JP2001119364A - Orthogonal frequency division multiplex signal transmitting system, transmitter and receiver - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problems that there is a fixed limit in much more accuracy improvement since the amplitude characteristic difference or phase characteristic difference of I signal and Q signal becomes the direct cause of code error and that a characteristic change can not be corrected while a reference signal is not transmitted. SOLUTION: On the transmitting side, the known reference signal is transmitted by a specified carrier wave among plural carrier waves composing an OFDM signal, the carrier wave for transmitting the known reference signal is designated by a symbol number to be transmitted by a prescribed carrier wave, and the reference signal is transmitted while being successively and cyclically changed at the interval of a fixed period. On the receiving side, the reference signal is demodulated by a demodulation circuit 311 and transmission line characteristics are detected by a detecting circuit 312. A first correction expression is calculated from the detected transmission line characteristics and stored by a first correction expression deriving and holding circuit 313. The signal is decoded by a first correcting circuit 314 while using this correction expression. Further, the output signal of this correcting circuit 314 is corrected by newly calculating a second correction expression so that high-speed change characteristics can be corrected.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an orthogonal frequency division multiplexing signal transmission system, a transmitting apparatus and a receiving apparatus, and more particularly to an orthogonal frequency division multiplexing (OFDM) for a coded digital video signal in a limited frequency band. Frequen
The present invention relates to an orthogonal frequency division multiplex signal transmission system that converts a signal into a cy division multiplex signal and transmits / receives the signal, and relates to a transmission device and a reception device.

[0002]

2. Description of the Related Art One method of transmitting coded digital video signals in a limited frequency band is as follows.
6 Quadrature Amplitude Modul (QAM)
The OFDM system, which transmits multilevel modulated digital information as OFDM signals using a large number of carriers, has been known from the past because of its features such as being resistant to multipath, less susceptible to interference, and having relatively good frequency use efficiency. Have been. The OFDM system is a system in which a large number of carriers are arranged orthogonally and independent digital information is transmitted on each carrier. Note that "carriers are orthogonal" means that the spectrum of an adjacent carrier becomes zero at the frequency position of the carrier.

[0003] In a transmitter for transmitting this OFDM wave,
A plurality of digital baseband in-phase signals (I signals) and quadrature signals (Q signals) obtained by performing an inverse discrete Fourier transform (IDFT) operation on data to be transmitted are converted into a D / A converter,
After each passing through a low-pass filter (LPF), the signal is converted and synthesized into an intermediate frequency (IF) by a quadrature modulator, and then converted into an RF signal band by a frequency converter, and an unnecessary frequency is converted by a band-pass filter (BPF). The component is removed, the power is amplified by the transmission unit, and the radio wave is emitted from the transmission antenna.

On the other hand, in the receiving apparatus, the frequency converter converts the frequency of the received RF signal to an intermediate frequency, amplifies the intermediate frequency with an amplifier, removes unnecessary frequency components with a BPF, and then outputs the I / O signal with a quadrature demodulator. The signal and the Q signal are separated. After that, the separated I signal and Q signal are converted to LPF, A / D
Transformer and discrete Fourier transform (DFT) circuit, QAM
The data is restored by performing QAM decoding or the like through the respective decoding circuits.

In these OFDM signal transmission processes,
At the time when both the transmitting device and the receiving device are separated into an I signal and a Q signal, if there is a relative amplitude characteristic difference and a phase characteristic difference between the systems, or a quadrature modulator or a quadrature demodulator If a modulated or demodulated wave having an accurate 90-degree phase difference is not supplied, a code error will occur. That is, an amplitude change and a phase change of its own frequency and a crosstalk (image component) to the symmetrical opposite frequency occur.

Therefore, conventionally, in order to prevent the occurrence of the above-mentioned code error, an amplitude characteristic difference and a phase characteristic difference at the time of separation into an I signal and a Q signal, or a quadrature modulator,
A characteristic difference (error) between the I signal and the Q signal that a quadrature demodulator is not supplied with an accurate 90-degree phase difference modulation / demodulation wave.
Various methods have been proposed for eliminating (for example, JP-A-6-350658, JP-A-3-76623, JP-A-5-227239, JP-A-5-110369,
JP-A-3-53735, JP-A-6-188932,
JP-A-4-290337).

[0007] Further, corrections for setting the frequency amplitude characteristic and the frequency phase characteristic of the I signal and the Q signal itself to specified characteristics have been conventionally proposed.
Signal, Q signal frequency amplitude characteristics and frequency phase characteristics, quadrature modulated wave phase, IF wave band signal processing frequency amplitude characteristics and frequency phase characteristics, RF wave band signal processing frequency amplitude characteristics and frequency phase characteristics, and reception The frequency amplitude characteristics and frequency phase characteristics of I and Q signals, the phase of quadrature demodulated waves, the frequency amplitude characteristics and frequency phase characteristics of IF band signal processing, and the frequency amplitude characteristics and frequency phase characteristics of RF wave band signal processing Further, it has been proposed to correct one or more of various characteristics such as a frequency amplitude characteristic and a frequency phase characteristic in a multipath environment of radio wave propagation (JP-A-6-311134, JP-A-6-311134). No. 5-219021).

[0008]

However, in each of these conventional correction methods, the cost is increased due to the complexity of the circuit configuration, and the accuracy is also limited. Compensating for all errors is even more complicated. Some of them are not suitable for OFDM signals. Further, a conventional method of compensating for a deviation in orthogonality by a transceiver (Japanese Patent Laid-Open No. 6-188)
No. 932) does not compensate for the deviation between the amplitudes of the I signal and the Q signal, or arranges a digital quadrature modulator in the transmitter in front of the D / A converter to reduce errors in the transmission system in principle. A conventional method (Japanese Unexamined Patent Publication No. 4-2903)
No. 37) has a problem in that usable devices are limited due to restrictions on the operation speed and bit width of the D / A converter.

As described above, conventionally, the amplitude characteristic difference, the phase characteristic difference, and the orthogonality error of the I signal and the Q signal directly cause a code error of the transmission / reception signal, and these errors are removed or compensated. However, there is a certain limit to further improving the accuracy. Furthermore, OFD such as compensation of frequency characteristics
Conventionally, there has not been any consideration up to the correspondence to the M signal.

[0010] The present invention has been made in view of the above points,
An object of the present invention is to provide an orthogonal frequency division multiplexed signal transmission system, a transmission device, and a reception device that can prevent occurrence of a code error without complicating a circuit configuration by allowing an error and correcting the error. .

Further, another object of the present invention is to provide, in addition to the error,
Orthogonal frequency division multiplexing signal transmission method suitable for OFDM signal transmission by correcting characteristics by frequency,
A transmitting device and a receiving device are provided.

[0012]

In order to achieve the above object, in the orthogonal frequency division multiplexing signal transmission system of the present invention, a plurality of carriers having different frequencies are assigned to each carrier on the transmitting side. The in-phase signal and the quadrature signal obtained from the information signal to be transmitted are separately modulated, and the frequency-division multiplexed orthogonal frequency division multiplex signal is generated and transmitted in symbol units. In the quadrature frequency division multiplexing signal transmission system for receiving and decoding each modulated carrier into an in-phase signal and a quadrature signal, and then demodulating the information signal, on the transmission side,
A known reference signal is transmitted with a set of a positive carrier on the high frequency side and a negative carrier on the low frequency side symmetrical with respect to the center carrier of the plurality of carriers, and the positive and negative carriers transmitting the known reference signal are transmitted. A set is designated by a specific carrier or a symbol number to be transmitted by a transmission parameter, and is sequentially and cyclically changed at regular intervals and transmitted.The receiving side refers to the received frequency division multiplexed signal based on the symbol number. The signal is decoded, and from this reference signal, the channel characteristics are detected based on the leak components to the real part and the imaginary part of the positive carrier and the real part and the imaginary part of the negative carrier, respectively. , A correction formula is calculated and stored, and the in-phase signal and the quadrature signal decoded using the correction formula are corrected.

Further, in the frequency division multiplex signal transmission system of the present invention, the first set of reference signals and the second set of reference signals transmitted by a set of positive and negative carrier waves by alternately switching the reference signals in predetermined symbol units. The values of the real and imaginary parts of the reference signal transmitted on the positive carrier are p and q, respectively.
When the values of the real part and the imaginary part of the reference signal transmitted on the negative carrier are r and u, respectively, the values of the real part and the imaginary part of the received reference signal of the positive carrier are p ′ and q ′, respectively. When the values of the real part and the imaginary part of the received reference signal of the negative carrier are r ′ and u ′, respectively, the receiving side obtains the following equation as a transmission path characteristic.

[0014]

(Equation 8) The coefficients S0 to S7 expressed by the following formulas are calculated from the known reference signal for the received reference signal.

[0015]

(Equation 9) (However, H0 = + S0 (S6S6 + S7S7) -S2 (S4S6 + S5S7) + S3 (S4S7-S
5S6), H1 = + S1 (S6S6 + S7S7) -S3 (S4S6 + S5S7) -S2 (S4S7-S5S
6), H2 = + S4 (S2S2 + S3S3) -S6 (S0S2 + S1S3) + S7 (S0S3-S1S2), H
3 = + S5 (S2S2 + S3S3) -S7 (S0S2 + S1S3) -S6 (S0S3-S1S2), H4 = + S
2 (S4S4 + S5S5) -S0 (S4S6 + S5S7) -S1 (S4S7-S5S6), H5 = + S3 (S4
S4 + S5S5) -S1 (S4S6 + S5S7) + S0 (S4S7-S5S6), H6 = + S6 (S0S0 + S
1S1) -S4 (S0S2 + S1S3) -S5 (S0S3-S1S2), H7 = + S7 (S0S0 + S1S1)
-S5 (S0S2 + S1S3) + S4 (S0S3-S1S2), det A = S0 × H0 + S1 × H1
+ S4 × H2 + S5 × H3) The value calculated by is stored and held, and the following formula is used for each set of positive and negative carriers using the correction formula.

[0016]

(Equation 10) (Where a and b are the real and imaginary parts of the corrected received data assigned to the positive carrier frequency,
c and d are the real part and the imaginary part of the corrected received data assigned to the negative carrier frequency, a ′ and b ′ are the real part and the imaginary part of the received data of the positive carrier frequency, c ′
And d 'are the real part and the imaginary part of the received data of the negative carrier frequency) and the corrected received data a, b, c
And d.

Here, the reference signal is composed of a first set of reference signals and a second set of reference signals, which are alternately switched in a predetermined symbol unit and transmitted as a set of positive and negative carrier waves. Of the values of the real part and the imaginary part transmitted on the positive carrier and the real part and the imaginary part transmitted on the negative carrier,
The first set of reference signals has only the real or imaginary part values transmitted on the positive carrier as predetermined values, and the other values are each set to zero. The second set of reference signals is transmitted on the negative carrier. It is preferable that only the value of the real part or the imaginary part is set to a predetermined value and the other values are set to zero, respectively, because calculation becomes easy in calculating the coefficient of the transmission path characteristic.

Similarly, the reference signal is composed of a first set of reference signals and a second set of reference signals, which are alternately switched in a predetermined symbol unit and transmitted as a set of positive and negative carrier waves, and a set of positive and negative carrier waves. Of the values of the real part and the imaginary part transmitted on the positive carrier and the real part and the imaginary part transmitted on the negative carrier,
The first set of reference signals has the values of the real part and the imaginary part transmitted on the positive carrier as predetermined values, and the other values are each set to zero. The second set of reference signals is transmitted on the negative carrier. Even if the values of the real part and the imaginary part are set to predetermined values and the other values are set to zero, the calculation in the calculation of the coefficients of the transmission path characteristics becomes easy.

In order to achieve the above object, an orthogonal frequency division multiplex signal transmitting apparatus according to the present invention is arranged such that a digital information signal to be transmitted represented by a complex number is input to a plurality of real part input terminals and a plurality of imaginary part input terminals, respectively. An inverse discrete Fourier transform to generate an in-phase signal and a quadrature signal transmitted on a plurality of carriers having different frequencies in symbol units, and a value that changes for each symbol of the digital information signal to be transmitted A symbol number counting circuit for generating a symbol number and inputting it to a specific input terminal of the arithmetic unit, and a digital information signal transmitted on a positive carrier on a high frequency side symmetrical with respect to a center carrier among a plurality of carriers. The real part input terminal and the imaginary part input terminal of the operation part to be processed, and the real part input of the operation part to which the digital information signal transmitted by the low-side negative carrier wave are input A reference signal insertion means for inputting a reference signal to a set of a child part and an imaginary part input terminal, and switching a set of a real part input terminal and an imaginary part input terminal for inputting the reference signal at regular time intervals; An output buffer for temporarily storing the phase signal and the quadrature signal; a conversion unit for converting the output in-phase signal and the quadrature signal of the output buffer into quadrature frequency division multiplex signals; and a transmission unit for transmitting the quadrature frequency division multiplex signal. It is configured.

Further, to achieve the above object, an orthogonal frequency division multiplex signal receiving apparatus of the present invention transmits a digital information signal to be transmitted on a plurality of carriers having different frequencies from each other, and transmits a digital information signal to a center carrier among the plurality of carriers. A known reference signal is inserted into a pair of a positive carrier on the high frequency side and a negative carrier on the low frequency side, which are symmetrical to each other, and a set of positive and negative carriers transmitting the reference signal is transmitted through a specific carrier. And receiving means for receiving orthogonal frequency division multiplexed signals sequentially and cyclically changed at regular intervals, and orthogonally demodulating the received orthogonal frequency division multiplexing signals from the receiving means, and expressing them in complex numbers, respectively. Demodulation means for obtaining the in-phase signal and the quadrature signal, the symbol number, and the reference signal, and the discrete-mode decoding of the in-phase signal, the quadrature signal, the symbol number, and the reference signal from the demodulation means. Decoding means for decoding a digital information signal by performing a Rier transform and decoding a symbol number and a reference signal; decoding a reference signal from the symbol number from the decoding means; and real and imaginary numbers of a positive carrier wave from the reference signal. , Detecting means for detecting transmission path characteristics based on leakage components to the real part and imaginary part of the negative carrier, and a correction equation for calculating and storing a correction equation from the transmission path characteristics detected by the detecting means. It comprises a calculating and holding means, and a correction circuit for correcting the in-phase signal and the quadrature signal decoded signal from the decoding means using the stored correction formula.

The generation of the in-phase signal (I signal) and quadrature signal (Q signal) errors generated in the orthogonal frequency division multiplexing signal transmission system causes a difference in amplitude characteristic, phase characteristic difference between the I signal and the Q signal in the transmitting / receiving apparatus. This is due to orthogonality errors. No error occurs after the I signal and the Q signal are combined by the transmitting device and until the I signal and the Q signal are separated by the receiving device. Further, as described in the section of the prior art, the correction for eliminating the characteristic difference (error) between the I signal and the Q signal and the correction for setting the characteristic of the I signal Q signal itself to a specified value are constant. And sufficient correction cannot be performed.

In view of the above points, the present invention differs from the conventional frequency division multiplexing signal transmission system and the transmitting apparatus and the receiving apparatus which perform correction for "do not generate an error". The correction is made. That is, in the present invention, a reference signal is inserted into a frequency division multiplexed signal as known reference data on the transmission side and transmitted, and a transmission path indicating an error between the I signal and the Q signal based on the reference signal on the reception side. The characteristic is detected, and a correction equation is obtained to correct the detected transmission path characteristic.

