JP3084368B1 - OFDM receiver - Google Patents

Abstract

Abstract: PROBLEM TO BE SOLVED: To provide an interpolation method which does not require a multiplier for increasing a circuit scale and is excellent in estimating a channel response as an interpolation method in a time direction when performing equalization in a frequency domain. SOLUTION: A transmission line response estimation value of a pilot signal H ~
(l, k _{p, l} ) and H ~ (l + 4, k _{p, l + 4} ) are added by the adder 21 and shifted by one bit by the bit shift circuit 22 to obtain H ~ (l
+ 2, k _{p, l + 2} ). Further, HＨ (l, k _{p, l} ) and the output of the bit shift circuit 22 are added by the adder 23, and the bit shift circuit 25 performs bit shift to obtain H ~ (l + 1, k _{p, l + 1} ). . Similarly, H ~ (l + 4, k _{p, l + 4} ) and the output of the bit shift circuit 22 are added by the adder 24, and the bit is shifted by the bit shift circuit 26, and H ~ (l + 3, k _{p, l + 3).} ).
This eliminates the need for a multiplier and can reduce the hardware scale. Further, it is only necessary to store two pilot signals in the time direction for each carrier through which the pilot signals are transmitted, and the memory size can be reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to OFDM (Orthogonal Frequency Division Multiplexing).
x) The present invention relates to an OFDM receiver for receiving a signal.

[0002]

2. Description of the Related Art OFDM is a terrestrial digital broadcasting system.
(Orthogonal Frequency Divi
sion Multiplex) method is being studied in Europe and Japan.
In this OFDM system, a large number of carriers are multiplexed and transmitted within the band of one channel (2 in the European DVB-T system).
1705 books in K mode, Reference 1: EBU / ETSI JTC: Digita
l Broadcasting system for television, sound and da
ta services; Framing structure, channel coding mod
ulation for digital terrestrial television, ETS 300
744, Mar. 1996). In the DVB-T system, a pilot signal whose amplitude and phase are known is transmitted in this subcarrier.

When a mobile receives an OFDM signal, the received signal undergoes Rayleigh fading in which the amplitude is distributed in a Rayleigh manner and the phase is uniformly distributed. As a technique for compensating for this effect, by interpolating the pilot signal in the time direction and observing the fluctuation between the pilot signals, the amplitude of the received signal, the transmission path response due to the temporal fluctuation of the phase is estimated,
There is a method of equalizing in the frequency domain based on the estimation result. This method has already been reported by academic societies (for example,
Literature 2: ITE Journal Vol.52, No.11 1998, Takada et al., "Transmission Characteristics of OFDM Symbol Length and Scattered Pilot in Digital Terrestrial Broadcasting", Literature 3: Wireless pers
onal Communications 2: 335-356, 1996, Hara "Transm
ission Performance Analysis of Multi-Carrier Modul
ationin Frequency Selective Fast Rayleigh Fading C
hannel ").

On the other hand, as the fading becomes faster,
Doppler spread of carriers cannot be ignored, and inter-carrier interference (ICI) occurs. In other words, this is equivalent to the fact that the channel response fluctuates within one OFDM symbol. As a technique for compensating for the transmission line response fluctuation within one OFDM symbol, a technique based on time domain equalization has been reported at academic conferences (Reference 4: 1997 IEICE, B5).
-282, Hashizume et al. "High-speed fading compensation method using guard interval in orthogonal multicarrier modulation").

First, equalization in the frequency domain will be described.

FIG. 13 shows a pilot signal arrangement of the European DVB-T system. k is the carrier number (carrier in
dex), and 1 is a symbol index. As the pilot signal, a CP (continuous pilot) continuously transmitted on the same carrier regardless of the transmission symbol number and an SP (distributed pilot scatte) transmitted on a different carrier for each transmission symbol
red pilot). In FIG. 20, SP indicates a carrier position k _{p, l} = 12 every 12 lines as indicated by a black circle symbol.
p + 3 * (l mod 4), p = 0,..., 142, l =
0,..., 67 (the number of symbols in one frame) and are cyclically inserted so that the same carrier arrangement is obtained every four symbols (however, k _{p, l}
Is 1704). The data signals are CP and S
Transmitted on 1512 carriers excluding P, the carrier arrangement is k _{d, l} , d = 0,..., 1511, l = 0,
, 67 (where k _{d, l} ≠ k _{p, l} ).

On the other hand, when observed in the symbol direction (time direction), it can be seen that the SP is transmitted every four symbols at the position of a multiple of 3 in the carrier number. Therefore, when performing frequency domain equalization, it is necessary to interpolate the channel response for three data signals between pilot signals.

FIG. 14 shows the equalization in the frequency domain.
This will be described with reference to FIG. FIG. 14 shows a configuration of a baseband demodulation unit including a frequency domain equalizer. A received OFDM signal Y (l, k) converted into a frequency domain signal by an FFT (fast Fourier transform) circuit 11 is as follows. Receive data signal Y
(l, k _{d, l} ) and the received pilot signal Y (l, k
_{p, l} ). The pilot signal has a known complex amplitude X (l, kp _{, l} ) and is stored in the memory 12 in advance. Therefore, when noise is not added in the transmission path, the received pilot signal Y (l, kp _{, l} ) is divided by the dividing circuit 13.
Is divided by the complex amplitude X (l, k _{p, l} ) stored in the memory 12 to obtain the transmission path response H (l, k _{p, l} of the subcarrier k = k _{p, l} for transmitting the received pilot signal _{. l} ) = Y (l, k
_{p, l} ) / X (l, kp _{, l} ). However, since noise is generally added in the transmission path, the transmission path response of the carrier transmitting the pilot signal has an estimated value H to (l, k
_{p, l} ).

