JP2008085379A - Ofdm transmitter - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enable peak suppression without any recursive trials when suppressing PAPR using a correction signal for peak suppression to reduce the outband radiation of an OFDM signal. <P>SOLUTION: The correction signal for peak suppression is added to a data signal converted to a time waveform by inverse Fourier transform by an adder 104. A correction signal generated by a correction signal generation section 103 is the time waveform of a single carrier base, and the phase of a correction signal is modulated corresponding to a point requiring peak suppression while the correction signal generated by a correction signal generation section 103 has a time waveform of a single carrier base. By the procedures, the correction signal can be generated only by successively determining the phase of the correction signal corresponding to the point requiring the peak suppression. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a transmitter using OFDM (Orthogonal Frequency Multiplexing).

  With the development of the information society, a large amount of information has been handled in familiar scenes, and the demand for communication speed has increased rapidly. In order to increase the communication speed, there are many examples in which a band that has not been used in the transmission medium is used. For example, the telephone line was used only for telephones until a long time ago, and its band was 8 kHz at most, but now xDSL can pass through and a band of several MHz is used. In addition, a low-speed communication of kbps order is currently possible on a power line for transmitting power, and communication of the order of 100 Mbps will be performed in the near future. In this way, when a band other than the band assumed originally is used, the shielding performance of the medium cannot catch up, and there is a high possibility that a large amount of unnecessary waves are radiated. When the frequency of the unnecessary wave is used for some purpose outside the medium, the unnecessary wave becomes an interference wave.

  In wireless communications, the law stipulates that a specific band be limited to a specific application, but some bands are open to multiple applications, and a wide band such as UWB is low power. Some frequencies have been determined to be open to overlap with other applications, limited to near field communication. In addition, studies have been started on a system called cognitive radio that detects that a licensed frequency band is not used for a specific system and a terminal of another system uses that frequency band. In such a case, a certain communication may become an interference wave with respect to another wireless system.

  Thus, with the increase in communication capacity, there is a high possibility that various interferences will occur. The performance with respect to interference on the side receiving the interference (interfered side) is various, and there is often no problem if there is a little interference, but there is also a case where even a small amount of interference such as radio astronomy has a great influence. For such a high-sensitivity system, it is necessary to consider that no interference wave is generated at that frequency. In addition to not using that frequency for communication, it is necessary to sufficiently suppress out-of-band radiation that appears at that frequency.

  The above systems need to be considered so as not to interfere with other systems, and most of them have a wider usable bandwidth than before, so in some cases, at the expense of part of the bandwidth. However, it is necessary to suppress unnecessary radiation so as not to interfere with other systems.

  Although there are various modulation systems for communication systems, the OFDM (Orthogonal Frequency Multiplexing) system has been frequently used because of its high frequency utilization efficiency. However, OFDM has a drawback that it tends to emit out-of-band unnecessary radiation for various reasons. One of them has a property that PAPR (peak-to-average power ratio) is large because a very large number of subcarriers are multiplexed. As a result of the large PAPR, clipping distortion occurs in the digital-analog (D / A) converter, and third-order intermodulation distortion (IM3) is likely to occur in the power amplifier.

Various methods are known for suppressing PAPR (for example, Patent Documents 1 and 2). In these methods, among subcarriers, a plurality of subcarriers with a small amount of data are used as correction signals for peak suppression, and PAPR is best improved among appropriate combinations of phases and amplitudes of the plurality of subcarriers. The combination to be selected is selected.
JP 2005-101975 A JP 2002-314502 A

  However, these methods are methods in which several combinations of phases and amplitudes are tried and an appropriate one is selected, and repeated trials are required to select an appropriate combination.

  The present invention provides an OFDM transmitter having a correction signal determination algorithm capable of sufficiently suppressing a peak without such repeated trials when suppressing PAPR using a peak suppression correction signal in order to reduce out-of-band radiation. The purpose is to provide.

  In order to solve such problems, the OFDM transmitter of the present invention assigns a group of adjacent subcarriers to a peak suppression correction signal, and transmits a group of subcarriers different from the subcarrier assigned to the correction signal. Subcarrier allocating means for allocating to the subcarrier for data, mapping means for mapping data to the subcarrier for data transmission, waveform converting means for converting the data transmission subcarrier mapped with the data into a time data waveform, A sine wave time waveform corresponding to one of the correction signal subcarriers is generated, the amplitude of the correction signal is determined, and the point to be peak-suppressed with respect to the time data waveform is the absolute amplitude of each point of the time data waveform List based on value and time corresponding to each point listed The phase of the correction signal is determined so that the amplitude absolute value of each of the listed points is substantially within a predetermined range, and the phase of the correction signal at a time other than each of the listed points is determined. Correction signal generating means for outputting the correction signal whose phase is determined, and addition means for adding the correction signal output from the time data waveform and the correction signal generating means and outputting an addition signal. It is characterized by.

  According to the OFDM transmitter of the present invention, it is not necessary to repeat the procedure of checking whether or not the peak suppression has been performed by changing various combinations of phase and amplitude, and selecting the optimal combination. The correction signal can be determined by a simple procedure while reducing external radiation.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  First, the principle and operation of the present invention will be described with reference to FIG. FIG. 1 is a block diagram showing a basic configuration of the OFDM transmitter 100. In this embodiment, the OFDM transmitter 100 generates a data mapping unit 101 to which data to be transmitted is input, an inverse Fourier transform unit 102 that performs inverse Fourier transform (IFFT) on the data-mapped subcarrier, and a peak suppression correction signal. A correction signal generation unit 103 that performs output, an adder 104 that adds the output signal of the inverse Fourier transform unit 102 and the output signal of the correction signal generation unit 103, and a D / A that converts the digital signal output from the adder 104 into an analog signal The analog unit 106 includes a converter 105, a filter, a frequency converter, an amplifier, and the like, and an antenna 107 that radiates the output of the analog unit 106 to the outside.

  The operation of each part will be described. First, data to be transmitted is input to the data mapping unit 101. Before the data is input to the data mapping unit 101, the data passes through the upper layer processing part, the part to which the error correction code is added, the connection part with the device having the application, etc. It is omitted because it is not directly related to the operation of the present application. Also in the following embodiments, portions not directly related to the operation of the present application are omitted.