Here, the transmission line characteristic may be detected by transmitting and receiving four reference signals with 16 transmission line characteristic coefficients. However, the following speed for the ever-changing transmission line characteristic becomes slow. In the present invention, there is a crosstalk effect on the center carrier of the plurality of carriers,
A pair of one positive carrier and one negative carrier, which are symmetrical to each other, are transmitted as a pair with these carriers. Further, according to the present invention, the positive and negative carriers into which the reference signal is inserted are designated by symbol numbers, and the set of the positive and negative carriers is switched at regular intervals so that the transmission path characteristics of all the carriers can be detected. It was done.

Thus, according to the present invention, the above-mentioned frequency characteristics of the I signal and the Q signal itself can be corrected in the same manner as in the prior art, and in addition to this, the difference between the I signal and the Q signal at each frequency of the transmitting device can be obtained. Relative amplitude characteristic difference, relative phase characteristic difference between I signal and Q signal at each frequency of the transmitting device,
The orthogonality error of the quadrature modulator of the transmitting device, the relative amplitude characteristic difference between the I signal and the Q signal at each frequency of the receiving device, and the relative phase characteristic difference between the I signal and the Q signal at each frequency of the receiving device It is also possible to correct characteristic errors such as the orthogonality error of the quadrature demodulator of the receiving device.

Further, according to the present invention, on the transmitting side, an information signal to be transmitted is separately modulated as a signal having a real part and an imaginary part to generate an in-phase signal and a quadrature signal. And a plurality of carrier waves having different frequencies from each other with orthogonal signals, and modulates a plurality of positive carrier waves on the high band side and a plurality of negative carrier waves on the low band side with respect to the center carrier, and performs frequency division multiplexing orthogonal frequency division. Generates and transmits a multiplex signal, and on the receiving side, receives an orthogonal frequency division multiplex signal, demodulates each modulated carrier into an in-phase signal and an orthogonal signal, and then decodes an information signal to transmit an orthogonal frequency division multiplex signal. In the system, the transmitting side transmits a known reference signal on a predetermined carrier of the plurality of carriers, and the predetermined carrier is sequentially and cyclically changed at regular intervals, and the receiving side receives the signal. A reference signal is decoded from the orthogonal frequency division multiplexed signal, a transmission path characteristic is detected from the reference signal based on a change component of a real part and an imaginary part of a predetermined carrier, and a first correction is performed based on the detected transmission path characteristic. Calculate and store the formula,
The decoded information signal is corrected using the first correction formula, and a high-speed change component of the transmission path is detected based on a difference between the corrected information signal and a predetermined signal point arrangement. And calculating and storing a second correction formula from the high-speed change component, and further correcting the corrected information signal using the second correction formula.

Thus, in the present invention, the first correction formula is used to determine the relative amplitude / phase characteristic difference between the I and Q signals and the orthogonality error of the quadrature demodulator of the receiving device at each frequency of the receiving device. , The IF of the transmitting device, the frequency amplitude / phase characteristics of RF waveband signal processing, the RF and IF of the receiving device
Frequency and amplitude characteristics of waveband signal processing, I of the receiver,
The frequency characteristics such as the frequency amplitude / phase characteristics of each Q signal, the frequency amplitude / phase characteristics under slowly changing radio wave propagation, and the phase of the quadrature modulation / demodulation wave of the transceiver can be corrected. In addition, according to the present invention, the frequency amplitude / phase characteristics under a rapidly changing radio wave propagation can be corrected by the correction using the second correction formula.

[0027]

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

FIG. 1 is a block diagram of a first embodiment of an orthogonal frequency division multiplex signal transmitting apparatus which is a transmitting apparatus of the transmission system according to the present invention. In FIG. 1, digital data to be transmitted is input to an input terminal 1. The digital data is, for example, M which is a color moving image coding system.
These are digital video signals and audio signals compressed by an encoding method such as the PEG method. The input digital data is supplied to an input circuit 2 and an error correction code is added as necessary based on a clock from a clock frequency divider 3. The clock divider 3 divides an intermediate frequency of 10.7 MHz from the intermediate frequency oscillator 10 and generates a clock synchronized with the intermediate frequency.

The digital data to which the error correction code has been added is supplied from the input circuit 2 to the arithmetic unit 4. The operation unit 4 is a circuit constituting a main part of the present embodiment, and converts digital data from the input circuit 2 into an inverse discrete Fourier transform (IDFT).
The in-phase signal (I signal) and the quadrature signal (Q signal) are generated by calculation, and the symbol number from the symbol number counting circuit 5 and the reference signal from the reference signal insertion circuit 6 are also respectively inputted to predetermined input as described later. IDFT
Calculate.

As an example, the arithmetic unit 4 has a data sequence N of 25.
When transmitted on six carriers, a signal is generated by performing IDFT operation of double oversampling. At this time, the input assignment to the arithmetic unit 4 is as follows when the numbers are sequentially assigned in the input frequency alignment type.

An information signal that modulates n = 0 to 128 carriers is provided.

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

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

That is, the number of input terminals of the arithmetic unit 4 is 512 for the real part (R) signal and 512 for the imaginary part (I) signal, and the first (n = 1) ) To the 127th (n = 127)
An information signal is input to a total of 127 input terminals of each of the 27th and 385th (n = 385) to 511th (n = 511) input terminals, and a DC signal is input to the 0th (n = 0) input terminal. Voltage (constant) is input and the 128th (n = M /
For example, a fixed voltage for a pilot signal is input to the 4th and 384th (n = 3M / 4) input terminals.

The operation unit 4 thus determines from the first to 127
A 4-bit R signal and a 4-bit I signal are input to the input terminal and the 385th to 511th input terminals, respectively, and a constant voltage is input to the 0th, 128th and 384th input terminals. After that, 0 is input to the 129th to 383rd input terminals, and a double oversampling IDFT operation is performed. As a result, an in-phase signal (I signal) and a quadrature signal (Q signal) are obtained. I
After inserting a guard interval for reducing multipath distortion into the signal and the Q signal, the signal and the Q signal are output to the output buffer 7.

Here, a total of 1 from the 1st to the 128th
The input information of the 28 input terminals is a carrier for information transmission on the upper side (higher frequency side) with respect to the center carrier frequency F0 transmitting the input information of the 0th input terminal (this is referred to as a positive carrier or a positive carrier in this specification). Carrier).
The input information of a total of 128 input terminals from the 384th to the 511th is a carrier for information transmission lower (lower side) with respect to the center carrier frequency F0 (this carrier is referred to as a negative carrier or carrier in this specification). ), Especially 1
As a result of the IDFT operation, the input pilot signals of the 28th and 384th input terminals are transmitted on carrier waves at both ends of the frequency, which is equivalent to half the Nyquist frequency,
0 is input to the remaining 129th to 383th input terminals (the ground potential), so that a carrier wave of that portion is not generated (not used for data transmission).

In the output buffer 7, the output operation result of the operation unit 4 is generated in a burst manner as 256 input information as 512 time axis signals (I signal and Q signal) in one IDFT operation. On the other hand, since the circuits subsequent to the output buffer 7 operate continuously at a constant reading speed of the content of the output buffer 7, they are provided to adjust the time difference between the two.

Based on the clock from the clock divider 3 in FIG. 1, I
The I and Q signals resulting from the DFT operation are supplied to a D / A converter / low-pass filter (LPF) 8, where they are converted into analog signals using the clock from the clock divider 3 as a sampling clock. , LPF, the I signal and the Q signal of the components in the required frequency band are passed and supplied to the quadrature modulator 9, respectively.

The quadrature modulator 9 uses the 10.7 MHz intermediate frequency from the intermediate frequency oscillator 10 as a first carrier wave, and sets the phase of this intermediate frequency to 90 ° by the 90 ° shifter 11.
A 10.7 MHz intermediate frequency shifted by 0 ° is used as a second carrier, and 257 waves (positive / negative 128 bits) are subjected to quadrature amplitude modulation (QAM) with the I and Q signals of the digital data input from the D / A converter and LPF 8, respectively. An OFDM signal composed of a set of carriers and one center carrier is generated. The OFDM signal output from the quadrature modulator 9 is frequency-converted by the frequency converter 12 into an RF signal in a predetermined transmission frequency band, and then subjected to transmission processing such as power amplification in the transmission unit 13 and radiated from an antenna (not shown). You.

FIG. 2 shows an example of the frequency spectrum of the OFDM signal. Here, "-256" and "+25"
Reference numeral 6 "denotes a pair of positive and negative carrier frequencies through which a signal (not used in this embodiment) input to the 256th (n = M / 2) input terminal of the arithmetic unit 4 is transmitted.
28 "is a negative carrier frequency at which the signal input to the {(3M / 4)} th input terminal is transmitted, and" +128 "is the negative carrier frequency at which the signal input to the {(M / 4)} th input terminal is transmitted. At the transmitted positive carrier frequency, these make up a set of positive and negative carriers.

The transmitted OFDM signal is received by, for example, a frequency division multiplexed signal receiving apparatus having a configuration as shown in FIG. This frequency division multiplexed signal receiving apparatus is characterized by the configuration of the FFT / QAM decoding circuit 31, as described later. In FIG. 3, O input through a spatial transmission path
The FDM signal is received by a receiving unit 21 via a receiving antenna, is then subjected to high-frequency amplification, and is further frequency-converted to an intermediate frequency by a frequency converter 22.
After that, the signal is supplied to a carrier extraction and quadrature demodulator 44 having a configuration described later.

The carrier extraction circuit portion of the carrier extraction and quadrature demodulator 24 is a circuit for extracting the center carrier (carrier) of the input OFDM signal as accurately as possible with a small phase error. Here, each carrier for transmitting information is placed adjacent to every 387 Hz that is a symbol frequency and OFD
Since the M signal is formed, the carrier for information transmission adjacent to the center carrier is also separated by 387 Hz from the center frequency, and in order to extract the center carrier, the carrier for information transmission adjacent to the carrier only 387 Hz is separated. A circuit with high selectivity is required so as not to be affected.

Therefore, the center carrier F0 is extracted by using a PLL circuit in the carrier extraction circuit. However, as the VCO constituting the PLL circuit in this case, a voltage-controlled crystal oscillation circuit (VCXO) using a crystal oscillator that oscillates at about ± 200 Hz whose variable range is about １ ／ of the adjacent carrier frequency is used. L used and constituting a PLL circuit
As the PF, an LPF having a sufficiently low cutoff frequency with respect to 387 Hz is used.

The center carrier F0 extracted by the carrier extraction and quadrature demodulator 24 is supplied to the intermediate frequency oscillator 25, where it is 10.7 phase-locked to the center carrier F0.
Generate an intermediate frequency of MHz. Intermediate frequency oscillator 2
5 is directly supplied to the quadrature demodulator 24 as the first demodulation carrier, while the phase is shifted by 90 ° by the 90 ° shifter 26, and then the carrier extraction and quadrature demodulation are performed as the second demodulation carrier. Is supplied to the vessel 24.

As a result, an analog signal (frequency division multiplexed signal) equivalent to the analog signal input to the quadrature modulator 9 of the transmitting apparatus is demodulated and taken out from the quadrature demodulator section of the carrier extraction and quadrature demodulator 24. , A low-pass filter (LPF) 28
, A signal of a required frequency band transmitted as OFDM signal information is passed, supplied to an A / D converter 29, and converted into a digital signal.

What is important here is the sampling timing for the input signal of the A / D converter 29, which is based on a sample synchronization signal having a frequency twice the Nyquist frequency, which is generated from the pilot signal by the synchronization signal generation circuit 27. Generated. That is, the pilot signal is set at a predetermined integer ratio with respect to the sample clock frequency, and the frequency of the pilot signal is multiplied according to the frequency ratio to obtain the timing of the sample clock.

The synchronization signal generation circuit 27 receives a demodulated analog signal and generates a sample synchronization signal by a PLL circuit which performs phase synchronization with a pilot signal transmitted as a continuous signal in each symbol period including a guard interval period. A generating circuit section, a symbol synchronization signal generating circuit section for examining a phase state of a pilot signal based on a signal extracted from a part of the sample synchronization signal generation circuit section, detecting a symbol period, and generating a symbol synchronization signal; A system clock generating circuit for generating a system clock such as a section signal for removing a guard interval period from the signal and the symbol synchronization signal.

The digital signal extracted from the A / D converter 29 is supplied to the guard interval period processing circuit 30, where the digital signal is less affected by multipath distortion based on the system clock from the synchronization signal generation circuit 27. Is obtained and supplied to the FFT / QAM decoding circuit 31.

The FFT (Fast Fourier Transform) circuit section of the FFT / QAM decoding circuit 31 performs a complex Fourier operation using the system clock from the synchronization signal generation circuit 27, and outputs the output signal of the guard interval period processing circuit 30 for each frequency. The signal levels of the real part and the imaginary part are calculated.

The real part of each frequency obtained by this,
Each signal level of the imaginary part is compared with the demodulated output of the reference carrier by the QAM decoding circuit section, whereby the level of the quantized digital signal transmitted by the carrier for digital information transmission is obtained. Decrypted. The decoded digital information signal is subjected to output processing such as parallel-serial conversion by the output circuit 32 and is output to the output terminal 33.

The frequency division multiplex signal transmission system including the frequency division multiplex signal transmitting apparatus shown in FIG. 1 and the frequency division multiplex signal receiving apparatus shown in FIG. Can be rewritten as shown in

In FIG. 4, the signal processing units 41 and 42 are respectively composed of the D / A converter / LPF 8 shown in FIG.
Is a circuit part composed of an amplifier and the like not shown.
The multipliers 43 and 44 and the adder 45 are circuit parts constituting the quadrature modulator 9 shown in FIG. 1, and the transmission system circuit 46 includes the frequency converter 12, the transmission unit 13, and the spatial transmission line shown in FIG. The characteristic is a transmission system component corresponding to a transmission line including the reception unit 21, the bandpass filter (not shown), the frequency converter 22, the intermediate frequency amplifier 23, and the like shown in FIG. 3 is a circuit part of the carrier extraction and quadrature demodulator 24 shown in FIG. 3, and the signal processing units 49 and 50 correspond to the circuit part including the LPF 28 and the A / D converter 29 shown in FIG.

Here, the complex number (p + jq) assigned to a certain carrier frequency + Wn and the complex number (r + ju) assigned to a negative carrier frequency −Wn symmetrical to the carrier frequency + Wn with respect to the center frequency F0 are calculated by the above-mentioned calculation. The signals are generated as I and Q signals of the time axis waveform shown in the following equations by the IDFT calculation by the unit 4.

I signal = Acos (Wnt + a) (1) Q signal = Asin (Wnt + a) (2) I signal = Bcos (−Wnt + b) (3) Q signal = Bsin (−Wnt + b) (4) where A = √ (P2 + q2), a = tan-1 (q / p), B = √ (r2 + u2), b = tan-1 (u / r).

Equations (1) and (2) are respectively given by the frequency + W
Equations (3) and (4) are the I and Q signals of frequency -Wn, respectively. When the I signals represented by the equations (1) and (3) are input to the signal processing unit 41 of FIG. 4, respectively, they receive the amplitude change X1 and the phase change c1, and receive the equations (5) and (7). And is taken out as an I signal represented by On the other hand, when the Q signals represented by the equations (2) and (4) are input to the signal processing unit 42 in FIG. 4, respectively, they undergo the amplitude change X2 and the phase change c2 (6).
The signal is extracted as the Q signal represented by the expression (8).

