JPH09116521A - Generation and decoding method for frequency division multiplex signal - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the peak power by controlling the arithmetic method based on the arithmetic result obtained at a prescribed stage of an inverse discrete Fourier transform operation. SOLUTION: An arithmetic part applies the inverse discrete Fourier transforms to the digital information signals which are inputted to plural input terminals. The arithmetic result obtained at a stage right before the final stage of the arithmetic part is divided into the first half parts E0 to E255 and the second half parts H0 to H255 . Then an operation is carried out with the parts E0 to E255 used as they are and the parts H0 to H155 set at zero respectively, so that a 1st arithmetic result F1(t) of the final stage is acquired. At the same time, a 2nd arithmetic result F2 (t) of the final stage is obtained through an operation carried out with the parts E0 to E255 set at zero and the parts H0 to H255 used as they are respectively. The normal final stage result F(t), if defined so, is equal to F1(t)+F2(t). If the result F(t) includes the peak value higher than a prescribed level, the data string of the result F2(t) is shifted and added to the result F1(t). As a result, the peak value can be reduced.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a frequency division multiplex signal generation method and a decoding method, and more particularly to orthogonal frequency division multiplexing (OFDM) of encoded digital video signals and the like in a limited frequency band.
The present invention relates to a method for generating and decoding a frequency division multiplex signal which is converted into a signal and transmitted and received.

[0002]

2. Description of the Related Art One of the methods for transmitting coded digital video signals in a limited frequency band is 25
6 Quadrature Amplitude Modul (QAM)
ation) and other multi-value modulated digital information is transmitted as an OFDM signal using a large number of carriers, the OFDM system is more resistant to multipath, less susceptible to interference, and has relatively good frequency utilization efficiency. Are known. This OFDM system is a system in which a large number of carriers are arranged orthogonally and independent digital information is transmitted on each carrier. The phrase "carriers are orthogonal to each other" means that the spectrums of adjacent carriers become zero at the frequency position of the carrier.

According to this OFDM system, since a guard band period (guard interval) is set and information of the period is transmitted redundantly, transmission distortion caused by multipath of radio waves can be reduced. That is, in the reception of this OFDM signal, the amplitude and phase modulation components of the signal transmitted within the symbol period are detected, and the information value is decoded by these levels, so the signal of the first guard interval period is By removing and decoding
Since the frequency components of the multipath signal in the same symbol section and the signal to be received are the same, decoded digital data with less transmission distortion can be transmitted in a relatively narrow frequency band.

[0004]

However, the above OF
In the conventional frequency division multiplexing signal generation method for generating a DM signal, a large amount of power is rarely generated because no measure is taken against the peak power that occurs instantaneously for an OFDM signal formed by combining a large number of information carriers. It may be done. For example, the power of an OFDM signal using 256 information carriers is an average power that is 256 times the power of one information carrier, so if the maximum amplitude voltage values of all the information carriers are generated in agreement, , 25 of one carrier
Transmission power of 6 times (or D / A converter, dynamic range of A / D converter, linearity of analog system, etc.) is required. Conversely speaking, the signal-to-noise ratio (S / N) per carrier decreases accordingly.

The probability that the phases of all the carriers are the same is very small and hardly occurs in reality, but the average power value is set to a low value with a margin, and the transmission power device also has an average power of 10 to 20 times. A device that can generate a large output signal with a certain margin is used, and it is considered that even a rarely generated high power signal can be transmitted without being saturated. Therefore, the conventional frequency division multiplex signal generator has a problem that the entire device is expensive and becomes large in size.

The present invention has been made in view of the above points, and by controlling the operation method according to a predetermined stage operation result of an inverse discrete Fourier transform operation for generating an orthogonal frequency division multiplex signal, the peak An object of the present invention is to provide a frequency division multiplex signal generation method and decoding method that can reduce power.

[0007]

In order to achieve the above object, the present invention is applied to a plurality of input terminals to generate a frequency division multiplexed signal composed of a plurality of carrier waves each modulated with a digital information signal. In a frequency division multiplexed signal generation method including an arithmetic unit for performing an inverse discrete Fourier transform of the digital information signal, the arithmetic result of the stage immediately before the final stage of the arithmetic unit is divided into two parts, a first half and a second half,
The first half is executed by setting the first half as it is and setting the latter half to zero to obtain the first calculation result F1.
(T) is generated, the first half is set to zero, and the second half is left unchanged to generate the second operation result F2 (t).
If the absolute value of the sum of the calculation results of
After at least one of the movement on the time axis and the multiplication by a predetermined value is performed on the calculation result F2 (t), the frequency division multiplexed signal is generated by adding and synthesizing to the first calculation result F1 (t). It is characterized by In the present invention, the peak value can be reduced to 1/2 of the normal value by performing the final stage twice.

Further, according to the present invention, when the X stages at the rear of the operation including the final stage of the operation unit are carried out, the operation result of the {(final stage) -X} th stage is divided into 2 ^X blocks. and, one of the divided blocks is used as is, to implement the X number of stages of the computation rear portion sets all the remaining blocks to zero was performed on all the blocks, 2 ^X number obtained When the absolute value of the sum of the calculation results of is greater than or equal to a predetermined value, at least one of movement on the time axis and multiplication of a predetermined value is performed on the 2 ^X calculation results, and then these 2 ^X calculation results are calculated. It is characterized in that the results are added and synthesized to generate a frequency division multiplexed signal. As a result, in the present invention, the peak value is reduced to 1 / th of the normal value by performing the stage operation X times from the Xth stage before the final stage.
Can be X.

