JP2005229207A - Ofdm receiver and offset correcting method of ofdm reception signal - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an OFDM receiver and an offset correcting method of an OFDM reception signal for adaptively changing a control quantity for correcting a reception signal depending on a noise quantity in correcting sampling offset, and as the result, capable of performing offset correction excellent in flexibility, correctness and responsive performance. <P>SOLUTION: The OFDM receiver has a weighting coefficient selecting circuit 71 for selecting a weighting coefficient μ adapted to a signal to noise ratio (SNR) estimated from a long preamble of an OFDM signal, and an offset corrector 72 for executing correction according to the weighting coefficient μ. If the noise quantity (SNR) changes, the weighting coefficient μ corresponding to the noise quantity (SNR) is adaptively changed, and the correction of the corrector 72 is executed again. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to an OFDM receiver having a function of performing correction for sampling offset data when sampling a received orthogonal frequency division multiplexing (OFDM) signal by obtaining a correction amount for each packet, and an OFDM received signal The present invention relates to an offset correction method.

  OFDM (Orthogonal Frequency Division Multiplexing) is a digital modulation scheme that multiplexes signals of a plurality of orthogonal subcarriers, and is characterized by extremely high frequency use efficiency and resistance to multipath. Currently, the OFDM modulation method is adopted in wireless LANs and terrestrial digital broadcasting of the IEEE 802.11 standard.

The OFDM modulation method is a type of multi-carrier modulation method, in which serial encoded data is converted into parallel encoded data (information symbols) corresponding to a number of subcarriers orthogonal to each other between adjacent ones, and the information symbols are mapped Assigned to orthogonal coordinates by processing (hereinafter, the assigned information symbols are referred to as “constellations”), converted into digital modulated waves by inverse Fourier transform (converting the frequency domain to the time domain), and then adding each of them. To generate and transmit an OFDM signal.
The OFDM receiving apparatus obtains the original encoded data by performing a process reverse to that at the time of transmission on the received OFDM signal.

FIG. 6 shows a schematic diagram of a conventional OFDM receiver.
6 includes an antenna 2, an RF receiver 3, an A / D converter (A / D) 4, a synchronization detection circuit (sync) 5, a fast Fourier transform (FFT) 6, and an equalizer. (equalizer) 7, demapper part (demapper) 8, de-interleave part (de-interleave) 9, de-puncture part (de-puncture) 10, Viterbi decoding part (vitabi) 11, and descrambler part (de-scrambler) 12 Have

  The RF receiver 3 has signal processing functions such as quadrature demodulation and gain adjustment. When receiving an OFDM signal from the antenna 2, the RF receiver 3 performs orthogonal demodulation on the OFDM signal using a local oscillation signal generated by a built-in local oscillator (not shown), thereby converting the OFDM signal in the radio frequency band to an intermediate frequency. Frequency conversion to a band or baseband OFDM signal. When converting to the intermediate frequency band, after performing various processes in the intermediate frequency band, the frequency band is further converted to baseband. The baseband OFDM signal is sent to the A / D converter 4. The frequency-converted OFDM signal is converted into a digital signal by the A / D converter 4 and is synchronously detected by a synchronization detection circuit (sync) 5. Here, “synchronization detection” is to determine the timing of the subsequent OFDM demodulation processing using the preamble signal (short preamble) present at the head of the packet.

The synchronization detection circuit (sync) 5 simultaneously corrects the frequency offset to some extent. Here, “frequency offset” means that the sum of the frequency of the local oscillator and the center frequency of the frequency signal down-converted in the RF section (R / F) 3 on the receiver side is that of the local oscillator on the transmitter side (not shown). This means that there is an error in the demodulated vector composed of real part data and imaginary part data of the signal output from the fast Fourier transform unit (FFT) 6 due to deviation from the frequency, resulting in mutual interference between subcarriers.
Data whose synchronization detection and frequency offset are corrected by the synchronization detection circuit (sync) 5 is input to a fast Fourier transform (FFT) unit 6, and is converted from a time domain signal to a frequency domain signal.

