. That is, the pulse associated with a data symbol will be superimposed on the pulses associated with the adjacent data symbols. Certain sequences of data will cause these superposition pulses to be added constructively, producing large peaks in the transmitted waveform, while other data sequences will cause these superposition pulses to cancel each other producing small values of the waveform transmitted. Amplifiers that are used to raise the power of the transmitted signal just before transmission work best when the signal remains at a fairly constant level. Large spikes in the transmitted signal lead to inefficient use of the power amplifier which in turn wastes battery life. The battery-operated communication devices employ a variety of techniques to save battery power in order to prolong the life of the battery. Increasing the efficiency of power amplifiers is a technique that designers use to extend the life of a communication device. Another scheme by which battery energy can be saved is the use of another efficient modulation technique in power. Various modulation techniques have different proportions of associated peak-to-average power. In general, it is highly desirable to have a peak-to-average ratio as close to zero dB as possible. However, many existing modulation formats result in relatively high peak-to-average power ratios. Two commonly used modulation formats are Phase Displacement Manipulation (PSK) and Quadrature Amplitude Modulation (QAM). The first uses a signal constellation where all the data symbols have the same magnitude while the latter varies both the phase and the magnitude of the individual data symbols. Binary signaling is a special case of PSK (that is, BPSK). In both modulation formats, the peak-to-average ratio depends on the shape of the pulse used. Quadrature Amplitude Modulation (QAM) uses both the phase and the amplitude of a carrier to transmit information and therefore has the potential to generate a greater peak-to-average power ratio. However, experiments have shown that, for example, a PSK constellation of sixteen symbols enjoys an improvement of 3-4 dB in the peak to average power ratio over a 16 QAM signal. However, this gain in efficiency improvement is accompanied by a 4 dB loss in sensitivity. Due to this loss of sensitivity, many system designers prefer to use the QAM modulation format despite their peak to average power ratio degraded. Referring to figure 1, a communication device is displayed as if it is currently available. Figure 2 shows a phase and magnitude path of a complex baseband 8 PSK signal. In other words, this figure represents the transition from one symbol to the next as the generated data changes state. A filter that is used to t sideband noise produces an undesirable overshoot as shown by reference 202. This overshoot 202 contributes to an increase in peak power, which results in an increase in the peak power ratio to average. This increase in the peak to average power ratio forces a designer to design an amplifier that can tolerate maximum peak power, which in turn makes the power amplifier more expensive to produce. In addition, the increase in the peak-to-average ratio reduces the power efficiency of the power amplifier. In the design of portable communication devices, the purpose of a designer is to use efficient components at the lowest possible cost. Power amplifiers have traditionally been some of the most expensive components of a communication device and have often withstood attempts aimed at decreasing their cost. A parameter that is directly related to the cost of the amplifiers is the peak to average power ratio. This is because the designer is forced to use an amplifier that can handle peak powers significantly greater than the average power. Therefore, the purpose of the designers has been to reduce peak to average power ratios as much as possible without degrading other performance parameters. Accordingly, there is a need for a modulation scheme having a minimum peak-to-average power ratio without undergoing another performance degradation. DESCRIPTION OF THE DRAWINGS Figure 1 shows a block diagram of relevant elements of a communication device as they are currently available. Figure 2 shows the magnitude and phase path of a complex baseband signal of the communication device of Figure 1. Figure 3 shows relevant portions of a communication device according to the present invention. Figure 4 shows the elements of a peak suppression algorithm according to the present invention. Figure 5 shows the magnitude and phase trajectory of a complex baseband signal according to the present invention. Figure 6 shows the optical performance diagram of the communication device according to the present invention. DETAILED DESCRIPTION OF THE PREFERRED MODE Referring to Figure 3, the relevant components of a communication device 300 according to the present invention are shown. A microphone 302 produces an analog signal which is coupled to a vocoder 304 where it is converted into a digital signal. The vocoder 304 generates a digital