Here, in the present invention, since it is necessary to transmit the calculation information to the receiving side, the input signal of the specific input terminal of the calculation unit is set to zero, the carrier wave assigned to the specific input terminal is used as the carrier hole, and the calculation is performed. As the signal transmitted through the carrier hole in the output frequency division signal of the second part, the operation information that is moved on the time axis and / or multiplied by the predetermined value is added and synthesized, or the initial input data of the second half part is added with the time. A specific carrier wave in which at least one reference data is inserted is set as the operation information obtained by the movement on the axis and / or the multiplication by a predetermined value, or the time is added to the initial input data of each of the 2 ^X blocks. A specific carrier wave in which at least one reference data is inserted is set as calculation information obtained by moving on the axis and / or multiplying by a predetermined value.

In order to achieve the above object, the decoding method of the present invention is a decoding method for decoding a digital information signal by performing a discrete Fourier transform on a frequency division multiplexed signal composed of a plurality of carrier waves each modulated with a digital information signal. In claim 3, the operation information transmitted in a predetermined carrier hole in the frequency division multiplex signal according to claim 3 is decoded, and the decoded digital information signal is corrected with the decoded operation information.

Further, in order to achieve the above object, the decoding method of the present invention decodes a digital information signal by performing a discrete Fourier transform on a frequency division multiplexed signal composed of a plurality of carrier waves each modulated by the digital information signal. Or decoding the reference data transmitted on a specific carrier in the frequency division multiplexed signal according to claim 4 and correcting the decoded digital information signal with the decoded reference data,
The reference data transmitted on a specific carrier in the frequency division multiplex signal according to claim 5 is decoded, and the operation result is corrected with the decoded reference data for each of the decoded 2 ^X blocks. is there.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, embodiments of the present invention will be described with reference to the drawings. First, before describing the frequency division multiplex signal generation method of the present invention, an outline of an OFDM signal transmission apparatus to which the frequency division multiplex signal generation method of the present invention is applied will be described. Here, transmission information is transmitted as an OFDM signal using 256 carriers. Further, in order to facilitate the design of the analog signal system in the latter stage, it is assumed that the double oversampling is used and the 512-point inverse discrete Fourier transform (IDFT) operation is executed to generate the OFDM signal.

In this transmitting apparatus, for example, digital data to be transmitted such as a digital video signal and an audio signal compressed by an encoding system such as an MPEG system which is a color moving image encoding display system is supplied to the arithmetic unit 4. This computing unit 4 performs inverse discrete Fourier transform (ID
FT) operation is performed to generate an in-phase signal (I signal) and a quadrature signal (Q signal). This arithmetic unit operates at a sample clock frequency higher than a predetermined frequency bandwidth. When transmitting transmission information with 256 carriers, double oversampling is used and 512-point IDFT operation is performed to generate a signal. The input allocation to the IDFT calculation unit at this time is as follows if the numbers are sequentially assigned in the input frequency alignment type.

N = 0-128 An information signal for modulating a carrier is provided.

N = 129 to 383 The carrier wave level is set to 0 and no signal is generated.

N = 384-511 An information signal for modulating a carrier is provided.

That is, the number of input terminals of the IDFT calculation unit is 512 for the real part (R) signal and 512 for the imaginary part (I) signal, of which 127 (n = 1) to 127 are the first.
127 pieces in total up to the nth (n = 127) and 1 pieces in total from the 385th (n = 385) to the 511th (n = 511)
Information signals are input to 27 input terminals each, and 0
The DC voltage (constant) is input to the th (n = 0) th input terminal and transmitted at the center frequency of the carrier wave.
For example, a fixed voltage for the pilot signal is input to the 2nd (n = M / 4) and 384th (n = 3M / 4) input terminals, and both ends are equivalent to a frequency that is 1/2 the Nyquist frequency. It is transmitted by a carrier wave of the frequency.

Here, a total of 1 from the 1st to the 128th
The input information of the 28 input terminals is transmitted by the information transmitting carrier wave on the upper side (higher band side) of the central carrier wave frequency F0.
The input information from a total of 128 input terminals from the 1st to the 511th is transmitted on the information transmission carrier below the center carrier frequency (low side). Also, the remaining 129th to 38th
0 is input to the third input terminal (set to the ground potential) so that the carrier wave in that portion is not generated (not used for data transmission).

That is, in the arithmetic unit, the transmission information from the external system is "AB", "CD", "E" in units of 8 bits.
In the order of F "," GH ", ... (each character represents a 4-bit block), the 1st to 128th real and imaginary part input terminals and the 384th to 5th
A 4-bit signal is input to each of the eleventh real part input terminal and the imaginary part input terminal. In this case, the carrier number, data of the real part input terminal, and data of the imaginary part input terminal are assigned in the following first arrangement.

[0020]

[Table 1] Furthermore, since the reference data for the amplitude / phase correction on the receiving side and the synchronization data etc. (including the transmission mode) are inserted on the specific carrier, the data corresponding to these will be written later. It is transferred to another carrier.