I signal = X1Acos (+ Wnt + a + c1) (5) Q signal = X2Asin (+ Wnt + a + c2) (6) I signal = X1Bcos (−Wnt + b + c1) (7) Q signal = X2Bsin (−Wnt + b + c2) (8) The Q signal is multiplied by multipliers 43 and 44 constituting the quadrature modulator 9 to obtain an I signal and a Q signal represented by the following equations. Here, the error of the orthogonality is c4, and the modulated wave is X3 cos (W0t + c3 + c
4) For the Q signal, it is -X4 sin (W0t + c3).
As a result, the multiplier 43 outputs an I signal generated from + Wn and an I signal generated from -Wn in the following equation.

I signal = X1Acos (Wnt + a + c1) × (+ X3cos (W0t + c3 + c4)) = (1/2) X1X3A [cos ((W0 + Wn) t + a + c1 + c3 + c4 ) + cos ((W0-Wn) ta-c1 + c3 + c4)] I signal = X1Bcos (-Wnt + b + c1) × (X3cos (W0t + c3 + c4)) = (1/2) X1X3B [cos ((W0-Wn) t + b + c1 + c3 + c4) + cos ((W0 + Wn) tb-c1 + c3 + c4)]. Similarly, the multiplier 44 outputs + W expressed by the following equation.
A Q signal generated from n and a Q signal generated from -Wn are output.

Q signal = X2Asin (Wnt + a + c2) × (−X4sin (W0t + c3)) = (1/2) X2X4A [cos ((W0 + Wn) t + a + c2 + c3) -cos ( (W0-Wn) ta-c2 + c3)] Q signal = X2Bsin (-Wnt + b + c2) × (-X4sin (W0t + c3)) = (1/2) X2X4B [cos ((W0-Wn) t + b + c2 + c3) -cos ((W0 + Wn) tb-c2 + c3)] The I signal and the Q signal are combined by the adder 45. This synthesized signal is referred to as (W0 + Wn) component and (W0
-Wn) can be expressed by the following equation. However, when there is an amplitude change and a phase change, the amplitude change is expressed by X in the I signal.
1 (or X3), the phase change is included in X2 (or X4) for the Q signal, the phase change is included in c3 + c4 for the I signal, and c3 in the Q signal. (W0 + Wn) component: (1/2) [X1X3 [Acos ((W0 + Wn) t + a + c1 + c3 + c4)] + Bcos ((W0 + Wn) tb-c1 + c3 + c4)] + X2X4 [Ac os ((W0 + Wn) t + a + c2 + c3) -Bcos ((W0 + Wn) tb-c2 + c3)]] (9) (W0-Wn) component: (1/2) [X1X3 [Bcos ((W0-Wn) t + b + c1 + c3 + c4)] + Acos ((W0-Wn) ta-c1 + c3 + c4)] + X2X4 [Bcos ((W0-Wn) t + b + c2 + c3)]-Acos ((W0-Wn) ta-c2 + c3)] (10) Next, by transmitting the I signal and the Q signal through the transmission circuit 46, the influence under the multipath environment is obtained. And is affected by the transmission system. Assuming that the characteristic changes of the frequencies corresponding to the frequencies + Wn and -Wn during this period are Y1 on the + Wn side and Y2 on the -Wn side, and the phase change is d1 on the + Wn side and d2 on the -Wn side. The (W0 + Wn) component and the (W0-Wn) component are respectively represented by the following equations. (W0 + Wn) component: (1/2) [X1X3Y1 [Acos ((W0 + Wn) t + a + c1 + c3 + c4 + d1)] + Bcos ((W0 + Wn) tb-c1 + c3 + c4 + d1) ] + X2X4Y1 [Acos ((W0 + Wn) t + a + c2 + c3 + d1) -Bcos ((W0 + Wn) tb-c2 + c3 + d1)]] (11) (W0−Wn) component: ( 1/2) [X1X3Y2 [Bcos ((W0-Wn) t + b + c1 + c3 + c4 + d2)] + Acos ((W0-Wn) ta-c1 + c3 + c4 + d2)] + X2X4Y2 [Bcos ( (W0-Wn) t + b + c2 + c3 + d2)]-Acos ((W0-Wn) ta-c2 + c3 + d2)] (12) Next, the I signal and the Q signal are Is subjected to quadrature demodulation. The orthogonality error at this time is g4, and the carrier for demodulation is 2Z3 cos (W0t + g3 + g) for the I signal.
4) If the Q signal is -2Z4 sin (W0t + g3), the + Wn component generated from the (W0 + Wn) component is (1)
Equations (3) and (14) are used (however, harmonic components are omitted).

I signal = (11) × (+ 2Z3cos (W0t + g3 + g4)) = (1/2) × [X1X3Y1Z3 [Acos (Wnt + a + c1 + c3 + c4 + d1-g3-g4) + Bcos (Wnt-b-c1 + c3 + c4 + d1-g3-g4)] + X2X4Y1Z3 [Acos (Wnt + a + c2 + c3 + d1-g3-g4) -Bcos (Wnt-b-c2 + c3 + d1-g3-g4)]] (13) Q signal = (11) formula x (-2Z4sin (W0t + g3)) = (1/2) x [X1X3Y1Z4 [Asin (Wnt + a + c1 + c3 + c4 + d1-g3) + Bsin (Wnt-b-c1 + c3 + c4 + d1-g3)] + X2X4Y1Z4 [Asin (Wnt + a + c2 + c3 + d1-g3) -Bsin (Wnt-b-c2 + c3 + d1-g3)]] (14) Similarly, the -Wn component generated from the (W0-Wn) component is as shown in the expressions (15) and (16) (where,
The harmonic components are omitted).

I signal = (12) × (+ 2Z3cos (W0t + g3 + g4)) = (1/2) × [X1X3Y2Z3 [Bcos (−Wnt + b + c1 + c3 + c4 + d2-g3-g4 ) + Acos (-Wnt-a-c1 + c3 + c4 + d2-g3-g4)] + X2X4Y2Z3 [Bcos (-Wnt + b + c2 + c3 + d2-g3-g4) -Acos (-Wnt-a- c2 + c3 + d2-g3-g4)]] (15) Q signal = (12) × (−2Z4sin (W0t + g3)) = (1/2) × [X1X3Y2Z4 [Bsin (−Wnt + b + c1) + c3 + c4 + d2-g3) + Asin (-Wnt-a-c1 + c3 + c4 + d2-g3)] + X2X4Y2Z4 [Bsin (-Wnt + b + c2 + c3 + d2-g3) -Asin (- Wnt-a-c2 + c3 + d2-g3)]] (16) Finally, the I signal output from the quadrature demodulator is subjected to the amplitude change Z1 and the phase change g1 by the signal processing unit 49 and becomes a demodulated signal I '. The Q signal is received by the signal processing unit 50 to receive the amplitude change Z2 and the phase change g2, and is output as a demodulated signal Q '. Accordingly, in the case of the signals I ′ and Q ′ transmitted and received and demodulated at the positive carrier frequency + Wn, the signals I ′ and Q ′ are transmitted at the negative carrier frequency −Wn and received at the expressions (17) and (18). The demodulated signals I ′ and Q ′ are shown in equations (19) and (20).

I ′ signal = (1/2) × [+ X1X3Y1Z1Z3Acos (Wnt + a + c1 + c3 + c4 + d1-g3-g4 + g1) + X2X4Y1Z1Z3Acos (Wnt + a + c2 + c3 + d1-g3- g4 + g1) + X1X3Y1Z1Z3Bcos (Wnt-b-c1 + c3 + c4 + d1-g3-g4 + g1) -X2X4Y1Z1Z3Bcos (Wnt-b-c2 + c3 + d1-g3-g4 + g1)] (17) Q ' Signal = (1/2) x [+ X1X3Y1Z2Z4Asin (Wnt + a + c1 + c3 + c4 + d1-g3 + g2) + X2X4Y1Z2Z4Asin (Wnt + a + c2 + c3 + d1-g3 + g2) + X1X3Y1Z2Z4Bsin (Wnt- b-c1 + c3 + c4 + d1-g3 + g2) -X2X4Y1Z2Z4Bsin (Wnt-b-c2 + c3 + d1-g3 + g2)] (18) I 'signal = (1/2) × [+ X1X3Y2Z1Z3Bcos (- Wnt + b + c1 + c3 + c4 + d2-g3-g4 + g1) + X2X4Y2Z1Z3Bcos (-Wnt + b + c2 + c3 + d2-g3-g4 + g1) + X1X3Y2Z1Z3Acos (-Wnt-a-c1 + c3 + c4 + d2-g3-g4 + g1) -X2X4Y2Z1Z3Acos (-Wnt-a-c2 + c3 + d2-g3-g4 + g1)] (19) Q ′ signal = (1/2) × [+ X1X3Y2Z2Z4Bsin (-Wnt + b + c1 + c3 + c4 + d2-g3 + g2) + X2X4Y2Z2Z4Bsin (-Wnt + b + c2 + c3 + d2-g3 + g2) + X1X3Y2Z2Z4Asin (-Wnt-a-c1 + c3 + c4 + d2-g3 + g2) -X2X4Y2Z2Z4Asin (-Wnt-a-c2 + c3 + d2-g3 + g2)] (20) These I 'and Q' signals are then subjected to a DFT operation to calculate the carrier frequencies + Wn and -Wn. The components are determined as complex numbers. This operation means that, when the expressions (17) to (20) are expressed by exponential functions, the magnitude and phase of the rotation vector of ej (+ Wnt) and ej (+ Wnt) are obtained. . Expressions (17) and (19) are real parts, (18)
When the expression and expression (20) are expressed and arranged as exponential functions as imaginary parts, the following two expressions are obtained. Where X '= X1X3,
X = X2X4, Z '= Z1Z3, Z = Z2Z4, Y' = Y1,
Y = Y2.

+ Wn component (1/4) ej (+ Wnt + c3-g3) × [Aeja (+ X′Y′Z′ej (+ c1 + c4 + d1-g4 + g1) + XY′Z′ej (+ c2 + d1-g4 + g1) + X'YZ'ej (+ c1-c4-d2 + g4-g1) -XYZ'ej (+ c2-d2 + g4-g1) + X'Y'Zej (+ c1 + c4 + d1 + g2) + XY'Zej (+ c2 + d1 + g2) -X'YZej (+ c1-c4-d2-g2) + XYZej (+ c2-d2-g2)) + Be-jb (+ X'YZ'ej (-c1 -c4-d2 + g4-g1) + XYZ'ej (-c2-d2 + g4-g1) + X'Y'Z'ej (-c1 + c4 + d1-g4 + g1) -XY'Z'ej (-c2 + d1-g4 + g1) -X'YZej (-c1-c4-d2-g2) -XYZej (-c2-d2-g2) + X'Y'Zej (-c1 + c4 + d1 + g2) -XY'Zej (-c2 + d1 + g2))] -Wn component (1/4) ej (-Wnt + c3-g3) × [Bejb (+ X'YZ'ej (+ c1 + c4 + d2-g4 + g1) + XYZ ' ej (+ c2 + d2-g4 + g1) + X'Y'Z'ej (+ c1-c4-d1 + g4-g1) -XY'Z'ej (+ c2-d1 + g4-g1) + X'YZej ( + c1 + c4 + d2 + g2) + XYZej (+ c2 + d2 + g2) −X′Y′Zej (+ c1-c4-d1-g2) + XY′Zej (+ c2-d1-g2)) + Ae-ja ( + X'Y'Z'ej (-c1-c4-d1 + g4-g1) + XY'Z'ej (-c2-d1 + g4-g1) + X'YZ'ej (-c1 + c4 + d2-g4 + g1 ) -XYZ'ej (-c2 + d2-g4 + g1) -X'Y'Zej (-c1-c4-d1-g2) -XY'Zej (-c2-d1-g2) + X'YZej (-c1 + c4 + d2 + g2) −XY Zej (-c2 + d2 + g2))] Both the + Wn component and the -Wn component display the magnitude and phase by combining 16 vectors. The equation for the upper + Wn component indicates A'ej (+ Wnt + a '), and the equation for the lower -Wn component is B'
ej (-Wnt + b '), p' + jq ',
Obviously it means r '+ ju'.

When these expressions are expressed again by complex numbers and newly arranged coefficients S0 to S7 are introduced, the result after the DFT operation is (p '+ jq') = (p + jq) (S0 + jS1) + (r −ju) (S2 + jS3) (21) The relationship of (r ′ + ju ′) = (r + ju) (S6 + jS7) + (p−jq) (S4 + jS5) (22) is obtained. That is,

[0064]

[Equation 11] It is.

From this, S0 indicates the rate at which the real part of the positive carrier is transmitted to the real part of the positive carrier, and indicates the rate at which the imaginary part of the positive carrier is transmitted to the imaginary part of the positive carrier. S1 indicates the rate at which the real part of the positive carrier leaks to the imaginary part of the positive carrier, and the rate at which the imaginary part of the positive carrier leaks to the real part of the positive carrier. S2 indicates the rate at which the real part of the negative carrier leaks to the real part of the positive carrier, and the imaginary part of the negative carrier is
The ratio of leakage to the imaginary part of positive carriers is shown.

S3 indicates the rate at which the real part of the negative carrier leaks to the imaginary part of the positive carrier, and the rate at which the imaginary part of the negative carrier leaks to the real part of the positive carrier. S4 indicates the rate at which the real part of the positive carrier leaks to the real part of the negative carrier, and the imaginary part of the positive carrier is
The rate of leakage of negative carriers to the imaginary part is shown. S5
Indicates the rate at which the real part of the positive carrier leaks to the imaginary part of the negative carrier, and the rate at which the imaginary part of the positive carrier leaks to the real part of the negative carrier. S6 indicates the rate at which the real part of the negative carrier is transmitted to the real part of the negative carrier, and indicates the rate at which the imaginary part of the negative carrier is transmitted to the imaginary part of the negative carrier. S7 indicates a rate at which the real part of the negative carrier leaks to the imaginary part of the negative carrier,
This shows the rate at which the imaginary part of the negative carrier leaks to the real part of the negative carrier. Here, the description of-and + of the rate was omitted.

That is, the above-mentioned coefficients S0 to S7 indicate the characteristics of the transmission path of the I signal and the Q signal. By calculating these coefficients S0 to S7, the characteristics of the transmission path can be detected. Further, by obtaining the inverse matrix of equation (23), it is possible to correct the received data and estimate the transmitted data.

Therefore, in the embodiment of the present invention, in the transmitting apparatus shown in FIG. 1, the symbol number counting circuit 5 sets 0, 1, 2, 3,. . . , 254, 255, 0,
1,2. . . A symbol number which is sequentially and cyclically increased is supplied to the reference signal insertion circuit 6, and is also supplied to the arithmetic unit 4 so that the symbol number is assigned to a specific carrier (for example, the first carrier). insert.

The reference signal insertion circuit 6 inserts a reference signal as known reference data into data transmitted at a certain carrier frequency + Wn, and may leak as an image component or crosstalk due to orthogonality error. A known reference signal (reference data) is also inserted into data transmitted at a certain negative carrier frequency -Wn symmetrical with respect to a center carrier frequency F0. The carrier frequency at which the reference signal is inserted and transmitted is determined in advance in association with the symbol number, and is switched at regular intervals.
This is because transmission characteristics are often different for each frequency.

As an example, two types of reference signals are inserted into the carrier specified by the symbol number. That is, the reference signal of the determinant represented by the expression (24a) for the even symbol and the expression (24b) for the odd symbol is inserted.