The data mapping unit 101 maps the input data to the subcarriers forming the OFDM symbol using a modulation scheme appropriate for the subcarriers such as QPSK and BPSK. FIG. 2 shows an example of the subcarrier arrangement of the correction signal and the data signal. As shown in FIG. 2, a part of the subcarrier group is reserved for the peak suppression correction signal and excluded from the data mapping target. For example, when one OFDM symbol consists of 128 subcarriers, data is mapped to the first 112 subcarriers, and the last 16 subcarriers are reserved for the peak suppression correction signal, and no data is mapped. Therefore, the last 16 subcarriers do not have a carrier component at this point.

Next, the subcarriers thus data-mapped are subjected to inverse Fourier transform by an inverse Fourier transform (IFFT) unit 102 to obtain a time-domain signal waveform having a length corresponding to the symbol length. FIG. 3 shows an OFDM time waveform and a correction signal waveform. The dotted lines in FIG. 3A are examples of waveforms when arbitrary data is mapped to 112 subcarriers out of 128 subcarriers by QPSK and converted into time domain signal waveforms. The inverse Fourier transform (IFFT) unit 102 outputs this waveform to the correction signal generation unit 103 and the adder 104.

  The correction signal generation unit 103 compares the input time waveform with the clipping level of the subsequent digital-analog (D / A) converter 105. In FIG. 3A, the solid line drawn horizontally above and below the graph is the clipping level of the D / A converter 105, and if a point having an amplitude exceeding this is input to the D / A converter 105 as it is, clipping is performed. And output as clipping level amplitude.

  The correction signal generation unit 103 does not perform the correction signal generation process if there is no point exceeding the clipping level in the input time waveform. Usually, however, some points exceed the clipping level. In this case, the following processing is performed.

  This correction signal generation processing will be described with reference to the flowchart of FIG.

  First, the correction signal generation unit 103 extracts a point where the absolute value of the amplitude is the maximum among points that exceed the clipping level, that is, a point that greatly exceeds the clipping level. In FIG. 3A, this is the point indicated by a1. The absolute value of the difference between a1 and the clipping level, that is, the amplitude amount at which a1 exceeds the clipping level is defined as the amplitude α of the correction signal. Here, the absolute value of the clipping level is CL (step S401).

  Next, the correction signal generation unit 103 lists points to be subjected to peak suppression processing. Levels whose amplitude absolute value is small by the correction signal amplitude α previously determined from the upper and lower clipping levels ± CL, that is, + CL−α, −CL + α (the levels shown as the list-up level in FIG. 3A) The points that exceed are listed in descending order of absolute amplitude. In FIG. 3A, they are indicated by a2, a3,. In FIG. 3 (a), up to a3 is shown, and illustration of the points below it is omitted (step S402).

  Next, for example, the correction signal generation unit 103 generates a sine wave signal corresponding to a subcarrier near the center of the region indicated as the peak suppression correction signal subcarrier 21 in FIG. 2 described above, for example, the 121st subcarrier. To do. The amplitude is α. This is, for example, a waveform as shown in FIG. The correction signal is a sine wave that is not modulated at all and is a repetitive waveform having a constant amplitude and a constant period. However, in most cases, there is a difference between the sampling points of the period and the time waveform, so that the waveform appears as if the amplitude increases and decreases in a slow period as shown in FIG.

  First, the phase offset of the correction signal is determined so that the sum of the data signal waveform and the correction signal is less than or equal to the clipping level at the point a1. Since the amplitude α of the correction signal is determined as the amount at which a1 protrudes from the clipping level, and a1 exceeds the upper clipping level, the phase offset of the correction signal at this point is the point a1. The phase is such that the correction signal is negative and the absolute value is the maximum (-π if the correction signal is expressed in cosine). Next, the phase of the correction signal at the point a1 is fixed to this value. Further, for a1 only, the phase is shifted by the amount changed from the initial value in a1 over the entire correction signal. Therefore, this phase change becomes a phase offset of the correction signal (step S403).

  Next, attention is focused on the next point a2 listed (step S404). The correction signal at that time is added to a2, and it is verified whether the added amplitude does not exceed the clipping level (step S405). If the clipping level is not exceeded, the phase of the correction signal at a2 is fixed to the current value. When the clipping level is exceeded, the phase of the correction signal is changed to the phase closest to the phase of the current correction signal among the phases in which a2 becomes the clipping level. The phase of the correction signal at a2 is fixed to this value (step S406).

  Next, from a2, the points on both sides of which are already fixed in phase are detected. That is, in the case of a2, one is a1. There is no point where the phase is fixed on the other side, but when it reaches the end of the time waveform, it turns to the opposite end and detects a1 from the opposite side. The same applies to the third and subsequent points, and if a point whose phase is fixed is not found by the end of the time waveform, it goes to the opposite end and searches for the next point whose phase is fixed. Then, the phase of the correction signal between a2 and the point where the adjacent phase is fixed is gradually changed. “Slow change” indicates a state in which the phase skips in the middle or is bent at an acute angle, or does not undergo an extreme change. FIG. 6 is a diagram showing how the phase of the correction signal is interpolated. For example, as shown in FIG. 6, the change in the phase of the correction signal changes from a2 to a1 on both sides in a sine wave shape corresponding to a half cycle. So that the shape is

The correction signal generated by the 121st subcarrier is, for example, cos (2π * 121 * j / L + φ _j ), L is the number of samples forming the OFDM1 symbol, j is the jth time waveform, and φ _j is a point j In FIG. 6, only the phase addition is plotted in the example of FIG. After the phase of a1 is determined, the phase of a2 is determined, and the phase in between is interpolated in a sine wave shape for a half cycle. Similarly, not only between a1 and a2, but also the portion that wraps around from the opposite side via the symbol end is also changed to a sine wave shape.

The reason why the phases at both ends of the symbol are connected is that the Fourier transform basically assumes a periodic waveform, and if the period breaks are discontinuous (including phase), the spectrum will spread. In the above description, the phase of the correction signal at a1 is assumed to be −π. In this case, the phase means that 2π * 121 * j / L + φj is −π, and therefore φ _j is not necessarily −π. Must not. It is this phase addition φ _j that is gradually changed in the above.

  Next, if there are still points listed, the same processing as in the case of a2 is performed for the next point. In the example of FIG. 3, the process of a3 is performed. First, it is the same from the step of verifying whether the amplitude obtained by adding the correction signal at that time, that is, the value at a3 of the correction signal whose phase is changed as shown in FIG. 6 to the data signal point does not exceed the clipping level. To process.

  Note that the point to be listed is not only the point that exceeds the clipping level but also the point from the clipping level to the inside by the correction signal amplitude α, as a result of adding the correction signal, the clipping level is exceeded. This is to list all possible points.