information signal and applies it to a Digital Signal Processor (DSP) 306. The combination of the vocoder 304 and the DSP 306 forms a digital modulator 301. The DSP 306 manipulates this digital information signal in accordance with the principles of the present invention. In addition to making peak and instantaneous power measurements, keep track of the time such peaks occur and combine the I and Q components; whose methods are known in the art, the DSP 306 carries out the graduation of the signal. In association with FIG. 4, further details of the operation of the DSP 306 will be discussed. The signal processed at the output of the DSP 306 is coupled to a digital-to-analog converter 308 where the signal is again converted to analog before being applied to a mixer. RF 310. This mixer 310, which could be a quadrature mixer, mixes the analog signal with a locally generated oscillator signal (LO). The output of the mixer is coupled to an amplifier 312, which amplifies the mixed signal before being transmitted through an antenna 314. Referring to FIG. 4, the essential elements of the DSP 306 according to the present invention are shown. In essence, a random binary data generator 401 coupled to a peak suppression algorithm 402 is shown. The generator 401 can be any digital data source such as the vocoder 304. The peak suppression algorithm includes a sputtering section of symbols 404 and a symbol ranking section 406. The digital information generated at 401 is represented on a constellation diagram 404 to produce data symbols each having a symbol interval and a start. These data symbols are represented by vectors 405, each having I and Q signal components. In other words, the data symbols are represented by vector components with orthogonal relationships. It is noted that the peak suppression algorithm can also operate on one-dimensional signals (for example, BPSK). Signal components I and Q collectively represent the magnitude and phase of vector 405. Each vector represents a symbol interval whose content is determined by the number of bits processed at each instant of time. Indeed, the data symbols are processed at a speed that separates them at their respective start through a symbol interval. For example, in a three-bit system, vector 405 represents three bits with eight different possibilities. In a four-bit system, a vector represents four bits and the signal constellation has sixteen symbol locations in it. In the preferred embodiment and in order to facilitate understanding of the principles of the present invention, a three-bit symbol interval is assumed. Once the symbols have been represented, a symbol graduation process is carried out. As part of this process, the magnitudes of the I and Q components are altered according to an algorithm that would minimize the overshoot in the subsequent filtering stage. This step is carried out through a pulse shape filter 408. The purpose of this filter is to reduce the high frequency components of the symbols before they are transmitted. However, due to its characteristics, this filter tends to produce signal peaks during transitions from one symbol to another. These signal peaks are shifted towards the additional peak power demanded from the amplifier 312. The magnitude of these peaks depends on both the sequence of the symbols and the characteristics of the filter. The present invention seeks to adjust or graduate these vectors (ie, 405) in such a way as to compensate or reduce the signal peaks. This compensation relieves the amplifier of having to operate at unnecessary peaks while maintaining the integrity of the system. The graduation of the data symbols can be implemented only in the magnitude or both in the magnitude and in the phase. In other words, the amplitude of the I and Q components can be altered in such a way as to keep the phase of the vector 405 constant. Alternatively, the amplitude of the I and Q components can be altered independently, thereby resulting in changes both in the magnitude and in the phase of vector 405. The magnitude of the non-graduated symbols of Figure 5 is shown through dotted circle 502. This circle represents the magnitude of the symbols as they are generated by the generator of random binary data 401 and the symbol projector 404. Ideally, amplifier 312 will have to amplify these signals of constant magnitude. But due to the pulse shape filter 408 these signal magnitudes are increased to the point where the concentric circle 504 is formed at the output of the pulse shape filter. This outer circle 504 shows the level of the overload placed on the amplifier 312. Indeed, the diametric distance between the two concentric circles 502 and 504 represents the difference in magnitude between the filtered and unfiltered symbols. This difference translates directly into undesirable peak power. The graduation of the symbols is equal to a contraction of the diameter of this circle despite a lower peak demand in the amplifier 312. The graduation algorithm looks in the sequence of symbols and determines the necessary alteration in each