[0071]

(Equation 12) The carrier frequency for inserting the reference signal is a set of positive and negative carrier frequencies symmetrical with respect to the center carrier frequency F0, and is switched every two symbols as described above. As we did, 2
56 symbols are transmitted at all 128 positive and negative carrier frequencies. That is, an arbitrary carrier frequency transmits a reference signal at a 256 symbol period.

On the other hand, in the frequency division multiplexed signal receiving apparatus shown in FIG.
The corrected transmission signals R ′ and I ′ are obtained as shown in the block diagram of FIG. That is, in FIG.
The received I signal I ′ and the received Q signal Q ′ extracted from the signal processing units 49 and 50 are respectively supplied to the FFT symbol number decoding circuit 311 in FIG. 5, where the FFT operation is performed and the symbol numbers are decoded first. Next, the reception value (reception reference signal) of the positive / negative carrier corresponding to the symbol number is obtained.

The set of received reference signals is p 1 for even symbols.
S ', q1S', r1S ', u1S', p2S 'for odd symbols,
q2S ', r2S', and u2S '. These symbol numbers and reception reference signals are supplied to the transmission path characteristic detection circuit 312. The FFT symbol number decoding circuit 311 outputs the I signal and Q obtained by performing IDFT operation in the transmitting apparatus.
The signal is subjected to an FFT operation to obtain modulated signal output R 'signal and I' signal of I signal and Q signal. The received reference signal and the received transmission information R ′ signal and I ′ signal other than the symbol number are supplied to the correction circuit 314.

The transmission path characteristic detecting circuit 312 divides the input received reference signal by a known reference signal as shown by the following equation to calculate the coefficients S0 to S7.

[0075]

(Equation 13) The above equation is a determinant obtained by substituting the equations (24a) and (24b) into the equation (23). Since there are eight unknowns, coefficients S0 to S7 can be obtained by transmitting and receiving two types of reference signals. Naturally, since the reference signal is known in the transmitting device and the receiving device, the value is not limited to the values of the expressions (24a) and (24b) as long as the coefficients S0 to S7 can be obtained. Thus, the characteristics of the transmission path can be detected from the coefficients S0 to S7.

Further, by obtaining the inverse matrix of equation (23), it is possible to correct the received data and estimate the transmitted data. Here, the inverse matrix of the equation (23) is expressed by the following equation.

[0077]

[Equation 14] Here, H0 to H7 and det A in the above equation are represented by the following equations.

H0 = + S0 (S6S6 + S7S7) -S2 (S4S6 + S5S7) + S3 (S4S7-S5S6) H1 = + S1 (S6S6 + S7S7) -S3 (S4S6 + S5S7) -S2 (S4S7-S5S6) H2 = + S4 (S2S2 + S3S3) -S6 (S0S2 + S1S3) + S7 (S0S3-S1S2) H3 = + S5 (S2S2 + S3S3) -S7 (S0S2 + S1S3) -S6 (S0S3-S1S2) H4 = + S2 ( S4S4 + S5S5) -S0 (S4S6 + S5S7) -S1 (S4S7-S5S6) H5 = + S3 (S4S4 + S5S5) -S1 (S4S6 + S5S7) + S0 (S4S7-S5S6) H6 = + S6 (S1S0 + S1) -S4 (S0S2 + S1S3) -S5 (S0S3-S1S2) H7 = + S7 (S0S0 + S1S1) -S5 (S0S2 + S1S3) + S4 (S0S3-S1S2) det A = S0 × H0 + S1 × H1 + S4 × H2 + S5 × H3 Then, the correction formula deriving and holding circuit 313 calculates det A and H0 to H7 from the input coefficients S0 to S7 based on the above formula.
Is calculated, and from these calculated values, the value of the following correction formula of the inverse matrix in the formula (26) is calculated and stored.

[0079]

(Equation 15) In this way, a correction formula for the corresponding positive / negative carrier is prepared. The corresponding positive / negative carrier is determined by the symbol number. Naturally, there is a correction formula for each carrier, and when 257 carriers are used as in this embodiment, about 12
Eight correction equations are sequentially calculated and held. In addition, the coefficient S0
Since S7 changes every moment, the correction formula is also updated every moment.

The correction circuit 314 converts the information received from the FFT symbol number decoding circuit 311 into the correction expression deriving and holding circuit 31.
Using the respective correction formulas derived and held by step 3, each of the positive and negative carrier pairs is used to calculate and correct the following formula, thereby outputting corrected transmission information R 'and I'.

[0081]

(Equation 16) However, in the expression (27), a to d are different values from the a to d in the expressions (1) to (20), and the complex number of the reception data (after correction) assigned to the frequency + Wn is (a + jb). ) And the complex number of the received data (after correction) assigned to the frequency −Wn is (c + jd). In the expression (27), a 'to d' represent the complex numbers of the reception data (before correction) assigned to the frequency + Wn by (a '+ j
b ′), and the value when the complex number of the received data (before correction) assigned to the frequency −Wn is (c ′ + jd ′).

The FFT operation in the FFT / QAM decoding circuit 31 is performed by a digital signal processor (D
When the correction is performed in (SP), the above-described correction can be performed according to the flowchart shown in FIG. That is, first, the reference signal of the frequency is extracted based on the symbol number (step 61), and the coefficients S0 to S7 of the frequency are calculated by the equation (3) (step 62). Further, a correction formula (inverse matrix) is calculated or updated for the coefficients S0 to S7 of the frequency based on the formulas (4) and (5) (step 63), and the received information signal is calculated from each correction formula at each frequency. R ′ and I ′ are corrected based on the equation (6)
4).

In the embodiment, the reference signal (reference data) is expressed by the following equations (24a) and (24b).
For even symbols, the positive carrier frequency + (as the first set)
A predetermined value ps is set only for the real part of the complex number transmitted by Wn, and the others are set to zero. For the odd symbols, only the real part of the complex number transmitted at the symmetric negative carrier frequency -Wn (as the second set) is a predetermined value. Although rS is set and others are set to zero, the present invention is not limited to this, and may be reference data represented by the following equation.

[0084]

[Equation 17] In this case, the transmission path characteristic detection circuit 312 divides the input received reference signal by a known reference signal as shown by the following equation to calculate the coefficients S0 to S7. Note that a value with a dash in the following equation indicates a received value.

[0085]

(Equation 18) Also, as the first set, only the real part and the imaginary part of the positive carrier frequency are set to predetermined values, the real part and the imaginary part of the negative carrier frequency are each set to zero, and the real part of the negative carrier frequency is set as the second set. A predetermined value may be set only for the imaginary part, and data represented by the following equation in which the real part and the imaginary part of the positive carrier frequency are zero may be used as the reference signal.

[0086]

[Equation 19] In this case, the transmission path characteristic detection circuit 312 divides the input received reference signal by a known reference signal as shown by the following equation to calculate the coefficients S0 to S7. Note that a value with a dash in the following equation indicates a received value.

[0087]

(Equation 20) Also, as examples of other reference signals, (X, Y, X, Y)
And (Y, X, Y, X) are transmitted, and the following equation is obtained: p1 ′ = S0 × X−S1 × Y + S2 × X + S3 × Y p2 ′ = S0 × Y−S1 × X + S2 × Y + S3 × X q1 ′ = S1 × X + S0 × Y + S3 × X−S2 × Y q2 ′ = S1 × Y + S0 × X + S3 × YS−X2 r1 ′ = S4 × X + S5 × Y + S6 × X−S7 × Y r2 ′ = S4 × Y + S5 × X + S6 × Y−S7 × X u1 ′ = S5 × X−S4 × Y + S7 × X + S6 × Y u2 ′ = S5 × Y−S4 × X + S7 × Y + S6 × X For the reference signal represented by the following equation, the correction coefficient S0
To S7 can also be obtained.

S0 = (p1′X−p2′Y−q1′Y + q2′X) / (2 (X2−Y2)) S1 = (p1′Y−p2′X + q1′X−q2′Y) / (2 ( X2−Y2)) S2 = (p1′X−p2′Y + q1′Y−q2′X) / (2 (X2−Y2)) S3 = − (p1′Y−p2′X−q1′X + q2′Y) / (2 (X2−Y2)) S4 = (r1′X−r2′Y + u1′Y−u2′X) / (2 (X2−Y2)) S5 = − (r1′Y−r2′X−u1′X + u2 ′) Y) / (2 (X2-Y2)) S6 = (r1'X-r2'Y-u1'Y + u2'X) / (2 (X2-Y2)) S7 = (r1'Y-r2'X + u1'X- u2'Y / (2 (X2-Y2)) By the way, in the above-described first embodiment, the reference signal is periodically transmitted, and the receiving system is thereby corrected. And a new correction formula In the period before the signal is output, the correction is performed by the correction formula based on the previous reference signal, so that the characteristic (environment) change during this period is suitable when the change is gradual. In some cases, sufficient correction cannot be performed for a changing multipath environment, so the second embodiment described below can cope with such a rapidly changing characteristic (environment) change. It was made.

FIG. 7 is a block diagram showing a second embodiment of the transmission side orthogonal frequency division multiplexing signal transmitting apparatus according to the present invention. In the figure, the same components as those of FIG. 1 are denoted by the same reference numerals, and description thereof will be omitted. In FIG. 7, the I and Q signals, which are the results of the IDFT operation continuously read from the output buffer 7, are supplied to the digital quadrature modulator 15, and the intermediate frequency of 42.8 MHz from the intermediate frequency oscillator 16 is applied to the carrier wave. As digital quadrature amplitude modulation (QA
M) and a digital O consisting of 257 information carriers
It is converted to an FDM signal.

This digital OFDM signal is converted into an analog signal by a D / A converter / bandpass filter (BPF) 17, and only a necessary band component is filtered and passed through a frequency converter 12 and a transmission unit 13. And is radiated from an antenna (not shown). The frequency spectrum of this transmission OFDM signal is the same as in FIG. However, as described later, the arithmetic unit 4 is operated by the symbol number counting circuit 14.
Has 9 bits.

The above-mentioned transmission OFDM signal is received and demodulated by a frequency division multiplexed signal receiving apparatus having the same block configuration as in FIG. This frequency division multiplexed signal receiving apparatus
As will be described later, the configuration of the FFT / QAM decoding circuit 31 is characteristic. The frequency division multiplexing signal transmission system including the frequency division multiplexing signal transmitting apparatus shown in FIG. 7 and the frequency division multiplexing signal receiving apparatus shown in FIG. 3 is rewritten as shown in FIG. 8 when focusing on the I signal and the Q signal. be able to.

In FIG. 8, the digital quadrature modulator 15
7 is a digital quadrature modulator 15 shown in FIG. 7, a signal processing unit 72 is a circuit portion including a D / A converter / BPF 17 and an amplifier not shown in FIG. 7, and a transmission system circuit 73 is shown in FIG. The frequency converter 12 and the transmission unit 1 shown in FIG.
3, the characteristics of the spatial transmission path, and the components of the transmission system corresponding to the transmission path including the receiving unit 21, the bandpass filter (not shown), the frequency converter 22, the intermediate frequency amplifier 23, and the like shown in FIG. The multipliers 74 and 75 are circuit portions of the carrier extraction and quadrature demodulator 24 shown in FIG. 3, and the signal processing units 76 and 77 are circuit components of the LPF 28 and the A / D converter 29 shown in FIG. Equivalent to.

Here, the I signal represented by the formulas (1) and (3) and the Q signal represented by the formulas (2) and (4) are respectively subjected to digital quadrature modulation by the digital quadrature modulator 15. After being converted into an OFDM signal, the OFDM signal is input to the signal processing unit 72 in FIG. 8 and receives an amplitude change X and a phase change c. Further, the transmission system circuit 73 receives an amplitude change and a phase change. Here, the amplitude changes of the carrier frequencies + Wn and -Wn of the OFDM signal in the transmission system circuit 73 are collectively represented as Y 'on the + Wn side and Y on the -Wn side, and the phase change is represented as +
It is d1 on the Wn side, and d2 on the -Wn side. Here, the digital quadrature modulator 14 is used to ignore errors in the transmission system.

The OFDM signal passed through the transmission system circuit 73 is subjected to quadrature demodulation by multipliers 74 and 75, respectively. The orthogonality error at this time is g4, and the carrier for demodulation is 2 cos (W0t + g3 + g4) for the I signal and -2s for the Q signal.
in (W0t + g3). The I signal quadrature-demodulated by the quadrature demodulator 24 receives the amplitude change Z 'and the phase change g1 in the signal processing section 76 and outputs it as a demodulated signal I'. The Q signal orthogonally demodulated by the quadrature demodulator 24 receives an amplitude change Z and a phase change g2 in the signal processing unit 77, and is output as a demodulated signal Q ′.

These I ′ signal and Q ′ signal are
The operation is performed, and the components of the carrier frequencies + Wn and -Wn are obtained as complex numbers. Thereby, the following equation is obtained.

+ Wn component A'ej (+ Wnt + a ') = (1/2) ej (+ Wnt + c3-g3) × [Aeja (+ XY'Z'ej (+ c + d1-g4 + g1) + XY 'Zej (+ c + d1 + g2)) + Be-jb (+ XYZ'ej (-c-d2 + g4-g1) -XYZej (-c-d2-g2))]] (31) -Wn component B'ej ( -Wnt + b ') = (1/2) ej (-Wnt + c3-g3) * [Bejb (+ XYZ'ej (+ c + d2-g4 + g1) + XYZej (+ c + d2 + g2)) + Ae- ja (+ XY'Z'ej (-c-d1 + g4-g1) -XY'Zej (-c-d1-g2))] (32) However, in the expressions (31) and (32), A '= √ (p '
2 + q'2), a '= tan-1 (q' / p '), B' = √
(R'2 + u'2) and b '= tan-1 (u' / r '). The values of the real part and the imaginary part of the reference signal transmitted on the positive carrier are p and q, respectively, and the real part of the reference signal transmitted on the negative carrier symmetrical to the positive carrier with respect to the center carrier. And when the values of the imaginary part are r and u, respectively, the values of the real part and the imaginary part of the received reference signal of the positive carrier are p ′ and q ′, respectively, and the values of the received reference signal of the negative carrier are The values of the real part and the imaginary part are r 'and u', respectively. Also, c3 and g3 are the phases of the modulation and demodulation waves of the transceiver, and g4 is the error of the quadrature demodulator 24 on the receiving side.

Expressions (31) and (32) are newly added to the coefficient S
When introducing and organizing 0 to S7, as a result after the DFT operation, the relationship between the above-mentioned expressions (21) and (22) is obtained.
The determinant expressed by the expression (23) is obtained.

Here, in this embodiment, the response speed to the characteristic that changes rapidly due to movement (multipath environment characteristic) is increased. The characteristic changes are Y ′, Y, d
1, appears at d2. Each change is MY ', NY,
It is assumed that d1 + d3 and d2 + d4. Thereby, (31)
The equation and the equation (32) are as follows.

A'ej (+ Wnt + a ') = (1/2) ej (+ Wnt + c3-g3) × [Aeja (+ XMY'Z'ej (+ c + d1 + d3-g4 + g1) + XMY 'Zej (+ c + d1 + d3 + g2)) + Be-jb (+ XNYZ'ej (-c-d2-d4 + g4-g1) -XNYZej (-c-d2-d4-g2))) (33) B'ej (-Wnt + b ') = (1/2) ej (-Wnt + c3-g3) × [Bejb (+ XNYZ'ej (+ c + d2 + d4-g4 + g1) + XNYZej (+ c + d2) + d4 + g2)) + Ae-ja (+ XMY'Z'ej (-c-d1-d3 + g4-g1) -XMY'Zej (-c-d1-d3-g2)))] (34) Where Mejd3 = V1 + jV2, Me-jd3 =
V1-jV2, Nejd4 = V3 + jV4, Ne-jd4 = V
Taking 3-jV4, the determinant shown in the above equation (23) is expressed as follows.