The estimated value H ~ (l, k _{p, l} ) of the transmission path response passes through a symbol filter 14 for interpolating the transmission path response in the time direction (symbol direction), and then, in a carrier filter 15, each received data signal. channel response H ~ a _(l, k _d, l) were interpolated to the carrier direction, to estimate the channel response in the carriers to transmit data signals. The estimated transmission path response H ~ (l, k _{d, l} ) obtained in this manner is used to receive the received data signal Y (l,
By dividing k _{d, l} ) by the division circuit 16, the data X (l, k _{d, l} ) after equalization = Y (l, k _{d, l} ) / H ~ (l, k _{d , l)} ) Can be obtained.

FIG. 15A shows an example of a transmission path response at a carrier position where a pilot signal is transmitted, and FIG.
Shows a schematic diagram of the pilot and data signal arrangement. In each case, the horizontal axis represents time (symbol direction), and a pilot signal is transmitted every four carriers. In the Rayleigh fading transmission path with the maximum Doppler frequency f _d , each transmission carrier spreads ± f _d , and the received signal fluctuates with time as shown in FIG.

Examples of reports on a method of estimating a channel response in a received data signal between pilot signals include a method of holding the channel response of a pilot signal until the next pilot signal is transmitted (Reference 2). And a method of linearly interpolating between pilot signals (Reference 3).

The method of holding the transmission path response of a pilot signal until the next pilot signal is transmitted and using it as the transmission path response of a data signal between the pilot signals is very simple in terms of hardware, and is time consuming. This is effective for a transmission line whose transmission line response does not fluctuate. However, when the transmission path response fluctuates with time as in the case of a fading transmission path, the characteristics deteriorate. On the other hand, a method of linearly interpolating between pilot signals is effective for a transmission path whose transmission path response fluctuates with time, but requires a multiplier and an adder, and increases the hardware scale.

Next, equalization in the time domain will be described.

As shown in FIG. 16, the OFDM signal has a configuration in which the end of the effective symbol period is added to the beginning of the effective symbol as a guard interval. This OFD
Signal processing such as AFC, timing reproduction, and mode determination is performed using the characteristics of the M signal. In addition, an example has been reported in which the number of carriers is smaller than that in digital broadcasting, but the present invention is applied to equalization of a signal subjected to fading interference (Reference 3).

FIG. 17 shows an example of an OFDM symbol configuration and a transmission line response in the case of receiving fading distortion. OFDM
The symbols are sampled at the timing T of IFFT (Inverse Fast Fourier Transform), and N points are sampled during the effective symbol period. Sample points from the N + 1 N _g + N of the effective symbol period, N _g -1 0 as a guard interval
Is copied to the sample point of

Assuming a time-invariant and noise-free transmission path,
The last part of the effective symbol period and the guard interval period are received as the same signal. However, in a general receiving apparatus, noise is added, and assuming a fading transmission path, as shown in a transmission path response example, each reception sample point is distorted in amplitude and phase, and the end portion of the effective symbol period is distorted. The guard interval period is received as a different signal.

The time domain equalization is a method of estimating a transmission path response from a difference between a received signal in the end portion of an effective symbol period and a received signal in a guard interval period, and performing signal equalization. Now
The received signal at each sample point is given by the following equation.

(Equation 1)

The change in signal between the end of the effective symbol period and the guard interval period is calculated by the following equation.

(Equation 2)

In the above equation, the denominator in Σ represents the received sample points up to k times before the last sample point in the guard interval, and the numerator is the received sample points up to k times before the last sample point of the effective symbol. Represents As described above, by averaging the ratio of a plurality of reception sample points to determine the amount of change in the signal, it is possible to reduce the effect of Gaussian noise independently added to each sample point.

Linear interpolation will be described as a technique for estimating the channel response at the receiving point in the effective symbol from the signal change h. FIG. 18 is a schematic diagram.

In the linear interpolation, the transmission path response of the sample number N _g −1 is set to 1 and the transmission path response of N _g + N is set to h, and the transmission path response of the sample number N _g to N _g + N is linearly interpolated. is there. In this case, the estimated transmission path response for each sample is given by the following equation.

(Equation 3)

With the obtained transmission path responses h ~ _n , the received signal x
By dividing ~ _n , a received signal x ~ _n 'after equalization is obtained.

(Equation 4)

Since the linear interpolation calculates the channel response for each sample point as shown in equation (3), it becomes complicated when hardware is considered.

As described above, in the OFDM method, equalization is performed by interpolating the transmission path response using a pilot signal and dividing the received data by the estimated transmission path response. Further, by using the correlation of the guard interval, equalization in the time domain is also possible. Now, the S / I ratio (signal to interference) of the signal after the received OFDM signal is equalized in the frequency domain.
Is defined as follows.