of the symbols as which are generated by the data generator 401. The algorithm uses the characteristics of the filter during this determination. Figure 5 shows a phase and magnitude trajectory of several symbols after they have graduated. The ungraded symbols are represented by 506 while their graded counterparts are shown by 508. In this example, we assume that five symbols are transmitted. The first symbol 501 is unchanged because no peak is generated. The following symbol is graduated radially downwards to avoid the signal peak that would normally result due to the interaction between the filtered symbols delayed in time. The third symbol graduates similarly downward to avoid a peak signal magnitude. The fourth graduates similarly downward. The fifth symbol is graduated upwards due to the small magnitude of the signal that occurs during the transition from the fourth symbol to the same. The symbol graduation is carried out in a way that maintains the integrity of the symbol and prevents the loss of information. The peak suppression algorithm determines the instantaneous power of the baseband signal during each symbol interval. The graduation of the signal will immediately follow the determination of the peak power and its time location in the symbol interval. Under these circumstances, the average power associated with the baseband signal is also determined. With the available peak power information the algorithm determines the time at which the composite baseband signal occurs. Next, the I and Q components of the symbol associated with the adjacent symbol intervals are altered. The amplitude of these components can be graduated in a radially equal manner at which time only the magnitude of the composite signal varies. Independent and unequal graduation of the I and Q signal is also possible, which would result in the graduation of the phase and the magnitude of the composite signal. In summary, the digital data symbols generated by the vocoder 304, and the symbol renderer 404 are processed through the peak suppression algorithm 402 in order to take advantage of the principles of the present invention. The data symbols generated as a result of this representation are represented by its components I (in phase) and Q (in quadrature). The I and Q components are dynamically graded through the symbol scaling portion 406 of the peak suppression algorithm block 402. The graduation of the I and Q components is before the filtering action that takes place. through the pulse shape filter 408. The symbol ranking simply maintains the record of the magnitude and phase path of the baseband signal (constituted by the I and Q components). As discussed, the problem with the prior art is that the pulse shape filter produces signal peaks during symbol transitions. The present invention provides a method for minimizing this peak signal problem. By graduating the I and Q components of the data symbols, the present invention is directed to decreasing the magnitude of the signal peaks, thereby reducing the peak power demand in the amplifier 312.

The algorithm used in the preferred embodiment accepts data symbols that have been produced by the constellation 404, processes the symbols and outputs them to the impulse shape filter 408. Specifically, the algorithm sequentially loads the data symbols into a block of data. input data for its iterative processing. Upon completion of the processing, the input data block is copied into an output data block and the graduated symbols are sequentially transmitted to the pulse shape filter 408. To maintain a constant symbol rate, the newly arrived data symbols they move to the unoccupied input data block while the graduated symbols move out of the output block. Therefore, if the processing time is assumed to be negligible, the transmission delay created by the algorithm is approximately equal to (block size) / (symbol rate) seconds. The block size must be large enough to ensure that the symbols within the block represent exactly the statistical characteristics of the transmitted, total data symbol sequence. After the successful population of the input data symbol block, the algorithm proceeds to determine several values for each symbol range defined by the input symbol block. These values are: (1) the peak transmitted signal magnitude, (2) the peak time location, and (3) the peak scale factor for the peak magnitude. The algorithm determines these values in a range of symbols by applying the pulse shape filter function to the appropriate data symbols. The number of data symbols used to calculate the signal over a particular symbol range depends on the impulse response of the pulse shape filter function. All symbols that combine with the pulse shape to produce a significant signal magnitude within the range of symbols of interest must be included in these calculations. The impulse response of the pulse shape filter 408 also determines how much symbol overlap must exist between the successive symbol blocks. The algorithm uses the peak amount of signal transmitted in a particular