[0100]

(Equation 21) Therefore, it is possible to take out a new correction formula separated from the conventional correction formula using the reference signal as it is. Further, although the expression (35) describes the reference signal, since this also applies to a normal information signal to be transmitted, the expression (35) can be expressed as the following expression.

[0101]

(Equation 22) Where a and b are real and imaginary parts of transmission data allocated to the positive carrier frequency, and c and d are real and imaginary parts of transmission data allocated to the negative carrier frequency. , A ′ and b ′ are the real and imaginary parts of the received data of the positive carrier frequency, and c ′ and d ′ are the real and imaginary parts of the received data of the negative carrier frequency.

The S matrix indicating the channel characteristics in equation (35) and the H matrix in equation (36), which is the inverse of the S matrix, can be obtained in the same manner as in the first embodiment. On the other hand, the K matrix in the equation (36), which is the inverse matrix of the V matrix showing the fast change characteristic in the equation (35), can be expressed by the following equation.

[0103]

(Equation 23) As can be seen from equation (37), this K matrix has many elements of 0, so that the amount of calculation is much smaller than in the case of obtaining the H matrix. Since this K matrix is updated at all frequencies for each symbol, it cannot be realized with a large amount of calculation, but can be realized according to the equation (37). Also, it can be seen that the K matrix is not a set of positive and negative frequencies, but is independent for each frequency.

As the processing means, the H matrix is updated every time a reference signal is received, as in the first embodiment. At the time of updating the H matrix, the K matrix is a unit matrix. Then H
With respect to the first correction information corrected by the matrix, a difference from a desired signal point arrangement, that is, a high-speed change in amplitude and phase is detected for each symbol, and based on that, the K matrix is updated for each symbol for each frequency. . The updated K matrix is used for the next symbol.

Next, as will be described in detail with respect to the flow for each symbol, it is assumed that a reference signal arrives at a predetermined frequency in a certain symbol n. At this point, the H matrix is updated. The K matrix is a unit matrix. The H matrix is obtained by the same method as in the first embodiment, and the correction equation is obtained by the following equation.

[0106]

(Equation 24) The next received and input symbol n + 1 is corrected by the following equation.

[0107]

(Equation 25) The signals <a>, <b>,
Decoded data is generated from <c> and <d>. here,
When there is no high-speed characteristic change (the transmission path characteristics are the same)

[0108]

(Equation 26) And the complete transmission signals a, b, c and d are corrected and decoded.

However, when mobile communication is taken into consideration, a characteristic change appears. This characteristic change can be expressed by the following equation.

[0110]

[Equation 27] Can be requested. Here, the left side of equation (41) is
The transmission information signals a, b, c, and d correspond to signals obtained by adding errors δp, δq, δr, δu, which are high-speed change components, respectively,
And this means that they correspond to the information signals a ″, b ″, c ″, d ″ corrected by the first correction formula. In the equation (43), K0, K1, K6 and K7 are represented by the following equations.

K0 = (aa ″ + bb ″) / (a ″ 2 + b ″ 2) (44a) K1 = (ab ″ −a ″ b) / (a ″ 2 + b ″ 2) (44b) K6 = (cc ″ + dd ″) ) / (C ″ 2 + d ″ 2) (44c) K7 = (cd ″ −c ″ d) / (c ″ 2 + d ″ 2) (44d) The K matrix in equation (43) is used for the next symbol n + 2. For the symbol n + 2, correction is performed based on the following equation, and decoded data is generated from <a>, <b>, <c>, and <d> obtained by the following equation.

[0112]

[Equation 28] Then, based on a ″, b ″, c ″, and d ″ of Expression (45), a new K matrix is calculated by the same expression as Expressions (42) and (43). The new K matrix thus obtained is used for the next symbol n + 3.

For the symbol n + 3, the correction is performed by the same expression as the expression (45), and the correction signal <
Decoded data is generated from a>, <b>, <c>, and <d>. In this symbol n + 3, (45)
Correction signals <a>, <b>, <c>, and <d> are obtained by the same equations as equation (45) except that the K matrix in the equation is the K matrix obtained with symbol n + 2. Can be calculated.

In the same manner as above, correction signals <a>, <b>, <c>, and (c) are obtained for each of the subsequent symbols in the same manner as described above, using the K matrix obtained for the previous symbol and the same equation as equation (45). <D>, and the calculated correction signal <a>
>, <B>, <c>, and <d>, and generate a new K matrix based on a ″, b ″, c ″, and d ″ for the next symbol (42). Equation (4)
Calculated by the same equation as equation 3). The above (3)
In the expressions 5) to (45), the transmission signals a, b, c, and d and the reception signals a ', b', c ', and d' are, of course, different numerical values.

In the above description, the method of updating the K matrix from the received information signal for each symbol without changing the H matrix until the next reference signal arrives has been described. The H matrix may be synthesized and replaced with new H matrices one after another as in the following equation.

[0116]

(Equation 29) In this case, the correction for a certain symbol

[0117]

[Equation 30] , And a new H matrix is generated again in the same manner as in equation (46) (in this case, the old H matrix is the H matrix in equation (47)). The new H matrix is used for the next symbol.

Next, a more specific processing example by the above method will be described. In the transmitting device of FIG.
In the same manner as in the first embodiment, a symbol number is inserted into a specific carrier, and a base reference signal (reference data) is inserted into another positive or negative carrier corresponding to the symbol number. Specifically, the symbol number counting circuit 14 in FIG. 7 is represented by 9 bits, and 0, 1, 2, 3,. . . , 511,0,
1, 2,. . As described above, the symbol number which sequentially and cyclically changes in each symbol period is counted and output.

Here, since accurate decoding of the symbol number is important, dedicated reference data (reference signal) is prepared and multi-level QAM (256 QAM) used for another carrier is used.
Multi-level modulation with a smaller number of multi-levels is performed. Specifically, among the above 9 bits representing the symbol number, 9, 8,
The third and second four bits are transmitted and received by 16QAM. On the receiving side, the symbol number from these 4 bits is changed to 9 bits. This decoding is easy because the symbol number is a sequence of numbers that are sequentially increased by one.

In FIG. 7, symbol number counting circuit 14
Of the 9 bits output from, four bits of the ninth, eighth, third, and second bits are input to the arithmetic unit 4, where an IDFT operation is performed so as to be transmitted on a specific carrier, for example, the first carrier. Further, the reference signal insertion circuit 6 receives the 9-bit symbol number output from the symbol number counting circuit 14, inputs the reference signal obtained based on the upper 7 bits of the 9-bit symbol number to the arithmetic unit 4, and inputs a specific carrier wave, for example, An IDFT operation is performed so that the signal is transmitted on the m1th carrier. Since the lower two bits of the symbol number are ignored, a reference signal having the same value is inserted between four symbols.

Also, based on the least significant bit of the 9-bit symbol number, reference signal insertion circuit 6 inserts the following two types of reference signals into odd symbols and even symbols.

[0122]

(Equation 31) The reference signal of the determinant represented by Expression (49a) is inserted into even symbols, and the reference signal of the determinant represented by Expression (49b) is inserted into odd symbols. Here, X is a known reference signal value. The carrier frequency into which the reference signal is inserted is transmitted as a set of positive and negative carrier frequencies symmetrical with respect to the center carrier frequency F0.

On the other hand, the frequency division multiplexed signal receiving apparatus in this embodiment has substantially the same configuration as that of FIG.
The FFT and QAM decoding circuit shown at 31 is configured as shown at 31 'in the block diagram of FIG.
And I. 9, the same components as those of FIG. 5 are denoted by the same reference numerals. 9, the signal processing unit 7 shown in FIG.
The received I signal I 'and the received Q signal Q' extracted from 6, 77 are supplied to an FFT symbol number decoding circuit 311. The symbol number is first decoded, and then a set of positive and negative carriers corresponding to the symbol number is obtained. Is obtained.

In this embodiment, since the predetermined 4 bits of the symbol number are subjected to 16 QAM, the symbol number can be decoded with a better error rate than other transmission information, and 0 to 0 represented by 9 bits. Symbol numbers up to 511 can be reliably decoded, and the insertion carrier of the transmission information reference signal can be reliably specified for each symbol.

The value of the received reference signal is p0s', q0s', r0s', u0 in the 0th symbol (even symbol).
In s', for the first symbol (odd symbol), p1
In s ', q1s', r1s ', and u1s', p2s ', q2s', r2s ', and u2s' are used in the second symbol (even symbol).
In the third symbol (odd symbol), p3s',
q3s', r3s', and u3s'.

The transmission path characteristic detection circuit 312 shown in FIG. 9 is based on the above equations (23), (49a) and (49b). Are calculated.

S0 = (p0s ′ + q1s ′) / (2X), S4 = (r0s′−u1s ′) / (2X) S1 = (q0s′−p1s ′) / (2X), S5 = (u0s′−) r1s') / (2X) S2 = (p0s'-q1s') / (2X), S6 = (r0s'-u1s') / (2X) S3 = (q0s' + p1s') / (2X), S7 = (u0s'-r1s') / (2X) (50) The coefficients S0 to S7 are obtained in the same manner for the second symbol and the third symbol. Thereafter, the coefficients S0 to S7 are respectively averaged to remove white noise. In this embodiment, an average among four symbols is obtained.

Since there are eight coefficients S0 to S7, the coefficients can be obtained by transmitting and receiving two types of reference signals. Of course,
Since the reference signal is known in the receiving apparatus, any reference signal may be used as long as the coefficient can be obtained. In this way, the transmission line characteristic detection circuit 312 detects the transmission line characteristic coefficients S0 to S7 and supplies them to the first correction expression derivation and holding circuit 313 together with the symbol numbers.

The first correction formula deriving and holding circuit 313 performs the same operation as in the first embodiment. That is, the coefficients S0
From S7, the above det A and H0 to H7 are calculated, and from these calculated values, the value of the first correction formula represented by the matrix shown in the above equation 14 is calculated and stored. Here, since 257 carriers are used, about 128 first
Are sequentially generated and updated every moment. First
Is obtained by averaging one corresponding positive / negative carrier with four symbols, the interval at which the average correction formula of the same positive / negative carrier is updated next is 512 symbols later (4 symbols × 128 pairs = 512 symbols).

After the transmission path characteristics are detected as described above, instead of calculating the inverse matrix, another method may be used to directly calculate the correction equation shown in Equation 14 from Equation (26). Good.

The first correction circuit 314 derives and holds the information received from the FFT symbol number decoding circuit 311 by the first correction formula derivation and holding circuit 313.
(27) as in the first embodiment using the correction equation
The following equation corresponding to the equation is performed, and a corrected signal is output.

[0132]

(Equation 32) Where a ″ and b ″ are real and imaginary parts of the corrected received data assigned to the positive carrier frequency, and c ″ and d ″ are assigned to the negative carrier frequency. The real part and the imaginary part of the corrected received data, a ′ and b ′ are the real part and the imaginary part of the received data of the positive carrier frequency, and c ′ and d ′ are the real parts of the received data of the negative carrier frequency. And the imaginary part.

Thus, the received signal is corrected in the same manner as in the first embodiment, and the corrected transmission information R ″ (a ″, c ″) and I ″ (b) are output from the first correction circuit 314. ”,
d ″), and the second correction circuit 316 and the second
, Respectively. Further, the symbol number is supplied from the first correction formula derivation and holding circuit 313 to the second correction formula derivation and holding circuit 315.

The second correction circuit 316 uses the K matrix which is a unit matrix for the next symbol after receiving the reference signal, and the K matrix generated for the previous symbol for the other symbols, using the following equation.

[0135]

[Equation 33] The input signals a ″, b ″, c ″, and d ″ are further corrected (second correction) based on <a>, <b>, <c>, and <d
>, Based on which the decoded data a, b, c and d are generated and output to the output circuit 32 in FIG.

The second correction formula deriving and holding circuit 315
Corrected signal a ″ input from the correction circuit 314 of
b ″, c ″ and d ″ and the first correction formula deriving and holding circuit 31
3 and the second correction circuit 31
Based on the decoded data a, b, c, and d input from No. 6, a K matrix is generated based on the above equation (43), and this is held as a second correction equation. This K matrix is generated for each received carrier frequency and used for the next symbol.

In this manner, the second correction circuit 316 corrects errors and characteristics that change relatively slowly, such as aging and temperature changes, by the first-stage correction using the first correction formula. In addition, accurate correction is performed by using a known reference signal, and subsequently, a relatively high speed such as a multipath environment generated in mobile communication or the like is corrected by a second-stage correction using a second correction formula. The decoded data a, which has been optimized for each symbol,
b, c and d (real part data R and imaginary part data I) are output.

Next, the operation of the correction method using the second correction formula will be described using specific numerical examples. Regarding the H matrix, the operation of the K matrix will be described in detail by simply expressing the description and omitting the description. In addition, since the H matrix is corrected by a set of positive and negative carriers, the K matrix is also expressed by a set. However, since the K matrix is independent of positive and negative carriers, only one of the descriptions is sufficient.
Here, description will be made focusing on only one frequency.

First, in a certain symbol n, (7.5, 7.5) is transmitted as a predetermined reference signal value (real part, imaginary part), and the receiving apparatus converts this reference signal to (6.25, 6.25). 2
5) It is assumed that the data is received at a value. Here, the H matrix is simply expressed by the following equation.

[0140]

(Equation 34) That is, the transmission system does not include an error, there is no change in the phase characteristic, and the amplitude characteristic is 6.25 / 7.5 times.

In the first-stage correction using the first correction equation, correction is performed using this H matrix, and thereafter, no change is made until a new reference signal arrives. Equation (53) is the first correction equation held by the first correction equation derivation holding circuit 313. At this point, the K matrix is

[0142]

(Equation 35) And This is the second correction formula held by the second correction formula deriving and holding circuit 315.

In the next symbol n + 1, it is assumed that the received transmission information has the following equation.

[0144]

[Equation 36] Accordingly, the second correction circuit 316 determines that the received data is data transmitted at (7.5, 6.5), and outputs the data of the real part and the imaginary part to the output circuit 32 shown in FIG. (7, 6) is passed as received data as a set (R, I) of decoded data. Note that 0.5 is a bias value added to simplify QAM decoding, and is a conventionally known method.

Finally, the second correction formula derivation holding circuit 315
Calculates a new K matrix based on the above equations (43) and (44), derives this as the second correction equation for the next symbol n + 2, and holds it. This newly calculated K
The matrix is represented by the following equation.

[0146]

(37) This K matrix represents a change in the amplitude characteristic and the phase characteristic generated in the transmission path from the transmission of the symbol n to the transmission of the symbol n + 1.

In the next symbol n + 2, it is assumed that the received transmission information has the following equation.

[0148]

(38) Therefore, the second correction circuit 316 determines that the received data is data transmitted at (7.5, 2.5), and outputs the real part data and the imaginary part to the output circuit 32 shown in FIG. (7, 2) is passed as received data as a set (R, I) of decoded data. Here, assuming that only the first-stage correction by the first correction formula is performed, the received data is (6.
In order to determine that the data is transmitted in (5, 2.5), (6, 2) is passed to the output circuit 32 as received data as a set (R, I) of decoded data. Compared with this embodiment, more reliable decoded data can be obtained.