(Equation 5)

That is, N is the number of FFT points, y _n
Is the n-th sample point in the effective symbol period of the received OFDM signal, and x _k and x ~ _k are the transmission point and the reception point of the k-th subcarrier, respectively, and the S / I is given by the following equation.

[0026]

(Equation 6)

[0027]

As described above, in the conventional receiving apparatus, as an interpolation method for estimating the transmission path response in the time domain in the frequency domain, the transmission path response of the pilot signal is calculated as follows. The method of holding the pilot signal until it is transmitted and using the transmission path response of the data signal between the pilot signals (Reference 2) is very simple in terms of hardware, and is a transmission in which the transmission path response does not fluctuate over time. Although it is effective on a channel, if the transmission channel response fluctuates with time like a fading transmission channel, the above S / S
Deterioration occurs in the I characteristic.

The technique of linearly interpolating between pilot signals (Reference 3) is effective for a transmission path whose transmission path response fluctuates with time, but requires a multiplier and an adder and increases the hardware scale. There is a problem that.

On the other hand, regarding time domain equalization, reference 4
When linear interpolation is performed on a signal within an effective symbol as shown in (1), a multiplier or the like is required to calculate a transmission path response for each sample point, which becomes complicated when hardware is considered. There is such a problem.

Therefore, in the present invention, in order to solve the above-mentioned problem, a multiplier for increasing the circuit scale is not required as an interpolation method in the time direction when performing equalization in the frequency domain, and the transmission path response is excellently estimated. It is an object of the present invention to provide an OFDM receiver using an interpolation method.

[0031]

In order to solve the above-mentioned problems, the present invention has the following characteristic configuration.

[0032]

[0033]

( 1 ) An OFDM signal having a guard period formed by copying the end of the effective symbol period to the beginning of the effective symbol period is received, and the end of the guard period and the end of the effective symbol period of the received OFDM signal are received. from calculates the variation in the channel response, a OFDM receiving apparatus which performs equalization on the time axis by using a variation of the channel response, before Symbol
Within valid symbol is divided into a plurality of blocks, and the channel response estimation means for estimating the amount of change in channel response channel response of each block in steps, based on the plurality of channel response obtained in this way A time domain equalizing means for performing time domain equalization of the effective symbol period.

( 2 ) In the configuration of ( 1 ), the transmission path response estimation detecting means performs an operation of obtaining a plurality of transmission path responses present in the effective symbol from a change in the transmission path response as an adder and a bit shifter. It is characterized by using a circuit device.

( 3 ) The last section of the effective symbol period is copied at the beginning of the effective symbol period to form a guard period, and pilot signals of known amplitude and phase are arranged at substantially equal intervals on the frequency axis and the time axis. An OFDM receiving apparatus for receiving an OFDM signal, comprising: calculating a change in a transmission path response from the end of a guard period and the end of an effective symbol period of the received OFDM signal, and dividing the effective symbol into a plurality of blocks. Time domain equalizing means for dividing and estimating the channel response of each block in steps from the change in the channel response, and performing time domain equalization of the effective symbol period based on the plurality of channel responses. Processing and estimating the transmission path response of the pilot signal as a binary digital signal, and interpolating the transmission path response of the received data signal to perform frequency domain equalization. Characterized by comprising the equalizing means.

( 4 ) In the configuration of ( 3 ), the frequency domain equalizer sequentially extracts first and second pilot signals continuous on the time axis from the reception output of the OFDM signal, and obtains the first and second pilot signals. The first and second transmission path responses of the second pilot signal are added, and the result of the addition is bit-shifted to obtain a third estimated transmission path response. 1st continuous on axis
And interpolating the transmission path response of the data signal between the second pilot signal and the second pilot signal, and performing frequency domain equalization based on the interpolated estimated transmission path response.

( 5 ) In the configuration of ( 3 ), the frequency domain equalization means sequentially extracts pilot signals continuous on the time axis from the reception output of the OFDM signal, and places the pilot signals in the data signal arrangement position between the pilot signals. The transmission line response is interpolated by inserting 0s, arranging them in a time series, and performing a convolution operation, and performing frequency domain equalization based on the interpolated estimated transmission line response.

( 6 ) In the configuration of ( 3 ), the time-domain equalizing means performs an operation of obtaining a plurality of transmission path responses present in the effective symbol from a change in the transmission path response by using an adder and a bit shifter. It is characterized by using a circuit device.

That is, in the interpolation method according to the present invention, in order to estimate the transmission path response of the data signal between the pilot signals (in the time direction), the pilot signals are added and multiplied by 、, and the result is further multiplied by the pilot signal. And add 1/2
, The transmission path response of the data signal is interpolated and estimated. The interpolation circuit includes an adder that adds a pilot signal and a multiplier that multiplies the addition result by a coefficient. In a receiving apparatus that performs signal processing with a binary digital signal, the multiplier has a simple configuration of bit shifting. By using a circuit, the hardware scale can be reduced.

On the other hand, regarding the time domain equalization, the hardware scale is reduced by dividing the effective symbol period into a plurality of blocks and assigning a fixed transmission path response to each block. In addition, a bit shift circuit is used instead of a multiplier to obtain a transmission path response to be assigned to each block, thereby reducing the hardware scale.