  If the same processing is performed until the end of the listed points, phase correction is applied to the correction signal as shown in FIG. 5 as shown in FIG. 3B.

  The correction signal thus generated is added to the data time waveform by the adder 104 of FIG. At this time, a slight delay is added to the data time waveform so that the data waveform and the correction signal waveform are added at the same timing, or the data time waveform is temporarily stored in the buffer, and the generated correction signal is synchronized with the timing. Output from the buffer (added so that the symbol time does not shift) and add. The adder 104 includes such a delay amount adjustment function.

  The solid line in FIG. 3A is a waveform obtained by adding the correction signal in FIG. 3B to the time waveform before correction indicated by the dotted line, and it can be seen that all the points are within the clipping level. According to the invention, it is possible in principle to ensure that all points of the summed signal fall within the clipping level, apart from the right in light of the final purpose.

  The spectrum of the corrected waveform is, for example, as shown by a solid line in FIG. A correction signal is present in a group of spectra at the end of a subcarrier group to which data is mapped. The first correction signal was a line spectrum corresponding to a specific subcarrier, but the phase was determined so that the phase was appropriate at several points on the data time waveform, and a moderate phase change was made between them. As a result, a relatively gradual phase modulation is applied, and a slight spectral broadening occurs. The reason why a group of subcarriers is first vacated for the correction signal is to allow for the spread of the spectrum so that data is not mapped to subcarriers having a large influence.

  In FIG. 7, no data subcarrier is present near the subcarrier number 50. This is because, in order to confirm the effect of the present invention, subcarriers are extracted without intentionally loading data, and the spurious level given by the correction signal is confirmed.

  An object of the present invention is to reduce PAPR in order to suppress out-of-band radiation due to clipping. According to the present invention, since the correction signal has a certain spectrum spread, it is necessary to confirm that the out-of-band radiation due to the spectrum spread is sufficiently small as compared with the case where clipping is performed without doing anything. is there. Since the calculation of the actual out-of-band radiation is somewhat time-consuming, in FIG. 7, some of the subcarriers in the band are extracted to form a spurious measurement window, where the spurious level is detected. Since this window is a convenient window for measuring the spurious level, when the present invention is carried out, the subcarriers are naturally mapped to the window and used.

  If clipping is done without doing anything, spurs with a relatively flat spectral shape are generated. A signal spectrum including spurious is shown by a dotted line in FIG. When the correction signal of the present invention is added, the spectrum rises in the vicinity of the correction signal frequency due to the presence of the correction signal, but the spurious is sufficiently reduced compared to the case where only clipping is performed in the spurious measurement window portion. It can be seen that out-of-band radiation can be sufficiently suppressed by the present invention.

  Returning to FIG. 1, the signal output from the adder 104 is converted by the D / A converter 105 from a signal in a numerical sequence to an analog signal expressed in voltage. Next, a signal is radiated from the antenna 107 via the analog unit 106 configured by a filter, frequency conversion, an amplifier, and the like. If the OFDM transmitter is applied to a wired system, it is output to a cable for transmission instead of the antenna 107. In the configuration of the present embodiment, the digital unit 108 processes the signal as a numerical string until it is converted into an analog signal by the D / A converter 105.

  Although a cyclic prefix (CP) is usually added to an OFDM signal, the CP adding unit is not shown because it is not directly related to the principle of the present invention. In order to insert a CP adding section, it is sufficient to insert it after the adder 104 and before the D / A converter 105.

  As described above, according to the present invention, after the data-mapped subcarriers are converted into time waveforms by IFFT, a peak suppression correction signal is generated and added. The correction signal is converted into a single carrier-based time waveform, and the correction signal is phase-modulated so that the amplitude after addition of the data signal and the correction signal is reduced by a necessary amount corresponding to the point where peak suppression is required. According to such a procedure, a correction signal can be generated simply by sequentially determining the phase corresponding to the point where peak suppression is required. Therefore, unlike the conventional method, it is not necessary to perform an iterative procedure of searching for an optimal combination of correction signals while checking whether peak suppression has been performed, and the correction signal determination procedure can be greatly simplified.

  In the above procedure, the amplitude α of the correction signal is the amount that the maximum amplitude point of the data time waveform exceeds the clipping level. By doing in this way, it is possible to make the correction signal amplitude necessary to fit the point most protruding from the clipping level into the clipping level, and to keep the necessary correction signal amplitude to the minimum necessary.

As can be seen from FIG. 7, the spectrum of the correction signal is caused by the phase modulation, and the spectrum spread becomes out-of-band radiation. Therefore, for the purpose of suppressing out-of-band radiation, it is desirable that the correction signal amplitude be the minimum necessary. The correction signal amplitude can be kept to the minimum necessary by the amplitude determination procedure as in the present invention.

  In the above procedure, the points of the data time waveform are listed as the peak suppression targets in descending order of amplitude, and the phase of the correction signal is fixed in this order. For example, the listed points can be processed in chronological order from the earliest time even if they are not in the order of amplitude. However, by taking the procedure as described above, the phase of the correction signal is determined in order from the point where more peak suppression is required. The later point in the processing order is that correction may be required due to the addition of the correction signal. According to the above procedure, the phase of the correction signal changes as the correction progresses, but depending on the phase state of the correction signal at that time, the phase of the correction signal corresponding to the later point needs to be changed. Often there is no. Therefore, by processing the points that need to be corrected within the first time, the phase of the correction signal given at the beginning of the processing step at the point where there is less need to change the phase is changed to the final correction signal. It can be close to the phase. As a result of the high probability that the phase need not be changed at such a point, the time required for processing the correction signal can be reduced.

  Furthermore, in the above procedure, the phase of the correction signal between the peak suppression target points whose phases are already fixed is gradually changed. This is for reducing the phase modulation applied to the correction signal and making the spectrum spread of the correction signal compact. If there is a portion where the phase of the correction signal is staggered or bent, the phase modulation applied to the correction signal has a harmonic component, so that the spectrum spread becomes large. A gentle phase change was applied to prevent this.

  In addition, in the above procedure, it is the same reason that a gradual phase change is determined by connecting one end of the correction signal time to the other end. In many cases, a cyclic prefix is added to the symbol after the addition of the correction signal. At this time, if the phase of one end of the correction signal and the other end are not smoothly continuous, the cyclic prefix and the symbol body are added. A phase discontinuity occurs at the boundary, and the spectrum of the correction signal spreads unnecessarily.