Finally, the second correction formula deriving and holding circuit 315
Calculates a new K matrix based on the above equations (43) and (44), derives this as a second correction equation for the next symbol n + 3, and holds it. This newly calculated K
The matrix is represented by the following equation.

[0150]

[Equation 39] The K matrix represents a change in the amplitude characteristic and the phase characteristic generated between the transmission of the symbol n and the transmission of the symbol n + 2. Hereinafter, the same operation as described above is repeated.

Next, each modified example of the second embodiment will be described.

(Modification 1) The frequency at which the reference signal is inserted is not limited to one set, but may be inserted at several sets. In the above description of the second embodiment, the reference signals are assigned to the same frequency among a plurality of symbols, and the obtained coefficients S0 to S7
Were averaged, that is, the transmission path system from which Gaussian noise was removed was detected, and the first correction equation was used.

The second correction equation can be averaged to remove Gaussian noise. Here, a method of averaging the K matrix over five symbols will be described.

It is assumed that a reference signal is transmitted and received in symbol n. As initial values, K01 = 1, K02 = 1,
K03 = 1, K04 = 1, K05 = 1, K11 = 0, K12
= 0, K13 = 0, K14 = 0, and K15 = 0. Then, K0 = (K01 + K02 + K03 + K04 + K05) / 5 K1 = (K11 + K12 + K13 + K14 + K15) / 5 (56) For the symbol n + 1, the second average correction is performed by the equation (56). The method for obtaining a new K matrix is as described above. The coefficients obtained here are designated as K00 and K10.
Next, K0m is replaced with K0m + 1, and K1m is replaced with K1m + 1, and a second correction equation (averaged K matrix) averaged by equation (56) is generated (where m = 0, 1, 2,
3, 4).

By repeating such processing, the coefficients among the latest five symbols are held and averaged. This averaging process is performed by the second correction formula deriving and holding circuit 315.

(Modification 2) (43), (44a) to (4)
4d) When calculating the K matrix by the equations (52), a ″ to
In the error signal at d ″ (that is, in the case of the signal a ″, the difference between the signal a and the signal a ″, and the other signals b ″ to d ″), predetermined upper and lower limits are set, and Otherwise, if it is less than that, the K matrix is replaced with a predetermined value.
Specifically, in either the real part or the imaginary part, when the error signal value becomes 0.4 or more or -0.4 or less, 0.4 or -0.4 is set, respectively. Substitute the K matrix.

Second correction (or second average correction)
In the case where a change larger than the high-speed change that can be followed by the above is generated, or when the S / N is extremely deteriorated, an error signal is turned back and the second correction formula may be reversely corrected. If averaging is performed, the effect is small, but if the averaging occurs continuously, an error will also occur. The above-described limit value is provided to avoid the reverse correction. This restriction process is performed by the second correction formula derivation holding circuit 315.

(Modification 3) (43), (44a) to (4)
4d) When calculating the K matrix by the equations (52), a ″ to
A predetermined weighting is performed on the error signal at d ″. This is particularly effective when the S / N is extremely deteriorated.

If the S / N ratio is extremely deteriorated, the error signal may be turned back, and the second correction equation may be inversely corrected. If averaging is performed, the effect is small, but if the averaging occurs continuously, an error will also occur. However, it can be considered that the error signal has variation due to the influence of Gaussian noise. Therefore, the distribution may be normal. Therefore, the K matrix is generated by increasing the weight of the portion close to the median value, increasing the influence, and reducing the weight of the peripheral portion to reduce the influence.

As a simple example, a case where averaging is performed twice will be described. When the absolute value of the error signal is 0.1 or less, 5.
4 when the value is greater than 0.1 and 0.2 or less; 3 when the value is greater than 0.2 and 0.3 or less; 2 when the value is greater than 0.3 and 0.4 or less; A table for weighting such as 1 is provided, and then the whole is divided by the number of weights.

In a numerical example, the first error signal is 0.
15, if the second error signal is 0.45, the total error signal when averaging twice is 3 (= (0.15 + 0.45) / 2) without weighting, When weighting is performed, 0.21 (= (0.15 × 4
+ 0.45 × 1) / (4 + 1)). Such weighting is performed by the second correction formula deriving and holding circuit 315.

(Modification 4) Second correction (or second correction)
In the case where a change equal to or more than the high-speed change that can be followed by the average correction is performed, or when the S / N is extremely deteriorated, the error signal is folded back and the second correction formula may be inversely corrected. The effect is small if averaging is performed,
If they occur consecutively, an error will still occur.

Therefore, in this modification, an error is detected based on a signal generated from an error correction circuit (not described), and the second correction equation is not updated when an error is detected. This is because, when an error occurs, inverse correction is performed even if the K matrix, which is the second correction equation, is generated. In this case, the generation of the K matrix is not performed, and the matrix is replaced with a unit matrix and complemented by the effect of averaging.

(Modification 5) (43), (44a) to (4)
4d) When calculating the K matrix by the equations (52), <p> to <p>
In the signal point arrangement at u> (integer part), predetermined upper and lower limits are set, and in the case of signal point arrangements that are greater or less than that, the K matrix is replaced with a unit matrix.

In a simple specific example, when the value of any of the real part, the imaginary part, and the real part + imaginary part is equal to or more than a predetermined value, the K matrix is replaced with a unit matrix. To explain with a numerical example, in 256QAM, in a signal point arrangement where the absolute value of the real part or the imaginary part is 8 or more, or the absolute value of the real part + imaginary part is 6 or more, the error signal is not used and the unit matrix Substitute. Figure 1 shows an example
0 is shown. In the figure, black circles indicate signal points to be used, and white circles indicate signal points not to be used.

In particular, when a change equal to or more than a high-speed change that can be followed by the second correction (or the second average correction) occurs, the error signal is folded back, and the second correction equation is inversely corrected. There is. If averaging is performed, the effect is small, but if the averaging occurs continuously, an error will also occur. Since the high-speed change affects the signal point arrangement on the outer peripheral side, there is a high probability that an error also occurs on the outer peripheral side. Therefore, in this modification, only the more reliable error signal of the signal point arrangement on the inner peripheral side, which is indicated by a black circle in FIG. 10, is used.

(Modification 6) In the above embodiment, the H matrix, the K matrix, or the matrix obtained by combining the H matrix and the K matrix described in the equations (46) to (48) are averaged on the time axis. The conversion process has been described. In this modification, averaging processing is performed on the frequency axis.

Each carrier of the OFDM signal is set adjacent to each other, and shows similar characteristics between each other. That is, when the coefficients of each matrix are arranged in order of frequency, a curve of several orders is obtained. Further, the change of the curve is interlocked, and does not have a high-frequency component higher than a predetermined value. If there is a large change point in this curve, it may be considered that the deterioration of S / N or the averaging on the time axis is not sufficiently performed, and an inappropriate correction coefficient is calculated.

Therefore, in this modification, after calculating each coefficient for a matrix obtained by combining the H matrix and the K matrix, a predetermined amount of low-pass filter is inserted into a sequence in the order of the frequency axis to remove high-frequency components. Such a filtering process is performed by the first correction formula deriving and holding circuit 313 or the second correction formula deriving and holding circuit 3
15 can be easily implemented by digital filtering. This filtering may be configured by a two-dimensional filter using not only one-dimensional operation on the frequency axis but also the above-described arrangement of coefficients on the time axis.

The two-dimensional filter on the frequency axis and the time axis will be briefly described. The aforementioned H matrix is averaged by transmitting and receiving the reference signal for a plurality of symbols. Also, the K matrix is averaged among a plurality of symbols. As a result, averaging on the time axis is performed.
Then, newly

[0171]

(Equation 40) Is generated, and the coefficient of the E matrix is close to the coefficient of the own carrier (for example, ± 10 carriers).
Average the coefficients of Of course, the same averaging is performed for all carriers. Thereby, the average on the frequency axis is obtained. Then, a correction calculation is performed by the following equation.

[0172]

[Equation 41] Until now, a two-step calculation was performed using the H matrix and the K matrix, respectively. In this case, the processing for obtaining the H matrix and the K matrix is the same. (5
Correction is made by equation 8).

As another example, a set is formed for each of three adjacent carriers, a reference signal is inserted into only the center carrier, and a correction formula deriving process is performed. It is also possible to correct by a correction formula.
As a result, the amount of calculation can be reduced and the arrival period of the reference signal can be shortened, so that the device can be reduced in cost and high-speed followability can be achieved.

Next, a third embodiment of the present invention will be described. In each of the first and second embodiments and the modified examples described above, a reference signal having a known value is inserted in a pair of positive and negative carriers symmetric with respect to the center carrier of the OFDM signal. On the other hand, in the third embodiment, the reference signal is transmitted / received without being limited to the positive / negative set, and the second correction method of the second embodiment is used. Also performs OFDM signal correction that can cope with mobile communication or a multipath environment that changes at high speed. Further, in the third embodiment and the first and second embodiments, if the symbol number can be identified on the receiving side, the transmission signal form is not questioned. The present invention can also be applied to a device that transmits data by using a.

In the third embodiment, an OFDM signal is transmitted from a frequency division multiplex signal transmitting apparatus having the same configuration as that of the transmitting apparatus having the configuration shown in FIG. However, the reference signal is transmitted on one carrier in the OFDM signal. Further, the transmitting side specifies a specific carrier modulated with a known reference signal based on a symbol number or specific parameter information or synchronization symbol information transmitted on a predetermined carrier, and sequentially circulates at regular intervals. And then send it.

Here, information represented by a complex number (p + jq) assigned to a certain carrier frequency + Wn is calculated by the arithmetic unit 4
Are generated as I and Q signals represented by the above equations (1) and (2), which are converted into OFDM signals and transmitted at the frequency + Wn therein.

The I signal and the Q signal represented by the above equations (1) and (2) are transmitted through the transmission system circuit 80 schematically shown in FIG. Are received as I 'and Q' signals by the frequency division multiplexed signal receiving apparatus shown in FIG. However, as will be described later, this embodiment is characterized in that the FFT and QAM decoding circuit 31 in the frequency division multiplexed signal receiving apparatus shown in FIG. 3 has a configuration indicated by 31 ″ in FIG. ,
The transmission system circuit 80 is the frequency converter 1 shown in FIG.
2, the transmission unit 13 corresponds to the characteristics of the spatial transmission path. Also,
Here, the errors received by the frequency division multiplex signal transmitting apparatus and the frequency division multiplex signal receiving apparatus are ignored.

The above-mentioned I ′ signal and Q ′ signal are subjected to a DFT operation in a frequency division multiplex signal receiving apparatus, and are obtained as complex numbers represented by the following equations.

A′ej (+ Wnt + a ′) = Aej (+ Wnt + a) × (Yejd) (59) where A ′ = √ (p′2 + q′2), a ′
= Tan-1 (q '/ p') p 'and q' are received signals. Equation (59) is used to calculate new coefficients S0 and S indicating transmission path characteristics.
By introducing 1, the following determinant can be rewritten.

[0180]

(Equation 42) And, as described later, in the third embodiment,
In order to increase the response speed to the characteristic that changes rapidly due to movement (multipath environment characteristic), the amplitude change Y and phase change d of the transmission system circuit 80 are MY and d + d1, which are high-speed characteristic changes, respectively. Thus, the above equation (59) is expressed by the following equation.

A′ej (+ Wnt + a ′) = Aej (+ Wnt + a) × (Yejd) × (Mejd1) (61) Here, if it is newly set that Mejd1 = V1 + jV2, p ′ + jq ′ = ( p + jq) (S0 + jS1) (V1 +
jV2), which is represented by the following determinant.

[0182]

[Equation 43] Therefore, it is possible to take out a new correction equation separated from the correction equation using the reference signal as it is. Also,
Although equation (62) describes the reference signal, since this also applies to the transmitted normal information signal, (6)
The expression 2) can be expressed as the following expression.

[0183]

[Equation 44] Both the S matrix and the H matrix, which is the inverse of the S matrix, can be obtained by a conventional method. Here, the K matrix, which is the inverse matrix of the V matrix, can be expressed by the following equation.

[0184]

[Equation 45] Since the K matrix of equation (64) is updated at all frequencies for each symbol, it cannot be realized with a large amount of calculation. However, since equation (64) is a matrix with 2 rows and 2 columns, it is not sufficient. realizable.

As a processing means, the H matrix is updated every time a reference signal arrives as in the past. At the time of updating the H matrix, the K matrix is a unit matrix. Thereafter, with respect to the first correction information corrected by the H matrix, a difference from a desired signal point arrangement, that is, a high-speed change in amplitude and phase is detected for each symbol. , Updated for each symbol. The updated K matrix is used for the next symbol.

As described above, in this embodiment, the characteristic that changes at a high speed by a method similar to the correction using the second correction equation described in the second embodiment (multipath environment characteristic).
Therefore, the description of the flow for each symbol will be omitted.
However, the H matrix and the K matrix in this embodiment are different from the second embodiment in that they have 2 rows and 2 columns. The K matrix is composed of K0 and K1 in the equations (44a) and (44b).

In the third embodiment, the transmitting apparatus having the configuration shown in FIG. 1 inserts a symbol number into the first carrier,
The first carrier, that is, reference data used only for correcting the symbol number is inserted into the m1th carrier as a reference signal. The symbol number is sequentially increased for each symbol period and goes around. The symbol number counting circuit 5 is represented by 9 bits, and 0, 1, 2, 3,. . . , 511, 0, 1,
2. . . And the number of the circulating symbol is counted and output. Of these, of the 9 bits representing the symbol number, 9, 8,
Fourth bits of the third and second bits are modulated by 16QAM and transmitted and received by a predetermined carrier.

In this embodiment, reference signal insertion circuit 6 in FIG. 1 inserts a reference signal based on the upper 8 bits of 9 bits representing the symbol number. In order to ignore the least significant bit, that is, between two symbols, a reference signal is inserted into the same carrier. The reference signal is represented by the following equation.

[0189]

[Equation 46] Here, X and Y in Expression (65) are known reference signal values.

On the other hand, the frequency division multiplexed signal receiving apparatus in this embodiment has substantially the same configuration as FIG.
The corrected transmission signals R and I are obtained by using the FFT and QAM decoding circuit indicated by 31 in the block diagram of FIG. 12 as indicated by 31 ″ in FIG. 12. In FIG. 12, the same components as those in FIG. 12, the received I signal and the received Q signal extracted from the signal processing unit 80 in FIG.
The signal is supplied to the FT symbol number decoding circuit 311, where the symbol number is decoded first, and then the received reference signal value of the set of positive and negative carriers corresponding to the symbol number is obtained.

In this embodiment, since the predetermined 4 bits of the symbol number are subjected to 16 QAM, the symbol number can be decoded with a better error rate than other transmission information, and 0 to 0 represented by 9 bits. Symbol numbers up to 511 can be reliably decoded, and the insertion carrier of the transmission information reference signal can be reliably specified for each symbol. This received reference signal value is p0 in the 0th symbol (even symbol).
In s 'and q0s', the first symbol (odd symbol) is p1s 'and q1s'.

First, the correction processing circuit 317 of FIG.
Based on equation (0), coefficients S0 and S1 representing the channel characteristics expressed by the following equation are calculated by the 0th symbol and the 1st symbol.

S0 = (Xp0s ′ + Yq0s ′) / (X2 + Y2), S1 = (Xq0s ′ + Yq0s ′) / (X2 + Y2) (66) Alternatively, S0 = (Xp1s ′ + Yq1s ′) / (X2 + Y2), S1 = (Xq1s) (−Yp1s ′) / (X2 + Y2) (67) Thereafter, the coefficients S0 and S1 are respectively averaged to remove white noise. In this embodiment, an average between two symbols is obtained.