Further, by using time domain and frequency domain equalization in combination, a good equalization characteristic can be obtained even in a fading transmission path.

[0043]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 shows an OF according to a first embodiment of the present invention.
FIG. 4 is a block diagram illustrating a configuration of a transmission line response interpolation circuit used in a DM receiving device. The interpolation circuit of the present embodiment employs a method of estimating the transmission path response of the data signal between the pilot signals by substantially linearly interpolating the transmission path response of the pilot signal. Let's call it interpolation.

First, as an example, the estimated transmission path response Hａ (l, k _{p, l} ) of the pilot signal having the symbol l and the carrier number p in FIG. 15A and the pilot signal having the symbol number l + 4 and the carrier number p are shown. Transmission path response H ~ (l + 4, k
The estimation of the transmission path response of the data signal between ( _{p, l + 4} ) will be described. Each estimated transmission path response is given by the following equation.

(Equation 7)

The above equation shows that for the central data signal k _{d, l + 2} between pilot signals, the pilot signals k _{p, l} and k
The average value of the channel response of _{p, l + 4} is used as the estimated channel response. For the data signal k _{d, l + 1} , the estimated channel response of the data signal k _{d, l + 2} and the pilot signal k _p, The average value of the estimated transmission path response of _l is defined as the estimated transmission path response. Also, the data signal k
_{For d, l + 3} , the average value of the estimated transmission line response of the data signal k _{d, l + 2 and} the estimated transmission line response of the pilot signal k _{p, l + 4 is used as} the estimated transmission line response. By using this interpolation method, it is possible to interpolate the transmission path response between the pilot signals k _{p, l} and k _{p, l + 4} almost linearly.

Since the transmission path response is processed by a binary digital signal, a simple bit shift can be used for the calculation of 1/2 included in the above equation.

FIG. 1 is a block diagram for realizing the equation (6). The estimated value H 値 (l, k) of the transmission line response of the pilot signal is shown in FIG.
_{p, l} ) and HＨ (l + 4, k _{p, l + 4} ) are added by the adder 21, and the result is shifted by one bit by the bit shift circuit 22 to be halved. The output of the bit shift circuit 22 is assumed to be H ~ (l + 2, k _{p, l + 2} ). Further, H ~ (l, k _{p, l} ) and the output of the bit shift circuit 22 are added by an adder 23, the output of the adder 23 is bit-shifted by a bit shift circuit 25, and this value is H ~ (l + 1 , k _{p, l + 1} ). Similarly,
H ~ (l + 4, k _{p, l + 4} ) and the output of the bit shift circuit 22 are added by an adder 24, the output of the adder 24 is bit-shifted by a bit shift circuit 26, and this value is HＨ (l + 3). ,
k _{p, l + 3} ).

According to the configuration of this embodiment, a multiplier is not required in the interpolation circuit, and the hardware scale can be reduced.
In addition, in this method, it is only necessary to store two pilot signals in the time direction for each carrier through which the pilot signal is transmitted, and the memory size can be reduced.

FIG. 2 shows an OF according to a second embodiment of the present invention.
FIG. 4 is a block diagram illustrating a configuration of a transmission line response interpolation circuit used in a DM receiving device. This interpolation circuit uses an FIR filter as an interpolation filter in the symbol direction. Figure 2 is a FIR filter taps NTAP, delay units 27 ₁ ~ 27 _N, the multiplier 28 _0-2 to multiply the coefficients A ₀ to A _N to the output of the input and the delay unit 28 ₁ ~ 28 _N
8 _N , an adder 12 for adding the outputs of the multipliers 28 _{0 to} 28 _N
It is composed of

The pilot signal is supplied to the input of the delay unit 27 ₁ ..., H ~ (l, k _{p, l} ), 0, 0, 0, H ~ (l + 4,
kp _{, l + 4} ),..., and 0 when a data signal is transmitted. Adder 1
From the second output, ..., H ~ (l + 1, k p, l + 1), H ~ (l +
2, k _{p, l + 2} ), H ~ (l + 3, k _{p, l + 3} ),...

According to the configuration of the present embodiment, since a plurality of multipliers are required, the circuit scale becomes large, but the estimation accuracy of the channel response is higher than that of the first embodiment using the bit shift. Get better.

The interpolation using the FIR filter and the simple linear interpolation have been described in the embodiments as the method of interpolating the pilot signal in the time direction. However, using these interpolation methods, the transmission path response to the data signal position is described. Estimation was performed and S / I simulations after equalization were compared.

As a transmission line model, a one-wave Rayleigh fading transmission line was assumed, and the average reception C / N was 30 dB. FIG. 3 shows a transmission path model.

[0055] as the coefficient of Rayleigh fading a _n, J
The following values studied by akes were used:

(Equation 8)

In this case, since the real part and the imaginary part of α _n follow the Gaussian random process, x _{n of the} transmission path output
A _n is subjected to Rayleigh fading in which the amplitude is distributed in a Rayleigh manner and the phase is uniformly distributed.

The simulation model substantially conforms to the 6 MHz version of DVB-T as shown in Table 1 below.