  Therefore, when a cyclic prefix is not added as in the case of a zero padded prefix (Zero Padded Prefix), it is not necessary to connect one end of the correction signal to the other end if there is only an out-of-band radiation problem. In such a case, if there is no point where another phase is fixed from the point where the phase is fixed to the end of the symbol period, the above procedure may be performed by adding the same phase as the point where the phase is fixed to the end. , The procedure can be simplified.

However, when receiving and demodulating such a signal, if the phase of both ends of the correction signal is not continuous when the FFT frame is cut out, the phase of both ends of the correction signal spectrum after FFT will be continuous. As a result, the modulation accuracy of a subcarrier carrying data may be deteriorated. Therefore, if there is a margin in the processing capacity of the transmitter, it is desirable that the phases at both ends be continuous even in a system using a zero padded prefix.

  The gradual phase change in the above procedure means that there is no sudden change in the phase of the correction signal. In an extreme example, an abrupt change in phase is a phase jump or a phase bend, but vibration at a fast cycle can also cause the spectrum of the correction signal to widen. In order to prevent such a state, it is only necessary to prevent the second derivative regarding the time corresponding to the phase addition of the correction signal from deviating from a certain range.

  The added phase of the correction signal is the phase change itself. The one-time differentiation indicates how much the phase change changes. Therefore, if the rate of change is equal, the same numerical value is obtained even if the frequency of change is different. Therefore, there are many cases where the spectrum spread cannot be limited even if the single differentiation is within a certain range. Similarly, even if there is a bend in the phase change, if only one-time differentiation is detected, there is a possibility that it will fall within a certain range depending on the differential coefficient before and after the bend.

  On the other hand, if the second derivative corresponding to the added phase falls within a certain range, for example, if the phase change is bent, the second derivative of the bent portion becomes a δ function, that is, an impulse. In addition, when there is a phase change such as vibration at a fast cycle, since the change in the first derivative is large, the second derivative is sufficiently large. Therefore, it is possible to control the spectrum spread of the correction signal by making the second derivative of the phase change fall within a certain range.

  Differentiation is an operation applied to signals that are continuous in time. In the present invention, the correction signal generation unit 103 is in the digital unit 108 in the OFDM transmitter 100. Accordingly, the correction signal is in the form of a numerical sequence that is discrete in time. In such a case, a difference is used instead of differentiation. The second differentiation takes the form of difference. In the case of differentiation, the amount of change is usually divided by the time piece that defines the change. However, when the difference between discrete numerical sequences is taken, if the sample interval of the numerical sequence is constant, it is simply Just take the difference. Corresponding to the sample interval, the value of the range (upper limit, lower limit) in which the difference should enter may be adjusted.

Specifically, the difference is defined as shown in FIG. FIG. 8 is a plot of the added phase φ in discrete time. There is a phase addition φ _j of the j-th sample point, and similarly there are j + 1 and j + 2 φ _{j + 1} and φ _{j + 2} . The phase differences are defined as dif _j = φ _{j + 1} −φ _j and dif _{j + 1} = φ _{j + 2} −φ _{j + 1} , respectively. Difference of the difference between the phase corresponding to the second derivative is defined by _{dif2 j = dif j + 1 -dif} j. All the sample points of the correction signal need only be within a certain range determined by the upper limit and the lower limit. The difference difference can take either plus or minus sign, but in the light of the purpose of limiting the spectrum spread of the correction signal, the effect on the spectrum spread is the same whether plus or minus, so the absolute value of the difference difference And this should be less than a certain upper limit value.

  As described above, a function that gives such a gradual phase change includes a sine wave corresponding to a half cycle. The derivative of a half-cycle sine wave has zero at both ends. In the procedure of sequentially fixing the phase as described above, the phase between the adjacent points where the phase has already been fixed is interpolated with a sinusoidal function corresponding to a half cycle. At that time, the phase is interpolated in the same half-cycle sine wave shape on the other side of the adjacent point, and the differential coefficient of the phase change at the end is zero. Therefore, when the phase is interpolated to the adjacent point, the phase is not bent in the vicinity of the adjacent point. Therefore, it can be said that the half-cycle sine wave has a change shape that can be easily applied to the method of sequentially fixing the phase.

  In the above procedure, if the correction signal is added to the listed point and the result is within the clipping level, the point is skipped without changing the phase of the correction signal only by fixing the phase. It was. After that, when the correction signal phase of the listed point adjacent to the point is changed and the surrounding phase needs to be interpolated, the phase of the skipped point is only fixed, Phase interpolation is not performed. Therefore, when interpolating between sine waves for half a cycle, the phase may be bent at that point. In order to avoid such a situation, all phase interpolation steps may be performed for all listed points. Alternatively, when it is determined that the curve is bent, the phase in the vicinity of the curve is adjusted so that the differential coefficient does not become discontinuous (the difference between the added phases does not have a value exceeding the upper limit). .

  In addition, examples of the shape that gives a gradual change include a polynomial whose degree is limited. If the derivative near the point where the phase is fixed does not become zero as in the case of a sine wave, the derivative at the end is matched with the derivative on the other side of the point in the procedure for interpolating the phase, or vice versa. In addition, it is preferable to perform a step such as changing the derivative of the other side of the point again according to the derivative of the end.

  In this way, the phase is sequentially interpolated, and it is only necessary to determine the phase in accordance with the procedure, and it is not necessary to select an appropriate one from a number of combinations, so that the procedure can be simplified.

  Note that the above procedure is a basic procedure, and in order to limit the spectrum spread of the correction signal and sufficiently reduce spurious due to the correction signal, some exception processing is performed to achieve an effect more in line with the object of the present invention. Is easy to obtain.

  In the above procedure, the amplitude of the correction signal is set to an amplitude that reduces the largest protruding point from the clipping level to the clipping level. However, in the OFDM signal, the phase of each subcarrier is easily aligned at the end of the symbol period, so that it is likely that the amplitude jumps out significantly at the end of the symbol and then jumps out moderately. If the amplitude of the correction signal is increased in order to return the point that protrudes significantly by one point to the clipping level in this way, the power of the correction signal increases, resulting in an increase in the spurious of the correction signal. Therefore, in the present invention, an upper limit is provided for the amplitude of the correction signal. When the amplitude necessary to retract the point that is the largest protruding from the clipping level to the clipping level exceeds this upper limit value, the upper limit value is set as the amplitude of the correction signal. Of course, if this upper limit is not exceeded, the amplitude maximum point is set to the amplitude required to return to the clipping level.