Since there are two coefficients, S0 and S1, they can be obtained by transmitting and receiving one type of reference signal. Of course,
Since the reference signal is known in the receiving apparatus, any reference signal may be used as long as the coefficient can be obtained. In this way, the correction processing circuit 317 sets the transmission path characteristic coefficient S0
And S1 are first detected, and then, based on the transmission path characteristic coefficients S0 and S1 and the symbol number, an average correction formula (first correction formula) for the corresponding carrier is derived and stored based on the following formula.

[0195]

[Equation 47] Here, in the equation (68), H0 = S0, H1 = S1, de
tA = S02 + S12.

This first correction equation is provided for each carrier (carrier). Here, since 257 carriers are used, about 256 first correction equations are sequentially generated and updated every moment. . Since the first correction equation is obtained by averaging one corresponding carrier with two symbols, the interval at which the average correction equation of the same carrier is updated next is 51
Two symbols later (2 symbols × 256 = 512 symbols).

Instead of calculating the inverse matrix after detecting the transmission path characteristics as described above, a correction formula may be calculated directly from the following formula as another method.

[0198]

[Equation 48] The first correction circuit 318 performs the following operation on the reception information from the FFT symbol number decoding circuit 311 using the first correction expression derived and held by the correction processing circuit 317, and corrects the information. Output a signal.

[0199]

[Equation 49] Naturally, the counts S0 and S1 are 512
Since it changes for each symbol, the above-described first correction equation also changes every moment for every 512 symbols. The correction itself using the first correction formula has been known in the past, and it is thereby possible to correct a characteristic on a transmission line that changes relatively slowly, such as a change over time or a change in temperature.

As in the second embodiment, the second correction circuit 316 generates a K matrix which is a unit matrix for the next symbol to which a reference signal has been transmitted / received, and generates a K matrix as a unit matrix for other symbols. The corrected data <a> and <b> are generated by the following equation using the K matrix thus obtained, and the decoded real part data R and imaginary part data I are generated from this and output to the output circuit 32.

[0201]

[Equation 50] The second correction equation calculation and holding circuit 315 includes the first correction circuit 3
The correction signals a ″ and b ″ output after being corrected and calculated based on the equation (70) by 18, the symbol numbers output from the correction processing circuit 317, and the decoded data a obtained from the second correction circuit 316. Based on b, a K matrix represented by the following equation is generated and stored.

[0202]

(Equation 51) Here, K0 in the equation (72) is a value represented by the equation (44a), and K1 is a value represented by the equation (44b), which is an error between the decoded data p and q and a desired signal point arrangement. 7 is a second correction equation corresponding to the transmission path characteristic of the high-speed change component of the transmission path based on the following equation.

As described above, the third embodiment also applies the second correction to the signal corrected based on the first correction equation.
Since the correction operation is performed by the second correction equation corresponding to the high-speed change characteristic similar to that of the embodiment, the characteristic that changes relatively quickly such as a multipath environment generated in mobile communication is corrected. Optimal decoded data can be obtained for each symbol.

Next, the operation of the correction method using the second correction equation will be described using specific numerical examples. Regarding the H matrix, the operation of the K matrix will be described in detail by simply expressing the description and omitting the description.

First, at a certain symbol n, the same (7.5, 7.5) as in the second embodiment is transmitted as a predetermined reference signal value (real part, imaginary part). It is assumed that the reference signal is received at a value of (6.25, 6.25). At this time, the H matrix does not include an error in the transmission system,
The phase characteristic is not changed, and the amplitude characteristic is represented by the above equation (53) showing a state of being multiplied by 6.25 / 7.5.

In the first-stage correction using the first correction equation, correction is performed using this H matrix, and thereafter, no change is made until a new reference signal arrives. Equation (53) is the first correction equation held by the first correction equation derivation holding circuit 313. At this point, the K matrix is the same unit matrix as shown in Equation 35.

When the received transmission information is a '= 6.10 and b' = 5.30 in the next symbol n + 1, as described above, the correction signals <a> = 7.32 and <b> = 6 .
36 is obtained, the second correction circuit 316 determines that the received data is the data transmitted in (7.5, 6.5), and outputs the real part data and the data to the output circuit 32 shown in FIG. (7, 6) is passed as reception data as a set (R, I) of decoded data of an imaginary part. Note that 0.5 is a bias value added to simplify QAM decoding.
This is a conventionally known method.

Finally, the second correction formula deriving and holding circuit 315
Calculates a new K matrix represented by the equation (54) based on the equations (43) and (44), and substitutes it for the next symbol n
It is derived and held as a second correction equation for +2. The newly calculated K matrix is represented by the following equation. This K matrix represents a change in the amplitude characteristic and the phase characteristic generated in the transmission path from the transmission of the symbol n to the transmission of the symbol n + 1. Hereinafter, the same operation as described above is repeated, and the same result as in the second embodiment is obtained.

Next, each modified example of the third embodiment will be described. Also in the third embodiment, the following modifications similar to those described in the second embodiment can be considered.

(Modification 1) The frequency at which the reference signal is inserted is not limited to one set, but may be inserted into several sets. In the description of the third embodiment, the coefficients S0 and S1 obtained by allocating the reference signal to the same frequency among a plurality of symbols are used.
Were averaged, that is, the transmission path system from which Gaussian noise was removed was detected, and the first correction equation was used.

The second correction equation can also be averaged to remove Gaussian noise. Here, a method of averaging the K matrix over five symbols will be described.

It is assumed that a reference signal is transmitted and received in symbol n. As initial values, K01 = 1, K02 = 1,
K03 = 1, K04 = 1, K05 = 1, K11 = 0, K12
= 0, K13 = 0, K14 = 0, and K15 = 0. Then K0 = (K01 + K02 + K03 + K04 + K05) / 5 K1 = (K11 + K12 + K13 + K14 + K15) / 5 (73) For the symbol n + 1, the second average correction is performed by Expression (73). The method for obtaining a new K matrix is as described above. The coefficients obtained here are designated as K00 and K10.
Next, K0m is replaced with K0m + 1, and K1m is replaced with K1m + 1, and a second correction equation (averaged K matrix) averaged by equation (73) is generated (where m = 0, 1, 2,
3, 4).

By repeating such processing, the coefficients between the latest five symbols are held, averaged, and used. This averaging process is performed by the second correction formula deriving and holding circuit 315.

(Modification 2) Equations (71) and (72)
When calculating the K matrix by the equations (44a) and (44b),
In the error signals at a ″ and b ″ (that is, the difference between the signal a and the signal a ″ for the signal a ″ and the difference between the signal b and the signal b ″ for the signal b ″), A lower limit is set, and if it is higher or lower, the K matrix is replaced with a predetermined value. Specifically, in either the real part or the imaginary part, when the error signal value becomes 0.4 or more or -0.4 or less, 0.4 or -0.4 is set, respectively. Substitute the K matrix.

Second correction (or second average correction)
In the case where a change larger than the high-speed change that can be followed by the above is generated, or when the S / N is extremely deteriorated, an error signal is turned back and the second correction formula may be reversely corrected. If averaging is performed, the effect is small, but if the averaging occurs continuously, an error will also occur. The above-described limit value is provided to avoid the reverse correction. This restriction process is performed by the second correction formula derivation holding circuit 315.

(Modification 3) Equations (71) and (72)
When calculating the K matrix by the equations (44a) and (44b),
A predetermined weight is applied to the error signals at a ″ and b ″. This is particularly effective when the S / N is extremely deteriorated.

When the signal-to-noise ratio is extremely deteriorated, the error signal may be turned back, and the second correction equation may be inversely corrected. If averaging is performed, the effect is small, but if the averaging occurs continuously, an error will also occur. However, it can be considered that the error signal has variation due to the influence of Gaussian noise. Therefore, the distribution may be normal. Therefore, the K matrix is generated by increasing the weight of the portion close to the median value, increasing the influence, and reducing the weight of the peripheral portion to reduce the influence. Such weighting is performed by the second correction formula deriving and holding circuit 315.

(Modification 4) The second correction (or the second correction)
In the case where a change equal to or more than the high-speed change that can be followed by the average correction is performed, or when the S / N is extremely deteriorated, the error signal is folded back and the second correction formula may be inversely corrected. The effect is small if averaging is performed,
If they occur consecutively, an error will still occur.

Therefore, in this modified example, when an error is detected by an error correction circuit not described above, even if the K matrix as the second correction formula is generated, the inverse correction is performed. Therefore, the K matrix is not generated. Substitute with an identity matrix and complement with the averaging effect.

(Modification 5) Equations (71) and (72)
When calculating the K matrix by the equations (44a) and (44b),
In the signal point arrangement in a> and <b> (integer part),
A predetermined upper limit value and a lower limit value are set, and in the case of a signal point arrangement of more or less than that, the K matrix is replaced with a unit matrix. For example, as shown in FIG. 10, in 256QAM, in a signal point arrangement where the absolute value of the real part or the imaginary part is 8 or more, or the absolute value of the real part + the imaginary part is 6 or more, the error signal is not used. Replace with an identity matrix. Since the high-speed change affects the signal point arrangement on the outer peripheral side, there is a high probability that an error also occurs on the outer peripheral side. Therefore, in this modification, only the more reliable error signal of the signal point arrangement on the inner peripheral side, which is indicated by a black circle in FIG. 10, is used.

(Modification 6) In this modification, averaging processing is performed on the frequency axis. In this modification, after calculating each coefficient, a predetermined amount of low-pass filter is inserted into a sequence in the order of the frequency axis to remove high-frequency components. Such filter processing can be easily performed by digital filtering in the correction processing circuit 317 or the second correction formula derivation and holding circuit 315. This filtering may be configured by a two-dimensional filter using not only one-dimensional operation on the frequency axis but also the above-described arrangement of coefficients on the time axis.

The two-dimensional filter on the frequency axis and the time axis will be briefly described. The aforementioned H matrix is averaged by transmitting and receiving the reference signal for a plurality of symbols. Also, the K matrix is averaged among a plurality of symbols. As a result, averaging on the time axis is performed.
Then, an E matrix similar to the above equation (57) is newly generated, and among the coefficients of the E matrix, the coefficient of the own carrier and the coefficient in the vicinity (for example, ± 10 carriers) are averaged. Of course,
Similar averaging is performed for all carriers. Thereby, the average on the frequency axis is obtained. Then, a correction calculation is performed by the following equation.

[0223]

(Equation 52) Up to now, two-stage calculation has been performed using the H matrix and the K matrix, respectively. In this case, the process of obtaining the H matrix and the K matrix is the same. to correct.

As another example, a set is formed for each of three adjacent carriers, a reference signal is inserted only into the center carrier, and a correction formula deriving process is performed. It is also possible to correct by a correction formula.
As a result, the amount of calculation can be reduced and the arrival period of the reference signal can be shortened, so that the device can be reduced in cost and high-speed followability can be achieved.

Next, a third embodiment of the present invention will be described. The first and second embodiments described above and the modifications thereof are all symmetrical with respect to the center carrier of the OFDM signal. In other examples, a set is formed for every three adjacent carriers, and only the center carrier is used. It is also possible to perform a process of deriving a correction formula by inserting a reference signal, and to correct the carrier on both sides thereof by the correction formula of the center carrier. As a result, the amount of calculation can be reduced and the arrival period of the reference signal can be shortened, so that the device can be reduced in cost and high-speed followability can be achieved.

[0226]

As described above, according to the present invention,
A set of one positive carrier and one negative carrier, which are symmetrical to each other and have a crosstalk effect on the center carrier of the plurality of carriers, are referred to by these carriers. Since the signal is transmitted, the channel characteristics can be detected quickly, and the channel characteristics of all the carriers can be detected by switching the set of positive and negative carriers into which the reference signal is inserted at regular intervals.

Thus, according to the present invention, the aforementioned I
The frequency characteristics of the signal and the Q signal itself can be corrected in the same manner as in the prior art. In addition to this, the relative amplitude characteristic difference between the I signal and the Q signal at each frequency of the transmitting device, and the I Signal, the relative phase characteristic difference between the Q signals, the orthogonality error of the quadrature modulator of the transmitting device, the relative amplitude characteristic difference between the I signal and the Q signal at each frequency of the receiving device, Since characteristic errors such as a relative phase characteristic difference between the I signal and the Q signal and a quadrature error of the quadrature demodulator of the receiving device can be corrected, the I signal and the Q signal can be corrected at each frequency of these transmitting devices. , The I signal at each frequency of the transmitter,
Relative phase characteristic difference between Q signals, quadrature error of quadrature modulator of transmitting device, I signal, Q at each frequency of receiving device
Even if the relative error characteristic difference between the signals, the relative phase characteristic difference between the I signal and the Q signal at each frequency of the receiving device, the characteristic error such as the orthogonality error of the quadrature demodulator of the receiving device need not be improved. In addition, it is not necessary to consider these changes in temperature and changes with time, so that a higher quality orthogonal frequency division multiplexed signal can be transmitted as compared with the related art.

Further, according to the present invention, by using a known reference signal, the decoded data can be changed with time or temperature using a first correction equation consisting of a correction coefficient obtained accurately. Compensate for errors and characteristics that change relatively slowly, such as changes, and then, for this corrected data, receive a reference signal and then receive the next reference signal to calculate a new correction equation The transmission path characteristic corresponding to the change in the period up to is corrected based on the second correction equation obtained by detecting the transmission path characteristic. Can be corrected, and optimization for each symbol can be achieved.

Further, according to the present invention, a reference signal of a known value transmitted as a set of a positive carrier and a negative carrier symmetrical to each other with respect to a center carrier of a plurality of carriers by using the first correction equation. And by sequentially performing the correction by the first correction formula and the correction by the second correction formula, it is possible to perform a high-precision correction corresponding to a change in the transmission path characteristic, More accurate decoding can be performed.

Further, according to the present invention, by performing averaging in one or both of the first and second correction equations, Gaussian noise can be removed, and a more reliable correction coefficient can be derived. Thus, the accuracy of correction can be improved.

Further, according to the present invention, a predetermined upper limit value and a predetermined lower limit value are provided in the second correction equation, predetermined weighting is performed according to the error signal, and the second correction formula is used when the error correction circuit detects an error. By stopping the updating of the correction formula and substituting the second correction formula with a unit matrix, or providing an upper limit value and a lower limit value to a correction signal used for calculating the second correction formula, When the (average) correction formula 2 is calculated, it is possible to avoid a reverse correction that may not be able to follow a high-speed change or may occur due to an extremely low S / N.

Further, according to the present invention, the number of times of averaging on the time axis is reduced, and averaging is performed on the frequency axis, thereby realizing a device with higher speed and higher reliability.

[Brief description of the drawings]

FIG. 1 is a block diagram of a first embodiment of a transmission device of the present invention.

FIG. 2 is a diagram showing a frequency spectrum of an example of an OFDM signal transmitted in the present invention.

FIG. 3 is a block diagram of an embodiment of the receiving device of the present invention.

FIG. 4 is a block diagram rewritten focusing on an I signal and a Q signal transmitted in the first embodiment of the system of the present invention.

FIG. 5 is a block diagram of a first embodiment of a main part of FIG. 3;

FIG. 6 is a flowchart of an embodiment when the FFT operation of FIG. 3 is performed by software.

FIG. 7 is a block diagram of a receiving apparatus according to a second embodiment of the present invention.