  An equalizer 7 shown in FIG. 6 performs signal processing after FFT, (1) correction of initial phase and initial amplitude by multipath, (2) correction of residual frequency offset, and (3) sampling offset. Correction is being performed. Here, the “sampling offset” means that the sampling frequency in the A / D conversion unit 4 and the fast Fourier transform unit (FFT) 6 shown in FIG. It means that an error occurs in the demodulated vector after FFT.

  The constellation of the data corrected by the equalizer 7 is demapped by a demapper 8. That is, the original encoded data before being associated as information symbols on the orthogonal coordinates in the mapping process on the transmitter side is restored. In the subsequent deinterleaving unit 8, the data interleaved (bit rearranged) on the transmitter side is restored to the original encoded data in order to be strong against transmission path errors that occur in bursts. In the subsequent de-puncture unit (de-puncture) 10, processing to reverse the puncture processing (bit thinning-out) performed in order to increase the coding rate of convolutional coding on the transmitter side is performed, and a Viterbi decoding unit ( Vitabi) 11 executes a Viterbi composite method for returning the convolutional coding, and then descrambler 12 returns the scramble coding, thereby obtaining the original transmission coded data.

Here, the “initial phase and initial amplitude due to multipath” corrected by the equalizer 7 means that when a multipath exists, a plurality of radio waves interfere with each other at the receiving point and a sudden change in amplitude or phase occurs. It is. “Residual frequency offset” is a frequency offset generated because the reference frequency used for frequency conversion for obtaining a baseband signal is not accurate, and the frequency remaining by the synchronization detection circuit (sync) 5 without being fully corrected. It is an offset. Further, the “sampling offset” is an offset caused by a mismatch of sampling frequencies in A / D and D / A.
These widen the constellation of the mapped data and cause an increase in error rate. In the wireless LAN system, there is no guarantee that transmission and reception are always performed with each other, so that correction of the sampling offset cannot refer to past information, and it is necessary to cope with adaptation for each reception. Therefore, it is necessary to correct the sampling offset for each packet.

As a method of correcting the sampling offset, there are two methods, a method of correcting using a time domain signal before FFT and a method of correcting using a frequency domain signal after FFT.
As a method of correcting in the time domain, there is a detection method using a guard interval (see, for example, Patent Document 1).
As shown in FIG. 2, there is a repetitive section of a partial signal called a guard interval (GI) called a guard interval (GI) in one symbol of the OFDM signal. The method described in Patent Document 1 estimates the sampling frequency error by using the correlation between the guard interval (GI) and the signal portion that is the copy source.

  As a method of correcting in the frequency domain, a detection method using a pilot carrier can be mentioned. There are a plurality of signals called pilot carriers in the symbol after FFT. The pilot carrier is a known signal in a virtual situation where noise, multipath, sampling offset, and the like do not exist. The phase shift can be detected from these pilot carrier values, and the sampling frequency error can be estimated. This method is described in Non-Patent Document 1.

The sampling frequency error is estimated using the method described above. As a method of correcting the estimated frequency for the received OFDM signal, a correction method using digital signal processing and a frequency for controlling the frequency are used. There is a method of changing the clock frequency of A / D by providing a control circuit.
JP 09-321733 A M. Spetch, S. Fechtel, G. Fock and H. Meyer, “Optimum Receiver Design for OFDM-Based Broadband Transmission-Part II: A Case Study”, IEEE transaction on communication vol. 49, No. 4, April 2001

Regardless of which method is used, if there is noise in the OFDM signal, the phase difference due to the sampling offset is much smaller than the phase difference caused by the noise, so the sampling frequency error (sampling offset amount) can be estimated accurately. It is difficult to do. This sampling offset amount is usually on the order of 10 ⁻⁴ and is not a problem for short packets. However, since it accumulates in the symbol direction and the carrier direction, for example, when adopting a variable-length packet scheme as in the IEEE 802.11 standard, 1500 In order to transmit / receive a packet having a length of about a byte, tracking (phase fluctuation tracking function) by a pilot signal is required for each information symbol (data symbol) in each packet.
However, in the conventional sampling offset correction method, the sampling offset correction speed does not change according to the amount of noise. Therefore, if the correction speed is specified assuming that there is a lot of noise, the OFDM signal input from a transmission path with little noise is actually processed later, so that it is flexible and accurate and responsive. However, there was a problem that good correction could not be made.