  In this way, clipping cannot be completely eliminated. However, if the point that protrudes violently is slightly retracted and the other points are retracted, and one point that is protruding is clipped, the correction signal is used to completely lower that point. Out-of-band radiation can be reduced by increasing the amplitude.

  Similarly, it is preferable to limit the phase change. In the above procedure, the points listed are all determined to have such a phase that the point is below the clipping level. However, for example, as shown in FIG. 9, when the amplitude exceeds the clipping level continuously for two points and the correction signal is a high-frequency signal as indicated by the dotted line in the figure, the two points are continuously stored within the clipping level. In this case, the phase modulation applied to the correction signal becomes very large. That is, the difference in phase addition between the two points becomes very large. As a result, the spectrum of the correction signal is significantly widened. In this regard, the present invention does not attempt to keep both points within the clipping level.

  Specifically, for example, only the point having a larger amplitude is corrected, the other is not changed, and the phase of the correction signal is not fixed. Alternatively, the other point is a phase change in an allowable range from the phase of the point having a large amplitude, and is fixed by adding a phase in a direction of decreasing the amplitude to some extent. As another method, even if the clipping level is not lowered to each, the amplitudes are both lowered little by little, or the vicinity of the phase where the amplitude of the correction signal becomes 0 is selected, and the amplitude is changed little by little. You may take a method.

  The determination as to whether it is necessary to apply a restriction on the phase change may be performed as follows. First, the amount of phase addition necessary to reduce the amplitude to the clipping level for that point is calculated.

  In many cases, the phase of the correction signal required to lower the amplitude of the listed points to the clipping level is also in the range of ± π. (If 0 or ± π is exactly one way). Among these, the phase closer to the phase of the correction signal given at the beginning of the processing at that point is selected. This is because no large phase change is given to the correction signal. Therefore, the closer phase may be selected even in a procedure that does not limit the phase change.

A phase addition is calculated from the phase determined in this way. Next, the phase addition of the adjacent point where the phase is already fixed and the time difference (number of samples) to that point are examined. When phase interpolation is performed with a fixed pattern such as a half-cycle sine wave, the secondary differential function of the fixed pattern is known. For example, a sine wave is represented by A · cos (B · t), and its second derivative is AB ² · cos (B · t). Therefore, the maximum point of the absolute value of the second derivative is also known in advance, and the maximum value is simply substituted into the equation if the difference in phase addition from the adjacent point where the phase is already fixed and the time difference are known. Can be obtained. The same applies to the difference, and if the maximum absolute value of the difference of the phase difference obtained in this way exceeds the predetermined upper limit value, the phase change is limited. It can be judged that it is necessary to add.

  When the interpolation function is not a fixed pattern, various methods can be used. Even if the interpolation function is not fixed, if it can be expressed by an expression, an expression corresponding to the second derivative is prepared in advance. Since the equation of the interpolation function is determined from the determined phase and, in some cases, the parameters of the surrounding phase, the parameter used for the equation of the second derivative is also determined. As a result, the value at each point of the second derivative is obtained, and the maximum value is obtained. If this exceeds a predetermined upper limit, it is determined that the phase change should be limited.

  When phase interpolation is performed up to the point where the adjacent phases on both sides are fixed, the both sides are naturally examined. If it is determined that the phase change should be limited in any case, the phase change is limited in accordance with the parameter requiring the phase change limitation.

  The example of FIG. 9 is a case where the clipping level has been exceeded for two consecutive points. On the other hand, when the restriction on the phase change is added, as in the case where two points are continuous, the phase change of the correction signal is stopped for the later point in the order of listing and processing, It is possible to apply a method of changing to a limit that does not limit the phase change, or giving a phase change little by little to both of the two adjacent listed points in question. In addition, when there is a margin in the phase of the point where the adjacent phase is fixed, that is, when the amplitude obtained by adding the data signal waveform and the correction signal at that point does not reach the clipping level, there is a direction to lower it below the clipping level. In the case where the restriction due to the phase change is relaxed, the adjacent fixed phase may be changed. In that case, it is necessary to perform the phase interpolation again on the other side of the point, and at that time, it is necessary to check whether or not the phase limitation occurs.

  In this way, a point that does not fall within the clipping level occurs, but rather than giving an unreasonable phase change to the correction signal, out-of-band radiation can be reduced as a result of giving up and clipping a small number of points.

  In the above procedure, when determining the phase of the correction signal for the listed points, the phase is determined so that the point is exactly the clipping level. From the viewpoint of amplitude, it may be within the clipping level, and need not be exactly. However, if the clipping level is exceeded in the phase of the correction signal given at the beginning of the processing at that point, the phase for making it within the clipping level with the smallest phase change from that phase is The sum of the data signal waveform and the correction signal is a phase that is just the clipping level. Therefore, unless there are special circumstances, the clipping level may be set exactly.

  In the above procedure, it was written that the phase was selected from the range of ± π, but actually, it was selected from any of the phases obtained by adding 2nπ (n is an integer) to the optimal phase within the range of ± π. That's fine. If the difference between the added phase and the adjacent point whose phase has already been fixed becomes smaller when a point shifted by −2π, 2π, or 4π is selected, such a point may be selected.

  As a method of realizing a gradual phase change over the entire correction signal, the following method is possible in addition to the method of sequentially fixing the phase for the listed points as described above.

  First, for the points listed, a range for phase addition such that each point falls within the clipping level is calculated. In FIG. 10, each point and a straight line connecting two points vertically indicate the range at each list-up point. FIG. 10 shows a case where a limit is applied to the correction signal amplitude, and the point where the clipping level exceeds the upper limit value of the amplitude is indicated by only the point with no range.

  Next, a gentle curve passing through these ranges is calculated. The gentle curve may be calculated with a spline curve of an appropriate degree. In addition, the range and point obtained by adding ± 2nπ to these ranges and points are also the ranges and points that can be added to the phase, and therefore, it is preferable to select a range and a point from which a gentle curve can be easily drawn.

  At this time, if it is not possible to connect all the ranges with the gradualness applicable to the curve, it is preferable that the point giving an unreasonable phase change is made close to it without trying to connect it forcibly. In the example of FIG. 10, if an attempt is made to pass the curve for the point V1, an unreasonable phase change is given, so this is kept close without being forced to pass.