The IDFT operation result (I signal and Q signal) of the above operation unit is supplied to the quadrature modulation means via the output buffer, where it is quadrature-modulated and 257 waves (128 sets of positive and negative carrier waves having mutually different frequencies) are supplied. And one central carrier) each of which has 256 QAM modulated OFD
After being converted into an M signal, it is frequency-converted into a transmission frequency band by a frequency converter, further power-amplified in the transmission unit, and radiated from the antenna.

In the frequency division multiplex signal receiver,
After the orthogonal demodulation and the DFT calculation, the DFT calculation result is corrected according to the reference data of the specific carrier.

Next, to explain an embodiment of the present invention, FIG. 1 shows a block diagram of an embodiment of the frequency division multiplexed signal generating method according to the present invention. In the figure, the arithmetic unit 1 is embodied by a digital signal processor (DSP), and a transmission information signal (such as the digital data) from an external system (not shown) corresponds to each carrier wave in the bit reverse order. It is supplied to the input terminal, the above-mentioned IDFT calculation is performed, and movement of the calculation result on the time axis is performed by the method of each embodiment described later so that a peak value of a predetermined value or more does not occur in the obtained calculation result. The multiplication of a predetermined value and the like are performed, and the calculation result obtained thereby is output to the output buffer 2.

At this time, reference data serving as a reference for decoding of the receiving system is input to the real number part and the imaginary number part input terminals for generating the first carrier, and IDFT operation is performed, and then output to the output buffer 2.

The IDFT operation result (I signal and Q signal) of the operation unit 1 is generated in bursts as time axis signals (I signal and Q signal) of 512 points of 256 pieces of input information in one IDFT operation. On the other hand, since the circuit in the subsequent stage needs to perform constant and continuous signal processing, the IDFT operation result is temporarily stored in the output buffer 2 in order to adjust the time difference between the two. After that, the I signal and the Q signal are continuously read from the output buffer 2 and input to the quadrature modulation means (not shown).

The frequency division multiplexed signal transmitted as described above is orthogonally demodulated by the orthogonal demodulator, and then supplied to the DFT operation section 10 shown in FIG. 10 for DFT operation to decode the transmission information. Here, the received OFDM signal is processed by the calculation unit 1 on the transmission side so as not to generate a peak value equal to or more than a predetermined value, as will be described later.
At the time of decoding by the FT calculation unit 10, reference data or calculation information for identification is added and transmitted so that the original data can be decoded. The DFT calculation unit 10 corrects the received data and restores the original data by a method described later based on the reference data or the calculation information.

(First Embodiment) The arithmetic unit 1 of FIG. 1 is assumed to perform 512 (= 2 ⁹ ) points of IDFT operations in radix 2, time thinning, input data bit reverse type, and output data aligned type. . This IDFT operation requires operations from the first stage to the ninth stage, but in this embodiment, it is the final stage (that is, the ninth stage).
When the first half is set to E _{0 to} E ₂₅₅ and the second half is set to H _{0 to} H ₂₅₅ among the 512 calculation results obtained by the above, the first final stage is obtained. ,
E ₀ the to E ₂₅₅ as it is used to obtain the operation result by calculating H ₀ to H ₂₅₅ as zero F1 a (t). Further, as the second final stage, the arithmetic unit 1 sets E _{0 to} E ₂₅₅ to zero, and H
₀ to H ₂₅₅ to accept to operation to operation result F2
(T) is obtained.

It is self-evident that if the usual final stage result is F (t), this is F1 (t) + F2 (t). Here, assuming that F (t) has a peak value equal to or greater than a predetermined value, in the present embodiment, the peak value is reduced by shifting the data string of F2 (t) and adding it to F1 (t). Is.

Next, the fact that the final stage result F (t) becomes F1 (t) + F2 (t) will be described in detail. Now, in order to simplify the explanation, the above IDFT
If the operation is a 16-point IDFT operation, the butterfly operation from the first stage to the fourth stage as shown in FIG. 2 is usually executed on the data input in the bit reverse order.

Here, the numerical values in FIG. 2 indicate the value of n of the twiddle factor W ⁿ = exp (−j2πn / 16) of the butterfly operation. When each butterfly operation is expressed by a complex number, (R1 + jI1) + of (R1 + jI1) and (R2 + jI2) of (R1 + jI1)
(R2 + jI2) W ^n, for (R2 + jI2) (R1 + jI1) - carrying out the calculation of (R2 + jI2) W ^n. It should be noted that the ½ scaling for each stage operation is omitted throughout the description.

In the 16-point IDFT calculation, four stages are carried out, but as described above, the fourth stage which is the final stage is left in this embodiment, and the calculation result of the third stage is divided into the first half parts E _{0 to} E _7. And the latter half H _{0 to} H ₇ are divided and the fourth stage is calculated as shown in FIG. 3, the calculation result shown in the same figure is obtained.

Here, the operation result of the fourth stage (final stage) is defined as F (t), and this is used as two functions F1.
(T) and F2 (t). That is, F (t) = F1 (t) + F2 (t)

[0033]

(Equation 1) To obtain F1 (t), use E _{0 to} E ₇ as they are,
It is obtained by performing the operation of the fourth stage on the assumption that H _{0 to} H ₇ are zero. Alternatively, as is clear from FIG. 3, E ₀
It can also be obtained by repeating ~ E ₇ .