FIG. 8 is a block diagram rewritten focusing on an I signal and a Q signal transmitted in a second embodiment of the system of the present invention.

FIG. 9 is a block diagram of a main part of FIG. 7 according to a second embodiment;

FIG. 10 is a diagram illustrating an example of a signal point arrangement used for generating a K matrix.

FIG. 11 is a block diagram rewritten focusing on an I signal and a Q signal transmitted in a third embodiment of the system of the present invention.

FIG. 12 is a block diagram of a third embodiment of a main part of the system of the present invention.

[Explanation of symbols]

 Reference Signs List 2 input circuit 3 clock divider circuit 4 operation unit 5, 14 symbol number counting circuit 6 reference signal insertion circuit 7 output buffer 8 D / A converter / low-pass filter (LPF) 9 quadrature modulator 10, 16, 25 intermediate frequency Oscillator 11, 26 90 ° shifter 12, 22, Frequency converter 15 Digital quadrature modulator 17 D / A converter / bandpass filter (BPF) 24 Carrier extraction and quadrature demodulator 27 Synchronization signal generation circuit 31, 31 ', 31 "FFT , QAM decoding circuit 41, 42, 72 Signal processing unit 43, 44, 47, 48, 74, 75 Multiplier 46, 73, 80 Transmission system circuit 49, 50, 76, 77 Signal processing unit 311 in receiving device FFT symbol number decoding circuit 312 Transmission path characteristic detection circuit 313 First correction formula derivation and holding circuit 314 First correction circuit 315 2 correcting equation deriving and storing circuit 316 the second correction circuit 317 correction processing circuit 318 first correction circuit

Claims (14)

[Claims] The transmitting side separately modulates each of a plurality of carrier waves having different frequencies with an in-phase signal and a quadrature signal obtained from an information signal to be transmitted allocated to each carrier, respectively, and After generating a division multiplexed orthogonal frequency division multiplexed signal and transmitting it in symbol units, the receiving side receives the orthogonal frequency division multiplexed signal and demodulates each modulated carrier into an in-phase signal and an orthogonal signal, respectively. In the orthogonal frequency division multiplexing signal transmission method for decoding a signal, on the transmitting side, a known combination of a high-side positive carrier and a low-side negative carrier symmetrical to a center carrier among the plurality of carriers is known. A reference signal is transmitted, and a set of positive and negative carriers for transmitting the known reference signal is designated by a specific carrier or a symbol number transmitted by a transmission parameter, and Transmit cyclically changed for each period, on the receiving side, decode the reference signal based on the symbol number from the received frequency division multiplexed signal, from the reference signal real part of the positive carrier and An imaginary part, a transmission path characteristic is detected based on a leak component to each of the real part and the imaginary part of the negative carrier, a correction equation is calculated from the detected transmission path characteristic and stored, and the correction equation is calculated. A quadrature frequency division multiplexing signal transmission system, wherein the in-phase signal and the quadrature signal decoded using the signal are corrected. 2. The reference signal comprises a first set of reference signals and a second set of reference signals, which are alternately switched in a predetermined symbol unit and transmitted as a set of positive and negative carrier waves. When the values of the real and imaginary parts of the transmitted reference signal are p and q, respectively, and the values of the real and imaginary parts of the reference signal transmitted on the negative carrier are r and u, respectively, The values of the real part and the imaginary part of the reference signal of the positive carrier are p ′ and q ′, respectively, and the values of the real part and the imaginary part of the received reference signal of the negative carrier are r ′ and u ′, respectively. At this time, the receiving side uses the following equation as the transmission path characteristic: The coefficients S0 to S7 are calculated from the known reference signal for the received reference signal, and the following equation is used as the correction equation. (However, H0 = + S0 (S6S6 + S7S7) -S2 (S4S6 + S5S7) + S3 (S4S7-
S5S6), H1 = + S1 (S6S6 + S7S7) -S3 (S4S6 + S5S7) -S2 (S4S7-S5S
6), H2 = + S4 (S2S2 + S3S3) -S6 (S0S2 + S1S3) + S7 (S0S3-S1S2), H
3 = + S5 (S2S2 + S3S3) -S7 (S0S2 + S1S3) -S6 (S0S3-S1S2), H4 = + S
2 (S4S4 + S5S5) -S0 (S4S6 + S5S7) -S1 (S4S7-S5S6), H5 = + S3 (S4
S4 + S5S5) -S1 (S4S6 + S5S7) + S0 (S4S7-S5S6), H6 = + S6 (S0S0 + S
1S1) -S4 (S0S2 + S1S3) -S5 (S0S3-S1S2), H7 = + S7 (S0S0 + S1S1)
-S5 (S0S2 + S1S3) + S4 (S0S3-S1S2), det A = S0 × H0 + S1 × H1
+ S4 × H2 + S5 × H3) is stored and held, and the following equation is used for each set of positive and negative carriers using the above-mentioned correction equation. (Where a and b are the real and imaginary parts of the corrected received data assigned to the positive carrier frequency,
c and d are the real part and the imaginary part of the corrected received data assigned to the negative carrier frequency, a ′ and b ′ are the real part and the imaginary part of the received data of the positive carrier frequency, c ′
And d 'are the real part and the imaginary part of the received data of the negative carrier frequency) and the corrected received data a, b, c
2. The orthogonal frequency division multiplexed signal transmission system according to claim 1, wherein d and d are obtained. 3. The reference signal comprises a first set of reference signals and a second set of reference signals, which are alternately switched in a predetermined symbol unit and transmitted as a set of positive and negative carrier waves. Of the set, each value of the real part and the imaginary part transmitted on the positive carrier and the real part and the imaginary part transmitted on the negative carrier, the first set of reference signals is a real number transmitted on the positive carrier. Only the value of the part or the imaginary part is a predetermined value, and the other values are each set to zero, and the second set of reference signals is a predetermined value only the value of the real part or the imaginary part transmitted by the negative carrier. 2. The orthogonal frequency division multiplexing signal transmission system according to claim 1, wherein each of the other values is set to zero. 4. The reference signal comprises a first set of reference signals and a second set of reference signals, which are alternately switched in a predetermined symbol unit and transmitted as a pair of positive and negative carriers, and Of the set, each value of the real part and the imaginary part transmitted on the positive carrier and the real part and the imaginary part transmitted on the negative carrier, the first set of reference signals is a real number transmitted on the positive carrier. The value of the part and the imaginary part are each a predetermined value, and the other values are each set to zero, and the second set of reference signals is such that the values of the real part and the imaginary part transmitted on the negative carrier are each a predetermined value. 2. The orthogonal frequency division multiplexing signal transmission system according to claim 1, wherein each of the other values is set to zero. 5. A digital information signal to be transmitted represented by a complex number is input to each of a plurality of real part input terminals and an imaginary part input terminal and subjected to inverse discrete Fourier transform, and transmitted by a plurality of carrier waves having different frequencies from each other. An operation unit for generating the in-phase signal and the quadrature signal to be transmitted in symbol units; generating a symbol number whose value changes for each symbol of the digital information signal to be transmitted and inputting it to a specific input terminal of the operation unit A symbol number counting circuit, a real part input terminal and an imaginary part input terminal of the arithmetic unit to which a digital information signal transmitted on a high-frequency positive carrier symmetrical to a center carrier of the plurality of carriers is input; And a reference signal to a set of a real part input terminal and an imaginary part input terminal of the arithmetic unit to which the digital information signal transmitted by the low-frequency side negative carrier is input. And a reference signal inserting means for switching a set of the real part input terminal and the imaginary part input terminal for inputting the reference signal at regular intervals, and an output for temporarily storing an output in-phase signal and a quadrature signal of the arithmetic unit. A buffer, a conversion unit for converting the output in-phase signal and the quadrature signal of the output buffer into quadrature frequency division multiplex signals, and a transmission unit for transmitting the quadrature frequency division multiplex signal. Multiplex signal transmitter. 6. A digital information signal to be transmitted on a plurality of carriers having different frequencies from each other, and a high-side positive carrier and a low-side negative carrier that are symmetrical with respect to a center carrier among the plurality of carriers. A known reference signal is inserted into the set of carrier waves, and the set of positive and negative carrier waves for transmitting the reference signal is designated by a symbol number to be transmitted on a specific carrier wave, and is sequentially and cyclically changed at regular intervals. Receiving means for receiving an orthogonal frequency division multiplexed signal, comprising: orthogonally demodulating a received orthogonal frequency division multiplexed signal from the receiving means; an in-phase signal and an orthogonal signal each represented by a complex number; and the symbol number and a reference signal. And a digital information signal by discrete Fourier transforming the in-phase signal and the quadrature signal, the symbol number and the reference signal from the demodulation means, respectively. Decoding means for decoding the symbol number and the reference signal, and decoding the reference signal from the symbol number from the decoding means, and real and imaginary parts of the positive carrier and the negative carrier from the reference signal. Detecting means for detecting a transmission path characteristic based on a leak component to each of a real part and an imaginary part of the correction part; and calculating and storing a correction equation for calculating and storing a correction equation from the transmission path characteristic detected by the detecting means. Means, and a correction circuit for correcting the decoded signals of the in-phase signal and the quadrature signal from the decoding means using the stored correction formula. 7. The reference signal comprises a first set of reference signals and a second set of reference signals, which are alternately switched in a predetermined symbol unit and transmitted as a set of positive and negative carrier waves, and wherein the detecting means comprises: The values of the real part and the imaginary part of the reference signal transmitted on the positive carrier are p and q, respectively, and the values of the real part and the imaginary part of the reference signal transmitted on the negative carrier are r and u, respectively. At some point, the values of the real and imaginary parts of the received positive carrier reference signal are p ′ and q ′, respectively.
When the values of the real part and the imaginary part of the received reference signal of the negative carrier are r 'and u', respectively, the transmission path characteristic is given by the following equation. Is a means for calculating the coefficients S0 to S7 represented by the following equation by dividing the received reference signal by the known reference signal. The correction formula calculation and holding means uses the following formula as the correction formula: (However, H0 = + S0 (S6S6 + S7S7) -S2 (S4S6 + S5S7) + S3 (S4S7-S
5S6), H1 = + S1 (S6S6 + S7S7) -S3 (S4S6 + S5S7) -S2 (S4S7-S5S
6), H2 = + S4 (S2S2 + S3S3) -S6 (S0S2 + S1S3) + S7 (S0S3-S1S2), H
3 = + S5 (S2S2 + S3S3) -S7 (S0S2 + S1S3) -S6 (S0S3-S1S2), H4 = + S
2 (S4S4 + S5S5) -S0 (S4S6 + S5S7) -S1 (S4S7-S5S6), H5 = + S3 (S4
S4 + S5S5) -S1 (S4S6 + S5S7) + S0 (S4S7-S5S6), H6 = + S6 (S0S0 + S
1S1) -S4 (S0S2 + S1S3) -S5 (S0S3-S1S2), H7 = + S7 (S0S0 + S1S1)
-S5 (S0S2 + S1S3) + S4 (S0S3-S1S2), det A = S0 × H0 + S1 × H1
+ S4 × H2 + S5 × H3) is a means for storing and holding the value calculated by the following formula: (Where a and b are the real and imaginary parts of the corrected received data assigned to the positive carrier frequency,
c and d are the real part and the imaginary part of the corrected received data assigned to the negative carrier frequency, a ′ and b ′ are the real part and the imaginary part of the received data of the positive carrier frequency, c ′
And d 'are the real part and the imaginary part of the received data of the negative carrier frequency) and the corrected received data a, b, c
7. An orthogonal frequency division multiplexed signal receiving apparatus according to claim 6, wherein the receiving means is a means for obtaining the following. 8. The transmitting side separately modulates an information signal to be transmitted as a signal comprising a real part and an imaginary part to generate an in-phase signal and a quadrature signal. Of a plurality of carriers having different frequencies from each other, a plurality of positive carriers and a plurality of negative carriers on the high band side with respect to the center carrier are modulated, and an orthogonal frequency division multiplexed signal is generated by frequency division multiplexing. In the orthogonal frequency division multiplexing signal transmission method of decoding the information signal after receiving the orthogonal frequency division multiplexing signal on the receiving side and demodulating each modulated carrier into an in-phase signal and an orthogonal signal, respectively, The transmitting side transmits a known reference signal on a predetermined carrier out of the plurality of carriers, and the predetermined carrier is sequentially and cyclically changed at regular time intervals and transmitted. The reference signal is decoded from a frequency division multiplexed signal, and a transmission path characteristic is detected based on a change component of a real part and an imaginary part of the predetermined carrier from the reference signal, and a first transmission path characteristic is detected from the detected transmission path characteristic. A correction formula is calculated and stored, and the decoded information signal is corrected using the first correction formula. Based on a difference between the corrected information signal and a predetermined signal point arrangement, a transmission path Detecting a high-speed change component, calculating and storing a second correction formula from the detected high-speed change component of the transmission path, and further correcting the corrected information signal using the second correction formula. An orthogonal frequency division multiplexing signal transmission method characterized by the above-mentioned. 9. The transmitting side specifies the specific carrier modulated by the known reference signal based on a symbol number or specific parameter information or synchronization symbol information transmitted on a predetermined carrier, and 9. The orthogonal frequency division multiplexed signal transmission system according to claim 8, wherein the transmission is performed by sequentially changing the transmission cyclically every predetermined period. 10. A coefficient is arranged in order of frequency with respect to each coefficient constituting a correction expression obtained by combining at least one of the first correction expression and the second correction expression, and a low-pass filter is inserted on a frequency axis. 9. The orthogonal frequency division multiplex signal transmission system according to claim 8, wherein the filter processing is performed. 11. The transmitting side uses a plurality of symbols to convert a reference signal known as a set of a high-side positive carrier and a low-side negative carrier symmetrical to a center carrier among the plurality of carriers. Continuously transmit, the receiving side averages the leakage components to the real part and the imaginary part of the positive carrier and the real part and the imaginary part of the negative carrier from the plurality of received reference signals. 9. The orthogonal frequency multiplexed signal transmission system according to claim 8, wherein the first correction formula is averaged based on a plurality of transmission path characteristics detected from the leakage component. 12. The receiver according to claim 1, wherein the information signal after the correction using the first correction formula is a plurality of symbol periods having a difference from a predetermined signal point arrangement obtained for each symbol. The fast change component of the transmission line detected from the average value, or the average value of the high speed change component of the transmission line detected from the difference from a predetermined signal point arrangement obtained for each symbol in the plurality of symbol periods. 9. An averaged second correction expression is calculated, and the corrected information signal is further corrected using the averaged second correction expression.
12. The orthogonal frequency division multiplex signal transmission system according to any one of claims 11 to 11. 13. The receiving side transmits information signals of the positive and negative pairs of carrier waves corrected using the first correction formula as (a ″ + jb ″) and (c ″ + jd ″). The information signals of the positive and negative pairs of carrier waves are (a + jb), (c + j
d), the following equation is obtained. (However, K0 = (aa "+ bb") / (a "2 + b")
2), K1 = (ab "-a" b) / (a "2 + b" 2),
K6 = (cc "+ dd") / (c "2 + d" 2), K7 =
9. The orthogonality according to claim 8, wherein a matrix represented by (cd "-c" d) / (c "2 + d" 2)) is generated as the second correction expression and used for the next symbol. Frequency division multiplex signal transmission method. 14. The method according to claim 8, wherein the smaller the absolute value of the error signal in the information signal corrected using the first correction formula, the larger the weight is, and the second correction formula is generated. The orthogonal frequency division multiplexed signal transmission system described in the above.