[0058]

[Table 1]

FIG. 4 shows the simulation results. As the interpolation method in the time direction, the simple linear interpolation described in the embodiment and the interpolation using the FIR filter are simulated, and for comparison, the technique of holding the pilot signal for four symbol periods described in the related art is used for comparison. Simulations were also performed. The horizontal axis indicates the maximum Doppler frequency, and the vertical axis indicates the S / I of the equalizer output.

In the hold type, the S / I of the output of the equalizer deteriorates rapidly as the Doppler frequency increases. However, by performing interpolation in the time direction, the S / I ratio close to 20 dB is obtained even at the maximum Doppler frequency of 50 Hz. Was obtained. Note that when the Doppler frequency increases,
Although the characteristics using the FIR filter are improved, at the Doppler frequency of 50 Hz, a result with almost no difference between the simple linear interpolation and the FIR filter is obtained, and the simple linear interpolation method is effective from the viewpoint of hardware scale. It became clear that it was.

Incidentally, the following two points are considered as causes of the characteristic degradation even when the interpolation is performed in the time direction.

First, in this simulation, a method is used in which the amplitude and phase fluctuate at each sample point in order to approach the actual mobile reception condition. Therefore, 1
Even within an OFDM symbol, the phase and amplitude change, and the orthogonality between the carriers is lost, causing the ICI (inter-carrier interfe
rence), and the characteristics are degraded.

Second, in the Rayleigh fading transmission path, the signal amplitude drops to near zero, and a period occurs in which the reception C / N extremely decreases. During this period, since the pilot signal also drops to the noise level, it becomes difficult to accurately estimate the transmission path response, and the characteristics deteriorate.

Generally, in the OFDM system, the above-mentioned 2
In order to deal with the second characteristic degradation, a method is used in which data is dispersed by time interleaving, and even if a certain transmission symbol falls to a noise level, it is restored by data of another symbol.

Also, as in the case of the 2kFFT mode,
If the OFDM symbol length is relatively short,
The second characteristic degradation is not a big problem,
When performing mobile reception in 4k or 8k mode, 1
It is necessary to take some measures for the second characteristic deterioration. Therefore, an embodiment relating to time domain equalization capable of removing ICI will be described below.

As for the time domain equalization method, the sample number N _g -1 shown in FIG.
Is 1 and the transmission path response of N _g + N is h.
There is a method of linearly interpolating a transmission path response of N _g + N from a sample number N _g corresponding to an effective symbol period of an OFDM symbol.

On the other hand, FIG. 5 shows an embodiment (third embodiment) of step interpolation for interpolating the transmission path response stepwise as a method for simplifying the linear interpolation. FIG. 7 (a) divides the effective symbol period into three parts, (b) divides it into five parts, and (c) divides it into nine parts, and assigns an estimated transmission path response to each sample in a stepwise manner. , Step 9. As an example, step 5 of (b) will be described.

First, the transmission path response 1 (C ₁ ) of the sample number N _g −1 and the transmission path response h of the sample number N _g + N
(C ₅₎ adding to by dividing by 2 a (1 + h) / 2 and this correction value is a sample number (2N _g + N) / 2 of the estimated channel response position (C _3). Next, the obtained sample number (2
(3 + h) / 4 is obtained by adding the estimated channel response (C ₃ ) at the position of N _g + N) / 2 and the channel response 1 (C ₁ ) of the sample number N _g −1 and dividing by 2. Sample number (4N _g
+ N) / 4 estimated channel response (C ₂ ), sample number (2N _g + N) / 2 estimated channel response (C ₃ ) and sample number N _g + N channel response h
(C ₅₎ adding to by dividing by 2 a (1 + 3h) / 4 to seek, and sample number (4N _g + N) / 4 of the estimated channel response position (C _4). The transmission path response obtained above is assigned to each sample signal as shown below.

[0069]

(Equation 9)

As the operation for obtaining the estimated values h ~ _n , the operation of dividing by 2 can be a simple bit shift operation as described in the first embodiment, so that the hardware scale is reduced. Can be reduced.

FIG. 6 is a block diagram showing a case where FIG. 5B is implemented by hardware.

First, the adder 32 adds the transmission path response 1 of the sample number N _g -1 (number C ₁ in the figure) and the transmission path response h of the sample number N _g + N (number C ₅ in the figure). The bit shift circuit 33 performs bit shift to obtain an estimated transmission path response (1 + h) / 2, and obtains a sample number (2N _g + N).
/ 2 estimated transmission path response (C ₃ ). Next, the estimated channel response (C ₃ ) at the position of the obtained sample number (2N _g + N) / 2 and the channel response 1 of the sample number N _g −1
(No. C ₁ in the figure) is added by the adder 30 and bit-shifted by the bit shift circuit 31 to obtain (3 + h) / 4, and the estimated transmission path response at the position of the sample number (4N _g + N) / 4 (C ₂ ). At the same time, the sample number (2N _g
+ N) / 2 and an adder 34 adds an estimated transmission path response (C ₃ ) at the position of (C ₃ ) and a transmission path response h of sample number N _g + N (number C ₅ in the figure), and performs a bit shift in a bit shift circuit 35. (1 + 3h) / 4 is obtained, and the sample number (4N
_g + N) / 4 is assumed to be the estimated transmission path response (C ₄ ).

As described above, according to the configuration of the present embodiment, the adders 30, 32, and 34 and the bit shift circuit 3
The hardware can be realized with a simple configuration of 1, 33, and 35.