On the other hand, to obtain F2 (t), H₀~ H₇
Is used as is, E₀~ E₇Is zero and the fourth step
It is obtained by performing the calculation of the page. Or from Figure 3
As you can see, W⁰ H₀~ W⁷H₇Calculate up to and its polarity
It can also be obtained by using inversion.

Here, consider a case where an input data string {a ₀ , a ₁ , ..., A ₁₅ } is input as the transmission information. These 16 data values are frequency-allocated according to a predetermined rule. However, in order to make the following description easy to understand, the allocated input data string is assumed to be {b ₀ , b ₈ ,
_{b 4, b 12, b 2} , b 10, b 6, b 14, b 1, b 9, b 5, b
₁₃ , b ₃ , b ₁₁ , b ₇ , b ₁₅ }. The relationship between a _n and b _n is outside the scope of the present invention. This data string is input from the left side of FIG. 2 and subjected to IFDT operation.

From another point of view regarding the determination of F1 (t) and F2 (t), {b ₀ , b ₈ , b ₄ , b ₁₂ ,
b ₂ , b ₁₀ , b ₆ , b ₁₄ , 0, 0, 0, 0, 0, 0, 0,
0} result of IDFT operation input sequence of corresponding to F1 (t), {0,0,0,0,0,0,0,0, b 1, b 9,
The input sequence of b ₅ , b ₁₃ , b ₃ , b ₁₁ , b ₇ , b ₁₅ } is IDF.
It is obvious that the result of T calculation corresponds to F2 (t).

The final result (OFDM wave) to be transmitted is F
1 (t) + F2 (t), but if the result has a peak value equal to or more than a predetermined value, F2 (t) is shifted by several samples and added.

Next, the above addition method will be described.
As the addition method, the maximum of F1 (t) and F2 (t),
F2 (t) is shifted forward or backward by a few samples so that the minimum peak values do not overlap. F2 (t) is aligned so that the minimum peak point of F2 (t) matches the maximum peak point of F1 (t). Shifting t) forward or backward by a few samples, the point where the level of F1 (t) is large (for example, 5 points)
So that points with a high level of F2 (t) (for example, 5 points) do not match, and points with a low level of F1 (t) (for example, 5 points) and points with a low level of F2 (t) ( For example, shift F2 (t) so that 5 points do not match,
The method such as is possible.

Both F1 (t) and F2 (t) are duplicated.
It is a prime function and F1 (t) = F1_R(T) + jF1_I(T) F2 (t) = F2_R(T) + jF2_ILet (t) be F1_R(T) + F2_R(T) is I (In Ph
as) signal sequence, F2_I(T) + F2_I(T) is Q (Qu
signal train), so F2
_R(T), F2_ICondition where (t) is shifted by the same number of samples
, The peak value is deleted considering both the I signal sequence and the Q signal sequence.
It goes without saying that it will decrease. The calculation unit 1 of FIG.
Like F2 (T) is shifted so that the peak value is below a predetermined value.
And then to the quadrature modulator (not shown) via the output buffer 2.
Send data.

Here, as an example, in the calculation result of FIG.
The following table shows a shift of F2 (t).

[0041]

[Table 2] As can be seen from the rearranged results in Table 2 above,
This is equivalent to the result of the third stage calculation being the following expression.

H _t '= W ¹ H _{t + 1} mod (t, 8) Eight points in the latter half of the second stage operation result are C _{0 to} C ₃ , D _0.
.. D ₃ is usually calculated as shown in FIG. 4, but is equivalent to the result of the second stage calculation shown in FIG. 5 when calculated backward from the above formula.

Further, when the calculation is performed up to the input data string, the calculation is normally performed as shown in FIG. 6. However, the backward calculation is performed as shown in FIG. In conclusion, the transmission information, that is, the input data string is {b ₀ , b ₈ , b ₄ , b ₁₂ ,
_{b 2, b 10, b 6} , b 14, b 1, b 9, b 5, b 13, b 3, b
_11, b _7, the case of b _15}, by way shift described above, the input data _{sequence, {b 0, b 8,} b 4, b 12, b 2, b 10, b
₆ , b ₁₄ , W ¹ b ₁ , W ⁹ b ₉ , W ⁵ b ₅ , W ¹³ b ₁₃ , W
^{_{3 b 3, W 11 b 11}} , W 7 b 7, W 15 b 15} is equivalent to enter. The first half of the frequency allocation is left as it is, and only the latter half of the waveform is advanced by one sample time.

In the above case, all the input data have the same amplitude, the phases of the input data are the same from b _{0 to} b ₁₄ , and b ₁ is the fundamental frequency (one cycle for one symbol). assigned because it is in phase (hereinafter, the angle also referred) one fraction of W ^1, b _9, since assigned to the fundamental frequency × 9, ¹ 9 pieces of the W is as an angle, b ₅ is the fundamental frequency Since it is allocated to x 5, 5 angles of W ¹ are allocated as angles, b ₁₃ is allocated to the fundamental frequency × 13, so 13 angles of W ¹ are allocated as angles and b ₃ is allocated to the basic frequency × 3. Therefore, since 3 angles of W ¹ and b ₁₁ are allocated to the fundamental frequency × 11 as angles, 11 angles of W ¹ are allocated to the angle and b ₇ are allocated to fundamental frequency × 7. 7 pieces of the W ¹ are as, b _15, since assigned to the fundamental frequency × 15, as the angle 15 of the phase of W ¹ is considered to have entered the data advanced.