  In FIG. 10, it can be seen that the point V1 is not so far from the range of nearby listed points. Nevertheless, it is difficult to pass here because the point V2 is passed. In such a case, it is preferable to check which of the points V2 and V1 exceeds the clipping level, and pass the curve that exceeds the clipping level more closely, and keep the other closer.

  In FIG. 10, the differential values for the added phase at both ends of the symbol are substantially the same. This is to prevent a discontinuous point from occurring in the correction signal phase after adding the cyclic prefix as described above.

  The gentleness applicable to the curve can be determined by verifying whether or not the absolute value of the value corresponding to the second derivative of the correction signal added to the phase falls within a predetermined upper limit as described above.

  When a gradual phase change is realized in this way, a more gradual phase change can be realized than in the case where the phase is fixed sequentially for each listed point, and the spectrum spread of the correction signal is kept small. , Out-of-band radiation can be made smaller.

  As other exceptional processing in the embodiment of the present invention, there are a case where only one point is listed, or a case where there is a plurality of points and it is not necessary to apply phase modulation to the correction signal by chance. It is done. In such a case, the correction signal has a complete line spectrum. This is very desirable in terms of reducing the spread of the correction signal. However, if the correction signal has a certain level of power at this time, the spectral density will be much higher than that of the other subcarriers. If the output power limit for the transmitter is based on total power, there is no problem in this situation. However, if the output is limited to a spectrum mask, the correction signal may deviate from the spectrum mask if it becomes a line spectrum.

  In such a case, for example, the correction signal may be intentionally subjected to phase modulation. In the case where only one point is listed, for example, it is preferable to apply a very gentle phase modulation with a period of one symbol length. When there are a plurality of points, it is preferable to extract two points having a relatively long time interval and rotate the phase one time between them.

  As another method for dealing with such a situation, there is a method of subjecting the correction signal to amplitude modulation. That is, the correction signal has an amplitude only in the vicinity of the listed points. An example is shown in FIG. In FIG. 11, only the 1st and 85th sample points are listed. Therefore, Gaussian amplitude modulation is applied around the point 85 so that there is an amplitude only in the vicinity thereof.

  Note that this method leaves the carrier component popping out. As a result, if the spectrum mask is not satisfied, the above-described phase modulation may be used in combination. Even when phase modulation is used in combination, the amount of spread by phase modulation is relatively small because the amplitude is small in the portion where the phase rotation is intense. As a result, even if amplitude modulation and phase modulation are used together, it is possible to realize a correction signal spectrum that is compact and has no line spectrum pop-out.

  The same is true when the phases of a plurality of listed points coincide by chance, leaving the amplitude only in the vicinity of each point. If the spectral mask is still not satisfied, phase modulation is used together.

  Not only in such exceptional cases, but also by using amplitude modulation, it is possible to suppress the generation of a line spectrum in the correction signal and to further reduce the spurious by reducing the power of the correction signal. .

  For example, in the case of the correction signal as shown in FIG. 3B, the correction is performed if the amplitude of the correction signal is 0 for the 80th to 180th sample points in comparison with FIG. 3A. It can be seen that there is no point that needs to be listed as the target of. On the other hand, it can be seen from FIG. 3B that there is almost no phase change in the correction signal within this range. Therefore, in this range, the amplitude of the correction signal is set to 0, and the amplitude is gradually reduced at the transition to this range.

  When amplitude modulation is used in this way, the simplest procedure is to apply the amplitude modulation collectively after generating the correction signal as in the case of FIG.

  On the other hand, by incorporating amplitude modulation into the procedure from the beginning, it is possible to reduce the number of points that require phase correction, and as a result, to realize a compact spectrum of the correction signal.

  FIG. 12 shows a flowchart of the process.

  Only the points that exceed the clipping level are listed up first (step S1201). Based on these, the phase of the correction signal is determined in the same procedure as before (step S1202). Next, the interval between each listed point, that is, the time interval between adjacent listed points is examined. If the time interval between adjacent points is equal to or greater than a predetermined value, the correction signal is subjected to amplitude modulation so as to reduce the amplitude between them (step S1203).

When the amplitude of the correction signal is determined in this way, next, for each point of the data waveform signal, a point where the value obtained by adding the amplitudes at that point of the correction signal exceeds the clipping level is listed (step S1204). The correction signal is expressed in a form such as A _j · cos (2π * 121 * t + φ _j ), but the amplitude of the correction signal to be verified by addition is A _j . That is, it is verified whether there is a possibility of exceeding the clipping level when the phase of the correction signal changes variously, and a list of possible points is listed. Next, the phase of the correction signal is determined for these points in the same procedure as before (step S1205).

  In this way, points that require a phase change are listed based on the amplitude of each point of the correction signal after being subjected to amplitude modulation, so the number of points to be listed is reduced. For example, in the waveform example of FIG. 3, when amplitude modulation is not applied, some points exceed the list-up level between points 80 and 180, and those points are listed. In the method of listing after modulation, some of them are not even listed. As a result, the amount of phase modulation is reduced, and the power of the correction signal itself is reduced due to the effect of amplitude modulation, so that the spectrum of the correction signal can be made compact.

  There are various ways of applying the amplitude modulation to the correction signal. For example, when the phase modulation processing is completed for the points listed first, a loose pulse-like amplitude modulation waveform having a predetermined shape is placed around each point. At this time, if there is not a sufficient interval between adjacent points, amplitude modulation is not applied. This will be described with reference to FIG. In the case of FIG. 13A, the four points a1 to a4 exceeded the clipping level. After applying the phase modulation, the interval between the points was examined and exceeded a predetermined value except between a3 and a2. A loose pulse-like amplitude modulation basic waveform is arranged around each point except between a3 and a2. If there is an overlapping part such as between a1 and a4, the overlapping part is added. Between a3 and a2, the original amplitude of the correction signal remains unchanged. The amplitude modulation waveform synthesized in this way is as shown in FIG. The correction signal is modulated with such an envelope.

  It is desirable that the amplitude modulation basic waveform itself has a compact spectrum. For example, a waveform based on a Gaussian waveform or a sine wave for one cycle. Since the spectrum is expanded even by amplitude modulation, it is desirable that the temporal expansion of the amplitude modulation basic waveform is so large that the spectrum is not excessively expanded.

  The time interval between the two points listed as a boundary for whether or not to apply amplitude modulation may be, for example, an interval corresponding to the full width at half maximum of the amplitude modulation basic waveform. That is, in a1 and a4 in FIG. 13A, there is an overlap between the amplitude modulation basic waveforms, and the overlapping portion is added when the amplitude modulation waveforms are synthesized, but any point of the amplitude after the addition is the amplitude modulation basic waveform. The width should not exceed the maximum value of the waveform (1 if normalized).