The above description has been made with reference to FIG.
However, interpolation can be performed with respect to FIG.
is there. This case is referred to as a fourth embodiment, and its block
The result is shown in FIG. This is in addition to the configuration shown in FIG.
Number N_gChannel response 1 (number C in the figure)₁) And sump
Number (4N_g+ N) / 4 estimated transmission path response C _TwoTo
The addition is performed by the adder 36 and the bit shift is performed by the bit shift circuit 37.
To obtain (7 + h) / 8 and sample number
(8N_g+ N) / 8 Estimated channel response C₆And
Constant transmission path response C_TwoAnd C_ThreeAre added by the adder 38, and the bit
(5 + 3h) /
8 and determine the sample number (8N_g+ 3N) / 8
Estimated transmission path response C₇And the estimated transmission path response C_ThreeAnd C_FourAdd
The addition is performed by the arithmetic unit 40 and the bit shift is performed by the bit shift circuit 41.
To obtain (3 + 5h) / 8 and sample number
(8N_g+ 5N) / 8 Estimated transmission path response C₈age,
Estimated transmission path response C_FourAnd C_FiveIs added by the adder 42, and the bit
Bit shift is performed by the shift circuit 43, and (1 + 7h) / 8 is calculated.
The sample number (8N_g+ 7N) / 8 position estimation
Transmission path response C₉It is assumed that. These determined transmission paths
The response is as follows for each sample in the effective symbol period:
Distribute as

[0075]

(Equation 10)

In this embodiment, the number of estimated transmission path responses (in other words, the number of divisions of the effective symbol period) can be increased by repeating the operation of adding adjacent transmission path responses and dividing by two. In general, the number is 1 + 2 ⁿ (where n is a positive integer). In this case, a simple bit shift operation can be used as described above, which is effective in reducing the hardware scale.

FIG. 8 shows an OF according to a fifth embodiment of the present invention.
It is a block diagram of a DM receiver. In FIG. 8, an OFDM signal received by an antenna 44 is converted into a time domain equalizer 47 via an analog signal processor 45 and a synchronous demodulator 46.
Undergoes time domain equalization. The guard interval of the equalized signal is removed by a guard removing unit 48, and the FFT 49
Is converted into a signal in the frequency domain by the frequency domain equalizer 50.
Undergoes frequency domain equalization. According to the configuration of the present embodiment, since equalization in the time domain and the frequency domain is performed in combination, good equalization performance can be obtained.

A simulation was performed on a one-wave Rayleigh fading transmission line using time domain equalization using the interpolation method described above.

First, a simulation system diagram will be briefly described with reference to FIG. In FIG. 9, transmission system 51 allocates random data generated from random data generation section 511 to a plurality of carriers orthogonal to each other at mapping section 512 (X _k ), and converts the data to time axis data at IFFT section 513 ( x _n ), a guard interval is added by the guard adding unit 514, and the guard interval is output. This transmission output is transmitted through a transmission line 53 and further subjected to AWGN (additive Gaussian noise).
Is transmitted to the receiving system 52 via an adding device 54 for adding

The receiving system 53 quantizes the received signal (A / A
D) The power is limited by the peak power limiting circuit 521 so that the peak is constant and converted into a digital signal (x
~ _n ), after undergoing time domain equalization by the time domain equalization section 522 (x ~ _n '), a guard interval signal is removed by the guard removal section 523, and converted to a frequency domain signal by the FFT section 524. The frequency domain equalization unit 525 performs frequency domain equalization and outputs the result. This output X _k is _supplied to the S / I calculator 5
Sent to 5. The S / I calculator 55 includes a mapping output X _k of the transmission system 51 and a frequency domain equalizer 5 of the reception system 52.
The simulation result is obtained by performing S / I operation from the 25 outputs X _k .

That is, here, the output S / I of the frequency domain equalizer 525 of the receiving system 52 was calculated by simulation. Table 2 shows the simulation model.
Simulations were performed for FFT size 4k and 8k modes. Rayleigh fading was generated using the jakes model.

[0082]

[Table 2]

First, the average number of sample points (the number of sample points for performing the correlation operation) described in the equation (2) for deriving h was compared and studied. FIG. 10 shows the simulation results. FIG. 5C shows an interpolation method for dividing the effective symbol period into nine as shown in FIG. 5C as an interpolation method used for time domain equalization, where the average reception C / N = 30 dB and the maximum Doppler frequency is 50 Hz. (Step 9) is used. Further, a 71-tap FIR filter is used for interpolation in the carrier direction used for equalization in the frequency domain, and the simple linear interpolation shown in the first embodiment is used for interpolation in the symbol direction.

As a result of the simulation, the best output S
The average number of points at which / I is obtained is 100 to 30
It turned out to be 0 points.

Next, the S / I of the equalized output due to the difference among the above-mentioned interpolation methods was compared and examined. FIG. 11 shows a simulation result. The results are obtained when the average reception C / N is 30 dB and the maximum Doppler frequency f _{d is} 50 Hz. Linear interpolation, step 3, step 5, and step 9 are used as interpolation methods used for time domain equalization. Also,
7 for carrier direction interpolation used for frequency domain equalization
A one-tap FIR filter is used, and simple linear interpolation is used for interpolation in the symbol direction.