symbol range to determine the peak scaling factor for that range. A peak scaling function is applied to the peak signal value. The peak scaling function is defined in such a way as to produce a negative peak scaling factor if the peak magnitude is greater than some reference value and a positive scaling factor if it is less than the reference value. The magnitude of this scale factor increases with the difference between the peak magnitude and the reference value. The reference value is normally set equal to the desired peak magnitude. The algorithm stores the peak scale factor and the corresponding peak time location for each symbol interval in two separate vectors. These values will be used subsequently to determine the symbol scale factor for the symbols in the block. After the successful completion of the peak scale factors and their associated time locations, the algorithm calculates the symbol scale factor for each of the data symbols. To determine a particular symbol scale factor, the algorithm uses the peak information from the two symbol intervals that are immediately adjacent to a particular symbol. These two intervals will be referred to as the intervals on the left and on the right. The symbol scale function weights the peak scale factor to the left by the relative time distance at which the peak of the particular symbol is located. Similarly, the peak scaling factor on the right is weighted by the relative distance that exists from the particular symbol. The two weighted scale factors are then summed together with a unit value to determine the symbol scale factor. In this way, the signal peaks that are located close to a particular symbol have a greater impact on the scale factor for that symbol. After each of the symbol scale factors has been determined, the algorithm normalizes the symbol scale factors to maintain the desired average power. Assuming that the impulse form has average unit energy and that the individual symbols are independent and are distributed identically, the average power (Ps) is calculated by the simple apportionment of the squared graded symbol quantities. The average power desired is normally equal to the average power of the transmitted signal without graduating (Pu). Therefore, the normalization factor is equal to Sqrt (Pu / Ps). In the case of a circular PSK constellation of unit symbol magnitude, Ps is simply equal to the average of the symbol scale factors. The algorithm repeats the symbol processing steps described above during a specific number of iterations or until a peak to a target averaging ratio is obtained. After one of these conditions has been met, the algorithm graduates the data symbols by the appropriate final symbol scaling factors and copies the graduated symbols to the output block. The algorithm then proceeds to sequentially emit the graduated symbols towards the pulse shape filter while simultaneously loading the input block with the new ungraded symbols from the constellation plotter. In an alternative mode, the peak suppression algorithm produces an imaginary sphere around each data symbol in order to create a limit to its graduation. This spherical boundary helps in the establishment of limits for movement and phase and magnitude gradation. Again this rating helps to minimize the peak power requirement in the amplifier 312. Simply stated, the ranking algorithm searches the phase and magnitude of symbols as they are generated by the vocoder 304 and the symbol renderer 404 and estimates the magnitude of the signal peaks (degree of overshoot) that will exist at the output of filter 408. This estimate of signal peaks is considered when determining the level and direction of the graduation to be implemented in each symbol. In doing so, the I and Q components are presented to the filter 408 with sufficient compensation to minimize the effect of the unavoidable signal peaks. This compensation minimizes the peak power requirement in amplifier 312. It is appreciated that without the benefit of the present invention, amplifier 312 must be able to handle the peak power demands as represented by circle 504. This additional requirement greatly increases the cost of amplifier 312. The increase in peak power vis-a-vis with the average power adversely affects the efficiency of amplifier 312. Portable radio devices are particularly disadvantaged in view of this further degradation in efficiency. The principles of the present invention provide a general method for suppressing the peaks in the transmitted waveform before being amplified. The magnitude of a data symbol is adjusted slightly according to the values of the surrounding symbols and the response of the pulse shape filter. The result is a transmitted waveform that retains a much more constant magnitude level. The algorithm works in a data block (normally it works better at approximately 50 to 500 symbols at a time). The peak suppression algorithm can be briefly described as follows: STAGE I: based on the data symbols for the block, and the pulse shape to be used, construct the transmitted waveform.