  Next, the frequency at which the correction signal is arranged will be described. In the example of FIG. 2 described above, a part of the subcarriers forming the OFDM symbol is allocated for the correction signal. The correction signal is assigned to a high frequency subcarrier with respect to a subcarrier for mapping data. If the correction signal is arranged at a high frequency, the correction signal becomes a signal that vibrates in a short cycle, so that it is difficult to affect the vicinity of the point to be corrected, and a fine operation is easily applied.

  Of course, it may be arranged on the low frequency side or near the center of the subcarrier group forming the symbol. However, since the correction signal is phase-modulated and the spectrum is broadened, it is necessary to arrange the correction signal subcarriers together. In order to reduce the sample rate of the D / A converter as a complex amplitude, it may be arranged on the high frequency side of either plus or minus frequency.

  Moreover, it is not always necessary to assign a part of the OFDM symbol to the correction signal. Since the OFDM symbol is created based on FFT and IFFT, the number of subcarriers is usually a factorial of 2. Therefore, in the case shown in FIG. 2, when IFFT is performed on subcarriers to which data is mapped, IFFT is performed at 128 points even if there are only 112 subcarriers. On the other hand, as can be seen from the above description, the correction signal is treated as a time domain signal from the beginning, and processing such as IFFT and FFT is not applied. Therefore, it is not necessary to arrange the correction signal on the subcarrier in the OFDM symbol. An OFDM symbol may be normally created, and a correction signal may be arranged outside the OFDM symbol. However, in this case, the OFDM symbol in the time domain needs to be oversampled so that the frequency outside the subcarrier can also be expressed. Further, considering that the object of the present invention is to suppress out-of-band radiation, when the correction signal is out of band, the frequency may be a frequency at which unnecessary radiation may be emitted with a specified spectrum mask. Need to be.

  If symbols are formed in the digital part with a very high oversampling rate and OFDM symbols with different frequencies are created at the same time, one correction signal should be used for these symbols. You can also. In this case, the correction signal may be a subcarrier within the band of any symbol or may be outside the symbol.

  Since the correction signal has no information, it need not be sent to the receiving side. Therefore, regardless of whether or not a part of the symbol is divided into the correction signal, if a specific condition is satisfied, a filter is arranged in the analog part after the D / A conversion so that a part of the correction signal is removed. A filter may be applied. For example, when the frequency of the end or outside of the symbol is assigned to the correction signal and the purpose of reducing the PAPR is only the clipping suppression of the D / A converter, and the PAPR in the power amplifier does not become a problem, for example, the power amplifier This is the case for transmitters that do not use.

  Further, when the correction signal is outside the band of the OFDM symbol and the frequency is sufficiently away from the OFDM symbol, the correction signal may be removed from the antenna after passing through the power amplifier.

  Next, a processing method in the case of forming OFDM symbols on a complex number basis in order to reduce the sample rate of the D / A converter will be briefly described. The configuration of the present invention is effective regardless of whether the subcarrier and the time waveform are complex numbers. This is because the amplitude can be defined as an amplitude and the phase can also be defined as a phase regardless of whether it is a complex number. However, just in case, the handling of complex numbers is described below.

  When handling symbols as complex numbers, the real part (I) and imaginary part (Q) of the complex signal in the time domain are separated before D / A conversion, and each is converted to an analog signal by a separate D / A converter. Thereafter, recombination is performed using a quadrature modulator. FIG. 14 shows the complex plane. The horizontal axis is the real part and the vertical axis is the imaginary part. One point of the data time waveform is expressed as 12 on the complex plane, for example. For the time being, the amplitude corresponding to the clipping level is shown as 16 in a circular shape.

  The point indicated by reference numeral 1401 in FIG. 14 is that the amplitude exceeds the clipping level. Therefore, it is necessary to add the correction signal to fall within the clipping level. The amplitude of the correction signal is set such that the point that exceeds the clipping level the most within the clipping level as before. For example, if the point indicated by reference numeral 1401 in FIG. 14 is a point of maximum amplitude, the amplitude of the correction signal and the phase offset which is the initial phase, when added to this vector, have a length without changing its phase. The amplitude and phase are determined by a vector that allows only the clipping level. It is assumed that the point 1401 is not the maximum point now, and the amplitude has already been determined by another point. It is assumed that the phase of the correction signal at the beginning of processing this point is indicated by 1402. Even if this vector is added to 12 vectors, it does not fall within the clipping level. Therefore, if the phase of this vector is rotated to a phase as indicated by 1403, the sum of the data and the correction signal is just the clipping level. In this way, the phase is corrected. As can be seen from the description of FIG. 14, the amplitude and phase are simply treated as complex numbers, and the procedure is exactly the same as described above. Accordingly, the order of listing, exception processing, gentle phase interpolation, amplitude modulation, etc. may be performed in exactly the same way.

  In FIG. 14, the maximum value of the allowable amplitude is the same value regardless of the phase and is shown by a circle, and this is called a clipping level for convenience. In the case of handling as a complex signal as described above, I and Q are converted by separate D / A converters. Accordingly, a clipping level is set for each D / A converter. Therefore, the true clipping level is not a circle but a square as shown in FIG. If the purpose of PAPR suppression is only clipping suppression by the D / A converter, a rectangular clipping level as shown in FIG. 15 may be set. In this case, whether a point is listed or whether it is within the clipping level after adding the correction signal is determined not only by the amplitude but also by the phase and by the clipping level at that phase. The level corresponding to the list-up level in FIG. 2 is a square inside the clipping level 1502 as indicated by 1503. Although the clipping level 1405 in FIG. 14 is shown as an allowable amplitude value 1504 in FIG. 15, one point 1401 of the data signal exceeding the clipping level shown in FIG. 14 is a square clipping level as shown in FIG. So, it is within the clipping level and exceeds the list level. By making the clipping level square corresponding to the two D / A converters in this way, the number of points that exceed the clipping level is reduced, and the amount of excess is reduced, so that the phase modulation to the correction signal is small. Thus, the correction signal can be realized with a small power and a compact spectrum, and the out-of-band radiation can be reduced.

Note that PAPR suppression is often intended not only for clipping in a D / A converter but also for reducing backoff of a power amplifier. Since the power amplifier handles the signal after combining I and Q, PAPR is determined by the amplitude of the combined signal. Therefore, it is desirable to use a circular clipping level if an effect on the power amplifier is expected.