Next, a decoding method by the DFT operation unit 10 of FIG. 10 will be described. As an example, the transmitting side inserts reference data into the basic frequency allocation data, and the correction amount for decoding is determined based on this. The value of this reference data may be any value as long as it is a predetermined reference phase value. For example, when the value of phase 0 degree is decided as the reference data,
It is assumed that the data of b ₁ 'detects the phase α'degree. This value corresponds to ± several W ¹ .

In this case, since the phase of b ₁ is advanced by α ′ degrees, in order to restore the data of b ₀ , the phase of the data of b ₉ ′ is delayed by α ′ × 9 degrees and the data of b ₅ is changed. In order to restore, the phase of b ₅ 'data is delayed by α' × 5 degrees, b
To restore the ₁₃ data, b _{1 3} delays 'phase the α data' × 13 degrees, in order to restore the data of the b _3,
b ₃ delayed 'phases of α data' × 3 times, in order to restore data b _11, 'phase the α data' × b ₁₁
It delayed 11 degrees, to restore data b _7, delayed 'phases of alpha data' × 7 degrees b _7, in order to restore data b _15, 'the phase of the data of the alpha' b ₁₅ × The operation may be delayed by 15 degrees.

Considering the same, b_ThreeInsert the reference data into
When the phase is advanced by β'degrees, b₁Data
B to restore₁'Data phase is (β' / 3) ×
 Delay once, b₉B to restore the data of₉’De
Delay the phase of the data by (β '/ 3) × 9 degrees, b_FiveData of
B to restore_FiveThe data phase of ‘(β’ / 3)
× delayed by 5 degrees, b₁₃B to restore the data of₁₃’
Delay the phase of the data of (β ′ / 3) × 13 degrees, b₁₁
B to restore the data of₁₁’Data phase
(Β '/ 3) × delayed by 11 degrees, b₇Restore data
B₇'Data phase is (β' / 3) x 7 degrees
Delay, b_FifteenB to restore the data of_Fifteen'Data of
If the operation of delaying the phase of (β ′ / 3) × 15 degrees
Good. Even if reference data is inserted at other frequencies, the same
Can be processed.

Thus, in the embodiment of FIG. 1, the second
As the final stage of₀~ E₂₅₅To zero
And H₀~ H₂₅₅Is used as it is for calculation and the calculation result F
2 (t) is obtained. The data string of this F2 (t) is {K₀,
K₁, K_Two, ..., K₅₁₁｝. Data string is +1
Samples that have been shifted {K₁, K_Two, ..., K₅₁₁,
K₀｝. This operation is F2 (t) (time axis waveform)
Indicates that the sample was advanced by 1 sample time. Therefore, each frequency
Are respectively received with a phase lead of that time.
It is. The first half and second half of the input data string to the arithmetic unit 1 are
First half = {b₀, 0,0, b₃₈₄, b₆₄, 0,0, b₄₄₈, b₃₂, 0,0, b₄₁₆, b
₉₆, 0,0, b₄₈₀, b₁₆, ・ ・ ・, B₅₁₀} Second half = {b₁, 0,0, b₃₈₅, b₆₅, 0,0, b₄₄₉, b₃₃, 0,0, b₄₁₇, b
₉₇, 0,0, b₄₈₁, b₁₇, ・ ・ ・, B₅₁₁}, The first half of the DFT operation unit 10 of FIG.
Data is decoded as it is, and the second half is {W¹b₁, 0,
0, W³⁸⁵b₃₈₅, W⁶⁵b₆₅, 0,0, W⁴⁴⁹b₄₄₉, W³³b₃₃, 0,0, W
⁴¹⁷ b₄₁₇, W⁹⁷b₉₇, 0,0, W⁴⁸¹b₄₈₁, W¹⁷b₁₇, 0,0, W⁵¹¹b₅₁₁｝
And the data is decrypted. Note that WⁿIs the number of IDFT operations
The transfer factor, Wⁿ= Exp (-j2πn / 512)
ing.

Therefore, as the correction when the +1 sample is shifted, W ⁻¹ and W ⁻¹ ,
W ^-385 , W ^-65 , W ^-449 , W ^-33 , W ^-417 , W ^-97 , W
^Multiply by the correction amount of ^-481 , W ^-17 , ..., W ^-511 .

Actually, as a method of obtaining the correction amount when the sample is shifted by ± n samples, as a first method, a specific carrier is provided as a carrier hole, and after IDFT calculation,
The information shifted to the signal corresponding to the specific carrier may be added (by the time axis function). This can be easily performed by using the SIN table prepared for the IDFT calculation. The receiver performs correction based on this information. Since there is a fixed relationship between the data of each frequency and the correction amount, the amount of information can be expressed in 256 ways in 8 pits and 512 ways which are all shifts in 9 bits.

Here, as described above, the second method and
Then, the above-mentioned b₁Always insert reference data into. b₁Divided into
The frequency to be applied is the fundamental frequency (f
₁). At the receiver, b₁’(This is b₁Receiving day
Data) is received, (b₁'/ B₁) From f ₁ Rank
The phase difference is known, and the correction amount is (b₁/ B₁’)
It is. Based on this, the received data b_n’, (B₁
/ B₁’)ⁿb_nIs calculated as b_nIs found.