STAGE 2: for each symbol interval in the transmitted waveform, calculate the peak value of the waveform in that interval, the position of that peak and the peak scale factor. STAGE 3: based on the peak scale factors and their positions, re-grade the heights of each data symbol. STAGE 4: repeat steps 1-3 using the graduated data symbols. Continue repeating this procedure until neither a peak suppression (or very few) can be achieved. The use of this peak suppression algorithm can in some cases double the efficiency of the power amplifier or equivalently double the life of the battery in a portable radio. Referring once again to Figure 5, the points 506 in the inner circle represent the magnitude of the unfiltered symbols. To prevent these symbols from suffering in some way from over-impulse towards the limits shown by the outer circle 504, they are graded as shown by 508. As can be seen, some of the symbols are graduated downwards while others are graduated upwards with object to minimize the magnitude of peak signal and error. The graduated symbols reduce the peak power demand and therefore improve the efficiency of the amplifier. In addition, the peak power requirement of the power amplifier is reduced. This reduction directly translates into a lower cost for the 312 amplifier. The improvement in system performance is carried out with minimal impact on the modulation accuracy. Figure 6 shows an optical diagram of a demodulated signal having its peaks suppressed. The optical aperture 602 shows to have an opening wide enough to maintain the performance of errors. This is highly significant since a modulation technique is only desirable when the demodulation techniques available for it are highly accurate. In addition to the modulations using phase or amplitude, the principles of the present invention are applicable to a QAM system that uses both the phase and the amplitude of a signal to carry information.

Claims (17)

1. NOVELTY OF THE INVENTION Having described the present invention, it is considered as a novelty and therefore the property described in the following claims is claimed as property. 1. A method for modulating digital information, comprising: generating digital information; projecting the digital information into a constellation diagram to produce data symbols each having a start; processing the data symbols at a speed that separates them at their respective start through a range of symbols; represent the data symbols in signal components I and Q; and graduating the I and Q signal components to reduce the peak power while maintaining the average power. The method according to claim 1, characterized in that the processing step includes the step of filtering the signal components I and Q to maximize the spectral efficiency and to produce baseband I and Q signals. The method according to claim 2, characterized in that it also includes the step of determining the average power of the baseband signal I and Q. The method according to claim 2, characterized in that it further includes the step of determining the instantaneous peak power of the baseband I and Q signal over each symbol range. The method according to claim 4, characterized in that it further includes the step of determining the moment at which the instantaneous peak power of the baseband I and Q signal occurs in each symbol interval. The method according to claim 2, characterized in that it further includes the step of combining the baseband I and Q signals to produce a composite baseband signal. The method according to claim 6, characterized in that it further includes the step of determining the average power of the composite baseband signal. The method according to claim 6, characterized in that it further includes the step of determining the peak power of the composite baseband signal in each symbol range. 9. A digital modulator comprising: a digital information generator; means for representing digital information in a constellation diagram to produce data symbols each having a range of symbols; means for representing the data symbols in signal components I and Q; and means for dynamically altering the amplitude of the I and Q signal components in order to reduce the peak to average power ratio. 10. The digital modulator according to the claim 9, characterized in that it also includes a filter to produce baseband I and Q signals with maximum spectral efficiency. 11. The digital modulator according to the claim 10, characterized in that it also includes means for determining the average power of the baseband I and Q signals. 12. The digital modulator according to the claim 10, characterized in that it further includes means for determining the instantaneous peak power of the baseband I and Q signal over each symbol range. The digital modulator according to claim 12, characterized in that it also includes a synchronizer for determining the moment at which the peak power of the baseband I and Q signals occurs in each symbol interval. The digital modulator according to claim 10, characterized in that it further includes a combiner for combining the baseband I and Q signals to produce a composite baseband signal. 15. A digital modulator comprising: a digital information generator; means for converting the digital information into a multidimensional signal constellation to produce data symbols each having a magnitude and a phase; and means for forming a sphere around each data symbol within which the magnitude and phase of the data symbol could be altered in order to minimize the peak to average power ratio. 16. A digital modulator comprising: a digital information generator; means for converting the digital information into a signal constellation having at least one dimension in order to produce data symbols each having a magnitude; and means for dynamically grading the magnitude of the data symbols in order to avoid unwanted peak power occurrences in symbol transitions and the minimize the peak to average power ratio. 17. A communication device comprising: a digital modulator comprising: a digital information generator; means for converting the digital information into a signal constellation having at least one dimension in order to produce data symbols each having a magnitude; and means for dynamically grading the data symbols in order to avoid unwanted peak power occurrences in symbol transitions and the minimize the peak to average power ratio; an amplifier to amplify the data symbols; and an antenna to transmit the data symbols.