As a result of the simulation, in the case of step 9 in which the effective symbol period is divided into 9 and interpolation is performed in a stepwise manner, there is almost no characteristic deterioration with respect to linear interpolation, but if the number of divisions is smaller than that, the characteristic deterioration will occur. Became clear.

Next, the effect of time domain equalization was compared and studied. FIGS. 12A and 12B show simulation results. (A) is the average reception C / N = 30 dB, FF
In T4k mode, (b) shows average reception C / N = 30 dB,
This is the result in the FFT 8k mode, where the horizontal axis represents the maximum Doppler frequency and the vertical axis represents the S / I of the equalizer output.

Here, the characteristics indicated by △ in FIGS. 12A and 12B are simple linear interpolation in the symbol direction and F in the carrier direction as an interpolation method used for frequency domain equalization.
It is obtained by performing interpolation using an IR filter. On the other hand, ×
The characteristic indicated by is obtained by performing interpolation using an FIR filter only in the carrier direction without performing interpolation in the symbol direction. Regarding the time domain equalization, ○ indicates that the time domain was not equalized, □ × indicates the result of performing linear interpolation, and Δ indicates the result of performing step 9 interpolation. Table 3 summarizes the interpolation methods used for each equalizer.

[0089]

[Table 3]

From these results, it was found that when interpolation was not performed in the symbol direction in the frequency domain equalization, the time domain equalization had no effect and the characteristics were degraded. on the other hand,
By performing equalization in the time domain and the frequency domain, the FFT
When 8k mode is used, the maximum Doppler frequency is 50H
In z, 2 dB S / I compared to equalization only in the frequency domain
It was found that an improvement was obtained.

From the above study, it can be seen that, as in the embodiment shown in FIG. 8, time domain equalization can be effectively used in combination with equalization using interpolation in the symbol direction in the frequency domain. It became clear.

According to the present invention, in the case of frequency domain equalization, a multiplier that increases the circuit scale is not required as an interpolation method in the time direction, and an interpolation method that is excellent in estimating a transmission path response is provided. In this method, in order to estimate a transmission path response of a data signal between pilot signals arranged in the time direction, a pilot signal is added and multiplied by 、, and the result is further added to the pilot signal to be で. By multiplying, the transmission path response of the data signal is interpolated and estimated. The interpolation circuit includes an adder that adds a pilot signal and a multiplier that multiplies the addition result by a coefficient. In a receiving apparatus that performs signal processing with a binary digital signal, the multiplier has a simple configuration of bit shifting. Since it can be constituted by a circuit, the hardware scale can be reduced.

On the other hand, with regard to time domain equalization, the hardware scale can be reduced by dividing the effective symbol period into a plurality of blocks and assigning a fixed transmission path response to each block. In addition, since a transmission path response to be assigned to each block is obtained, a hardware scale can be reduced by using a bit shift circuit instead of a multiplier.

Further, by using the time domain and frequency domain equalization in combination, there is an effect that good equalization characteristics can be obtained even in a fading transmission path.

[0095]

As described above, according to the present invention, a multiplier for increasing the circuit scale is not required as an interpolation method in the time direction when performing equalization in the frequency domain, and the transmission path response is excellently estimated. An OFDM receiver using an interpolation method can be provided.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a transmission line response interpolation circuit used in an OFDM receiver according to a first embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a transmission line response interpolation circuit used in an OFDM receiver according to a second embodiment of the present invention.

FIG. 3 is a block diagram showing a configuration of a transmission path model for comparing respective transmission path response estimation results in the first and second embodiments.

FIG. 4 is a characteristic diagram for explaining the effects of the first and second embodiments.

FIG. 5 is a timing chart for explaining a step interpolation process used in the third and fourth embodiments of the present invention.

FIG. 6 is a block diagram showing a configuration of a transmission line response interpolation circuit used in an OFDM receiver according to a third embodiment of the present invention.

FIG. 7 is a block diagram showing a configuration of a transmission line response interpolation circuit used in an OFDM receiver according to a fourth embodiment of the present invention.

FIG. 8 is a block diagram showing a configuration of an OFDM receiver according to a fifth embodiment of the present invention.

FIG. 9 is a block diagram showing a system configuration used for a simulation to show the effects of the above embodiments.

FIG. 10 is a characteristic diagram for explaining parameter dependence of the fourth and fifth embodiments from the relationship between the average number of points and S / I.

FIG. 11 is a characteristic diagram for explaining parameter dependence of the fourth and fifth embodiments from the relationship between the number of interpolation divisions and S / I.

FIG. 12 is a characteristic diagram for explaining the effects of the fourth and fifth embodiments.

FIG. 13 is a diagram showing a subcarrier transmission format of the DVB-T specification.

FIG. 14 is a block diagram showing a configuration of a frequency domain equalizer.

FIG. 15 is a diagram showing an example of a transmission path response and an example of a pilot signal arrangement.

FIG. 16 is a diagram showing a configuration of an OFDM symbol.

FIG. 17 is a diagram showing an example of an OFDM symbol and a transmission path response.