This will be described in more detail.
Now, assuming that the reference data S _TX to be transmitted is S _TX = x _S + jy _S , this is expressed by the following expression in polar coordinates.

X _S = _S cos (θ _S ), y _S = _S sin (θ _n ), where S = √ (x ² _S + y ² _S ), θ _S = tan ⁻¹ (y _S / x _S ). received reference data S _TX obtained by receiving the reference data S _TX 'is expressed by the following equation.

S _TX '= x _S ' + jy _S 'This is expressed by the following equation in polar coordinates.

X_S'= S'cos (θ_S'), Y_S'= S'sin (θ_S') However, S' = √ (x_S'^Two+ Y_S'^Two), Θ_S'＝ tan^-1(y_S'/
x_S') Basic frequency f₁The correction amount of is (b₁/ B₁’) = (X_S+ Jy_S) / (X_S’＋ Jy_s’) If this is expressed in polar coordinates, (b₁/ B₁') = (S / S') (cos (θ_S−θ_S') ＋ jsin (θ_S−θ
_S')) Here, only the phase is changed and the amplitude is the same, so S =
S ', f₁The correction amount of cos (θ_S−θ_S') ＋ jsin (θ_S−
θ_SGet '). Also, θ_SBy setting = 0
Is easier to calculate, cos (−θ_S') ＋ jsin (−θ_S')
Get. It should be noted that the angle has a positive direction to the left.
(In this case, θ_S'<0) On the other hand, transmission data D related to transmission information_TXn(= X + J
received data D obtained by receiving y)_TXn'= (X' + j
The correction for y ') is obtained from the following equation.

D _TXn = D _TXn ′ (cos (θ _S −θ _S ′) + jsin (θ _S −θ _S ′)) ⁿ = D _TXn ′ (cos (n (θ _S −θ _S ′)) + j sin (n (θ _S −
θ _S '))) In addition, by setting θ _S = 0, the received data can be calculated by the simpler calculation formula D _TXn = D _TXn ' ((cos (nθ _S ') -jsin (nθ _S ')). Is required.

When the reference data is set to b _m instead of b ₁ , the phase difference of f _m can be known from the calculation of (b _m '/ b _m ) and the correction amount is (b _m / b _m ' ) Is obtained. As is clear from the contents described above, the data correction is performed by D _TXn = D _TXn '(cos ((n / m) (θ _S −θ _S ')) + jsin
((N / m) (θ _S −θ _S '))).

Theoretically, θ _S −θ _S ′, (1 / m) (θ _S −
Since θ _S ') or θ _S ' is an integer multiple of the sample time, the SIN table for DFT calculation can be used for these calculations in the receiver, and the calculation can be performed easily.

(Second Embodiment) Next, the second embodiment of the present invention will be described.
An embodiment will be described. In the second embodiment, when the X stages of the arithmetic rear part including the final stage of the arithmetic unit 1 are executed, the arithmetic result of the {(final stage) -X} th stage is converted into a 2 ^X block. Divide and use one of the divided blocks as is, set all the remaining blocks to zero, and perform the X stages in the rear part of the operation, performed on all blocks and obtained 2 ^X When the absolute value of the sum of the calculation results is greater than or equal to a predetermined value, after moving on the time axis for 2 ^X calculation results,
A frequency division multiplexed signal is generated by adding and synthesizing these 2 ^X operation results.

Here, a case where the above X is "2" will be described. Now, the calculation result of the second stage is A
_It is divided into four blocks represented by _{0 to} A ₃ , B _{0 to} B ₃ , C _{0 to} C ₃ , and D _{0 to} D ₃ , and the operation result A ₀ to of the first block is divided.
A ₃ is used as it is, and the calculation results B ₀ to
B ₃ , C _{0 to} C ₃ , and D _{0 to} D ₃ are all set to 0, and the third and fourth stages are performed to obtain the calculation result F1 (t).

Similarly, the calculation result B of the second block₀~
B_ThreeIs used as is and the operation results of the remaining blocks A₀~
A_Three, C₀~ C_Three, D₀~ D_ThreeIs 0 for the 3rd and 4th
The calculation result F2 (t) and the third block
C calculation result C₀~ C_ThreeTo use the rest of the block
Calculation result A₀~ A_Three, B₀~ B_Three, D₀~ D_ThreeIs all 0
The third and fourth stages are performed as the calculation result F3 (t)
And the operation result D of the fourth block₀~ D_ThreeThe
Use as it is, operation result A of the remaining block₀~ A_Three, B₀
~ B_Three, C₀~ C_ThreeAre all 0, and the third and fourth stages
Then, the calculation result F4 (t) is obtained. Like these operations
The offspring are shown in FIG.

The normal final stage calculation result F (t) is F (t) = F1 (t) + F2 (t) + F3 (t) + F4
It is represented by (t). Here, in this embodiment, when a peak value equal to or larger than a predetermined value appears in the calculation result F (t), F
2 (t), F3 (t), and F4 (t) are shifted by ± samples and added, so that the peak value is less than the predetermined value. At this time, reference data is inserted in each of the second to fourth blocks so that the reception system can identify the number of samples shifted on the time axis.

In the receiver which receives the frequency division multiplexed signal thus generated, the correction value is determined according to the received value of the reference data of each block, and the data is decoded.