It is a block diagram which shows the structure of the OFDM transmitter of this invention. It is a figure which shows the example of the subcarrier arrangement | positioning of the correction signal and data signal in this invention. It is a figure which shows an OFDM time waveform and a correction signal waveform. It is a flowchart which shows the production | generation procedure of a correction signal. It is a figure which shows the waveform of a correction signal. It is a figure which shows a mode that the phase of a correction signal is interpolated. It is a figure which shows the waveform of the OFDM data signal after correction | amendment. It is a figure for demonstrating the method for adding a restriction | limiting to the amount of phase changes. It is a figure for demonstrating the exception process in this invention. It is a figure for demonstrating the method of determining the phase of a correction signal. It is a figure which shows the example which applies an amplitude modulation to a correction signal. It is a flowchart which shows the procedure which applies an amplitude modulation to a correction signal. It is a figure for demonstrating the example of the determination procedure of the envelope in the case of applying an amplitude modulation to a correction signal. It is a figure for demonstrating a mode that this invention is implement | achieved in complex amplitude. It is a figure for demonstrating the modification in the case of complex amplitude.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... OFDM transmitter 101 ... Data mapping part 102 ... Inverse Fourier transform part 103 ... Correction signal generation part 104 ... Adder 105 ... Digital-analog converter 106 ... Analog Part 107 ... Antenna 108 ... Digital part

Claims (13)

Subcarrier allocating means for allocating a group of adjacent subcarriers to a peak suppression correction signal and allocating a group of subcarriers different from the subcarrier allocated to the correction signal to a subcarrier for data transmission;
Mapping means for mapping data to the data transmission subcarrier;
Waveform conversion means for converting a data transmission subcarrier mapped with the data into a time data waveform;
A sine wave time waveform corresponding to one of the correction signal subcarriers is generated, the amplitude of the correction signal is determined, and the point to be subjected to peak suppression for the time data waveform is the amplitude of each point of the time data waveform Based on the absolute value, the phase of the correction signal at the time corresponding to each of the listed points is determined so that the absolute amplitude value of each of the listed points is approximately within a predetermined range. And a correction signal generating means for determining a phase of the correction signal at a time other than the listed points and outputting the correction signal with the phase determined,
An OFDM transmitter comprising: an adding means for adding the time data waveform and the correction signal output from the correction signal generating means and outputting an addition signal. The OFDM transmitter further includes digital / analog conversion means for converting the addition signal into an analog signal, and the correction signal generation means has a maximum amplitude absolute value in the time data waveform, and The absolute value of the difference between the amplitude of the maximum amplitude point and the clipping level for the maximum amplitude point exceeding the clipping level of the digital / analog converter is defined as the amplitude of the correction signal. The described OFDM transmitter. The correction signal generating means further uses the amplitude of the correction signal as the threshold when the absolute value of the difference between the amplitude of the maximum amplitude point and the clipping level exceeds a predetermined threshold. Item 5. The OFDM transmitter according to Item 2. The correction signal generating means lists, as a peak suppression target, a point whose amplitude absolute value is greater than a value obtained by subtracting the amplitude of the correction signal from the absolute value of the clipping level among the points of the time data waveform. The OFDM transmitter according to claim 2 or claim 3, wherein When the correction signal generating means determines that the amplitude absolute value of each of the listed points is approximately within a predetermined range, the correction signal generating means The OFDM transmitter according to claim 1, wherein the phase is determined. The correction signal generation means determines the phase of the correction signal at a time other than the listed points so that the phase of the correction signal gradually changes within a symbol period of the time data waveform. The OFDM transmitter according to claim 1. The correction signal generating means is an absolute value of a difference between the phase of the correction signal corresponding to the start time of the symbol period of the time data waveform and the phase of the correction signal corresponding to the end time of the symbol period of the time data waveform. The OFDM transmitter according to claim 6, wherein the phase of the correction signal is determined so that does not exceed a predetermined threshold. The correction signal generation means determines the phase of the correction signal for each of the listed points, and then determines an adjacent point and a phase of which the phase is determined among the listed points. When the adjacent point does not exist by the end of the correction signal long end, the correction signal of the correction signal is determined for each point between the adjacent points when the correction signal is regarded as a repetitive signal having a period corresponding to the symbol period. The OFDM transmitter according to claim 6 or 7, wherein the phase is determined so as to change gradually. When the correction signal generation means determines the phase of the correction signal for each point between the points listed and the adjacent point whose phase is determined, the difference between the phases of two adjacent sample points And calculating the difference between adjacent phase differences, and determining the phase of the correction signal so that the absolute value of the difference difference does not exceed a predetermined upper limit value. The OFDM transmitter according to claim 6. In the step of determining the phase of the correction signal so that the amplitude absolute value of each of the listed points is approximately within a predetermined range, the correction signal generating means is configured to determine the absolute amplitude of any of the listed points. When the phase of the correction signal is determined so that the value falls within a predetermined range, if a gradual phase change within the range corresponding to the data symbol period of the correction signal cannot be realized, or the absolute difference of the difference If the value exceeds the upper limit value, the absolute value of the difference between the listed differences is such that even if the amplitude absolute value exceeds the allowable range, the phase changes gradually even if the absolute value exceeds the allowable range. 10. The OFDM transmitter according to claim 6, wherein the phase of the correction signal is determined so as not to exceed the upper limit value. The correction signal generating means determines the phase range of the correction signal for all the points listed so that the amplitude falls within the predetermined range, and then the phase of the correction signal falls within the phase range. 2. The OFDM transmitter according to claim 1, wherein the phase of the correction signal is determined so as to change gradually within a symbol period of the time data waveform. When the correction signal determining means determines the phase of the correction signal so as to change gradually within a symbol period of the time data waveform, the phase of the correction signal is the phase for all the listed points. In the case where a gradual phase change cannot be realized if the phase is within the range, the symbol period of the time data waveform is set by making any one or more of the listed points out of the phase range. 12. The OFDM transmitter according to claim 11, wherein the phase of the correction signal is determined so as to change gradually. The subcarrier allocation means sets a frequency of a group of adjacent subcarriers allocated to the correction signal outside a band defined by a subcarrier having a minimum frequency and a subcarrier having a maximum frequency for the data transmission subcarrier. 2. The OFDM transmitter according to claim 1, wherein the OFDM transmitter is assigned.

Images