FIG. 18 is a diagram for explaining linear interpolation as time domain interpolation;

[Explanation of symbols]

11 ... FFT circuit 12 ... memory 13, 16 ... dividing circuit 14 ... symbol filter 15 ... carrier filter 21, 23, 24 ... adder 22, 25, 26 ... bit shift circuit 27 ₁ ~ 27 _n ... delayer 28 _{0-28 n} Multiplier 29 Adder 30, 32, 34, 36, 38, 40, 42 Adder 31, 33, 35, 37, 39, 41, 43 Bit shift 44 Antenna 45 Analog signal processing unit 46 ... Synchronous demodulation unit 47 ... Time domain equalization unit 48 ... Guard removal unit 49 ... FFT 50 ... Frequency domain equalization unit 51 ... Transmission system 511 ... Random data generation unit 512 ... Mapping unit 513 ... IFFT unit 514 ... Guard addition unit 52 .., Receiving system 521, quantization (A / D) peak power limiting circuit 522, time-domain equalizing unit 523, guard removing unit 524, FFT unit 5 5 ... Frequency domain equalization section 53 ... transmission line 54 ... adding device 55 ... S / I calculator

Continued on the front page (72) Inventor Takatoshi Josugi 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Pref. Multimedia System Development Division, Hitachi, Ltd. (56) References JP-A-11-163822 (JP, A) Kaihei 10-209931 (JP, A) JP-A-8-265293 (JP, A) "Simulation study of OFDM equalizer", Journal of the Institute of Image Information and Television Engineers, Vol. 52, No. 11, p. 1643-1649 “Simulation study of equalizer for OFDM”, Technical Report of the Institute of Image Information and Television Engineers, Vol. 21, No. 73, p. 37-42 “High-speed fading compensation method using guard interval in orthogonal Marcharid modulation”, Proc. Of the 1997 IEICE General Conference, Communication 1, p. 669, "Equalizer for OFDM", IEICE Technical Report, IE98-91, "Transmission Characteristics of OFDM Symbol Length and Scattered Pilot in Terrestrial Digital Broadcasting," Journal of the Institute of Image Information and Television Engineers, Vol. 52, No. 11, p. 1643 -1649 (58) Field surveyed (Int. Cl. ⁷ , DB name) H04J 11/00

Claims (6)

(57) [Claims] 1. An O which is obtained by copying a last section of an effective symbol period to a beginning of an effective symbol period to form a guard period.
FDM (Orthogonal Frequency Division Multiplex:
Orthogonal frequency division multiplexing) signal and the received OFDM
The change in the channel response from the end of the tail and the effective symbol period of the guard period of the signal is calculated and an OFDM receiving apparatus that performs equalization in the time domain by using a variation of the channel response, before Symbol within valid symbol is divided into a plurality of blocks, wherein the change of the channel response and the channel response estimation means for estimating a channel response of the plurality of blocks each stepwise transmission of each of the plurality of blocks obtained by this means Time domain equalizing means for performing time domain equalization of the effective symbol period based on a path response.
DM receiving device. 2. A transmission line response estimating means, comprising: a plurality of blocks existing in the effective symbol from a change in a transmission line response;
Wherein an adder and a bit shift circuit device are used as an operation for obtaining a transmission path response of each of the nodes.
2. The receiving device for OFDM according to 1 . 3. A guard period is formed by copying the last section of the effective symbol period to the beginning of the effective symbol period,
OFDM (Orthogonal Frequencies) in which pilot signals of known phases are arranged at substantially equal intervals on the frequency axis and the time axis.
An OFDM receiving apparatus that receives a quency division multiplex (orthogonal frequency division multiplex) signal, wherein the effective symbol is calculated by calculating a change in a channel response from the end of a guard period and the end of an effective symbol period of the received OFDM signal. Is divided into a plurality of blocks, and a transmission path response of each of the plurality of blocks is calculated from a change in the transmission path response.
Time-domain equalizing means for performing step-wise estimation and performing time-domain equalization of the effective symbol period based on the transmission path responses of the plurality of blocks , and processing the transmission path response of the pilot signal as a binary digital signal A frequency domain equalizing means for estimating and interpolating the transmission path response of the received data signal to perform frequency domain equalization. 4. The OFDM equalization means according to claim 1, wherein:
First and second continuous signals on the time axis from the reception output of the M signal
Are sequentially taken out, and the first and second channel responses of the first and second pilot signals are added,
A third estimated transmission path response is obtained by bit-shifting the addition result, and transmission of a data signal between the first and second pilot signals continuous on the time axis using at least the third estimated transmission path response. 4. The OFDM receiver according to claim 3 , wherein the channel response is interpolated, and frequency domain equalization is performed based on the interpolated estimated transmission channel response. 5. The frequency domain equalization means according to claim 1, wherein:
A pilot signal continuous on the time axis is sequentially taken out from the reception output of the M signal, 0 is inserted in a data signal arrangement position between the pilot signals and arranged in time series, and a convolution operation is performed to interpolate a transmission path response, 4. The OFDM receiving apparatus according to claim 3, wherein frequency domain equalization is performed based on the interpolated estimated transmission path response. Wherein said time-domain equalization unit, the plurality present within the effective symbol from the variation of the channel response
4. The OFDM receiver according to claim 3 , wherein an adder and a bit shift circuit device are used as an operation for obtaining a transmission path response of each block .