As an example, the calculation unit 1 advances F2 (t) by 2 sample times and F3 (t) by 3 sample times,
If it is determined that the peak value decreases when 4 (t) is delayed by one sample time, the calculation unit 1 moves the above sample on the time axis in order to make the peak value less than the predetermined value. Post-addition is executed to generate and output "data to be transmitted" shown on the right side of FIG. Thus, the transmission data described above is assumed to have been received as the data b _n 'shown on the left side of FIG. Here, it is assumed that reference data has been inserted into b ₂ , b ₁ and b ₃ .

In this case, since the first block has no advance or delay of samples, no correction is performed as it is. In the second block, as shown in FIG. 9, b ₂ '(= W ⁴ b ₂ ) is first received, but the transmission data b ₂ before shifting should be received, so b ₂ / b is used as the correction amount α. W ^-4 is obtained by calculating the ₂ '. Then, as described above from the relationship with the frequency of the reference data, each of the reception data of the second block is subjected to the correction of the following equation in order to correct the advance of two sample times.

B_Ten= (W^-20b_Ten’) = Α^(10/2)b_Ten’B₆= (W^-12b₆’) = Α^(6/2) b₆’B₁₄= (W^-28b₁₄’) = Α^(14/2)b₁₄'The third block is b, as shown in FIG.₁’(= W^Threeb₁)
Was received, but b₁Should be received, so the correction amount
b as β₁/ B₁’Calculate W^-3Is found. And
From the relationship with the frequency of the reference data, as described above, the third
For each received block data, advance 3 sample times
In order to correct the difference, the following formula is applied.

B₉= (W^-27b₉′) = Β^(9/91 b₉’B_Five= (W^-15b_Five′) = Β^(5/1) b_Five’B₁₃= (W^-39b₁₃′) = Β^(13/1) b₁₃'The fourth block is b, as shown in FIG._Three’(= W^-3b_Three)
Was received, but b_ThreeShould be received, so the correction amount
b as γ_Three/ B_Three’Calculate W^ThreeIs found. And
From the relationship with the frequency of the reference data, as described above,
One sample time delay for each received block data
In order to correct this, the following formula is corrected.

B ₁₁ = (W ¹¹ b ₁₁ ′) = γ ^(11/3) b ₁₁ ′ b ₇ = (W ⁷ b ₇ ′) = γ ^(7/3) b ₇ ′ b ₁₅ = (W ¹⁵ b ₁₅ ') = γ ^(15/3) b ₁₅ ' In the above explanation, it was explained that the peak value above the predetermined value is not moved so that the calculation result moves on the time axis. It is also possible to multiply by to change the amplitude. Although not described in detail, it is clear that the data can be restored by correcting the reference data.

(Third Embodiment) In this embodiment,
In addition to the method of shifting the phase on the time axis for the calculation result,
It also changes the amplitude.

In the addition method described in the first embodiment, as an example, the maximum peak point of F1 (t) is set to F2
It has been described that F2 (t) is shifted forward or backward by several samples so that the minimum peak points of (t) coincide. At this time, if the minimum peak point (absolute value) of F2 (t) is small, the amplitude of F2 (t) is increased. The correction method in this case can be explained as an extension of the contents described above.

That is, in the first embodiment, the correction amount of the fundamental frequency f ₁ is (b ₁ / b ₁ ′) = (x _S + jy _S ) / (x _S ′ + jy _s ′) as described above. ) Is required. When notation this in polar _{coordinates, (b 1 / b 1 '} ) = (S / S') (cos (θ S -θ S ') + jsin (θ S -θ
To _S ')) which, together with changing the phase in this embodiment, since the amplitude eg varied doubled transmits, 2S
= S ′, the correction amount of f ₁ is 2cos (θ _S −θ _S ′) + jsin (θ
_S −θ _S ').

On the other hand, transmission data D relating to transmission information_TXn
(= X + Jy) received data D obtained_TXn'=
The correction for (x '+ jy') is obtained from the following equation.

D _TXn = D _TXn '(S / S') (cos (n (θ _S −θ _S '))
+ Jsin (n (θ _S −θ _S ′))) The two-block division (first embodiment) and the four-block division (second embodiment) have been described above.
It is clear that division of more than blocks can be handled by the same kind of operation. Further, since the respective calculation results after these divisions have their respective regularities, it is possible to carry out the arithmetic utilizing the regularity instead of the normal IDFT arithmetic, and therefore, it is possible to expect a reduction in the arithmetic time.

[0074]

As described above, according to the present invention,
By performing the final stage twice, the peak value can be halved from the normal value, or the final stage X
Since the peak value can be reduced to 1 / X of the normal value by performing the stage operation X times from the previous stage, the generation of the peak value of the frequency-divided signal after addition can be suppressed significantly compared to the conventional case. In addition, it can be configured with an inexpensive electric system (adaptation of the dynamic range of the D / A converter and the A / D converter) without a drastic increase in the calculation time, and high reliability ( (Improvement of S / N) can be secured.

Further, according to the present invention, since the arithmetic processing and the decoding processing can be realized by the firmware of the transmitting device and the receiving device, it is possible to configure without adding any hardware, and to have a cost-effective configuration. Furthermore, according to the invention, the IDF
The regularity of the T operation can be used to shorten the operation time, and the correction process at the time of decoding can be easily performed by decoding the operation information and the reference data.

[Brief description of the drawings]

FIG. 1 is a block diagram of an embodiment of an apparatus to which a frequency division multiplex signal generation method of the present invention is applied.

FIG. 2 is a diagram illustrating an example of an IDFT calculation algorithm.

FIG. 3 is a diagram illustrating a calculation result of a fourth stage.

FIG. 4 is a diagram illustrating a calculation result of a third stage.

FIG. 5 is a diagram illustrating a calculation result of a third stage according to the first embodiment of this invention.

FIG. 6 is a diagram illustrating calculation results of first to third stages.

FIG. 7 is a diagram illustrating the calculation results of the first to third stages when the calculation results are shifted on the time axis.

FIG. 8 is an explanatory diagram of the operation result of each stage when the IDFT operation result is divided into four blocks.

FIG. 9 is a diagram illustrating a relationship between transmission data and reception data (before correction) that are shifted on a time axis.

FIG. 10 is a diagram illustrating an apparatus to which the decoding method of the present invention is applied.

[Explanation of symbols]

The latter part of the first operation unit 2 the output buffer 10 DFT calculation unit E ₀ to E ₇ third half portion H of the stage operation results ₀ to H ₇ third stage calculation result W ⁰ to ^W-7 twiddle factor A ₀ to A _3, B _{0 ~B 3, C 0 ~C 3} , D 0 ~D 3 second stage operation results

Claims (8)

[Claims] 1. An arithmetic unit for performing an inverse discrete Fourier transform of the digital information signals input to a plurality of input terminals in order to generate a frequency division multiplexed signal composed of a plurality of carrier waves each modulated by a digital information signal. In a frequency division multiplex signal generation method provided, the operation result of a stage immediately before the final stage of the operation unit is divided into a first half and a second half, the first half is left as it is, and the second half is set to zero. Then, the first final stage is performed to generate the first calculation result F1 (t), and the first half is set to zero, and the second half is left as it is to perform the second final stage. To generate a second calculation result F2 (t), and when the absolute value of the sum of the first and second calculation results is greater than or equal to a predetermined value, the second calculation result F2 (t) is timed with respect to the second calculation result F2 (t). On the axis After performing at least one of movement and multiplication of a predetermined value, the frequency division multiplexed signal generating method characterized by generating a frequency division multiplexed signal by adding synthetic first operation result F1 (t). 2. An arithmetic unit for performing an inverse discrete Fourier transform of the digital information signals input to a plurality of input terminals in order to generate a frequency division multiplexed signal composed of a plurality of carrier waves each modulated by a digital information signal. In the frequency division multiplex signal generation method provided, when the X stages of the operation rear part including the final stage of the operation part are performed, the operation result of the {(final stage) -X} th stage is a 2 ^X block. , And one of the divided blocks is used as it is, the remaining blocks are set to zero, and the X stages in the rear part of the operation are performed, and all the blocks are obtained. when the absolute value of the sum of 2 ^X number of operation result becomes a predetermined value or more, at least one of the multiplication of the movement and a predetermined value on the time axis with respect to 2 ^X number of operation result After applying, the frequency division multiplexed signal generating method, characterized in that to generate these 2 ^X number frequency division multiplexed signal by adding synthesizing operation result. 3. An input signal of a specific input terminal of the arithmetic unit is set to zero, a carrier wave assigned to the specific input terminal is used as a carrier hole, and is transmitted through the carrier hole in an output frequency division signal of the arithmetic circuit. The calculation information that has been moved on the time axis and / or multiplied by a predetermined value is added and combined as a signal to be added.
Alternatively, the frequency division multiplex signal generation method described in 2). 4. A specific carrier wave in which at least one reference data is inserted is set as the operation information obtained by performing movement on the time axis and / or multiplying by a predetermined value to the initial input data of the latter half part. The frequency division multiplex signal generation method according to claim 1, wherein 5. A specification in which at least one reference data is inserted as operation information obtained by performing movement on the time axis and / or multiplying by a predetermined value to the initial input data of each of the 2 ^X blocks. The frequency division multiplex signal generation method according to claim 2, wherein a carrier wave is set. 6. A decoding method for decoding the digital information signal by performing a discrete Fourier transform on a frequency division multiplexed signal composed of a plurality of carrier waves each modulated with a digital information signal, wherein the frequency division multiplexed signal according to claim 3. A decoding method, characterized in that the operation information transmitted through a predetermined carrier hole is decoded, and the decoded digital information signal is corrected by the decoded operation information. 7. A decoding method for decoding the digital information signal by performing a discrete Fourier transform on the frequency division multiplexed signal composed of a plurality of carrier waves each modulated by the digital information signal, wherein the frequency division multiplexed signal according to claim 4. A decoding method, comprising: decoding the reference data transmitted by the specific carrier, and correcting the decoded digital information signal with the decoded reference data. 8. A decoding method for decoding a digital information signal by performing a discrete Fourier transform on a frequency division multiplexed signal composed of a plurality of carrier waves each modulated by a digital information signal, wherein the frequency division multiplexed signal according to claim 5. The decoding method, comprising: decoding the reference data transmitted by the specific carrier, and correcting the calculation result with the decoded reference data for each of the decoded 2 ^X blocks.