  The problem to be solved is that in the tracking (phase tracking) when correcting the sampling offset in the conventional OFDM receiver, the speed at which the received signal is corrected according to the amount of noise cannot be changed adaptively. The correction is not flexible and accurate and responsive.

  An OFDM receiver according to the present invention is an OFDM receiver that corrects an offset that is a frequency or phase shift of a received orthogonal frequency division multiplexing (OFDM) signal, from a long preamble of the OFDM signal to a signal-to-noise ratio (SNR). ) Is estimated, a correction control unit that outputs a control amount for offset correction adapted to the signal-to-noise ratio, and an offset correction unit that performs the correction according to the control amount adapted to the signal-to-noise ratio. When the signal-to-noise ratio output from the SNR estimator is changed, the correction control unit outputs a control amount corresponding to the changed signal-to-noise ratio, and offset correction is performed based on the output control amount. The unit executes the correction again.

Preferably, in the present invention, the offset correction unit reads an accumulation value of the offset correction amount used for the previous correction in the accumulation target register and an accumulation register for accumulating the offset correction amount, and the read accumulation. A first correction unit that corrects an offset of an input signal according to a value, an offset detection unit that detects an offset from a signal corrected by the first correction unit, an offset output from the offset detection unit, and a previous correction time And the weighted addition average unit for calculating a new offset correction amount by adding the two weighting coefficients with a total of 1 to calculate the new offset correction amount, and the output signal of the first correction unit based on the new offset correction amount. And a second correction unit that corrects the phase of the offset by adding a new offset correction amount to the accumulated value of the accumulation register. While sequentially accumulating quantities, offset correction by the first correction unit, detection of the offset, calculation by the weighted addition averaging unit, and offset correction by the second correction unit are performed in units of symbols in the correction target packet. Repeat as.
In this case, more preferably, the correction control unit includes a weighting coefficient selection circuit that selects and outputs a weighting coefficient value adapted to the input signal-to-noise ratio for at least one of the two weighting coefficients.

Preferably, in the present invention, the offset correcting unit reads a storage register that stores the offset amount, and a phase that reads the stored value of the offset amount from the storage register and rotates the phase of the input signal based on the read storage value. Two weights for the sum of the rotation unit, the offset detection unit for detecting an offset from the signal after phase rotation, the offset output from the offset detection unit, and the offset amount already calculated before the phase rotation A weighted addition averaging unit that calculates and outputs a new offset amount by weighting with a coefficient, calculates a frequency for correcting an offset based on the offset amount output from the weighted addition average unit, and calculates the frequency And a frequency control unit that controls the oscillation frequency of the clock used for the reception process.
In this case, more preferably, the correction control unit includes a weighting coefficient selection circuit that selects and outputs a weighting coefficient value adapted to the input signal-to-noise ratio for at least one of the two weighting coefficients.

  An OFDM received signal offset correction method according to the present invention is an OFDM received signal offset correction method for correcting an offset that is a frequency or phase shift of a received orthogonal frequency division multiplexing (OFDM) signal. An SNR calculation step for calculating a signal-to-noise ratio (SNR) from a long preamble, a correction control step for obtaining a control amount for offset correction adapted to the signal-to-noise ratio, and the correction performed according to the control amount adapted to the signal-to-noise ratio Offset correction step, and when the noise level of the OFDM signal changes and the signal-to-noise ratio obtained in the SNR calculation step is changed, the correction control step is executed again to obtain the changed signal-to-noise ratio. A corresponding new control amount is obtained, and the offset correction step is executed again based on the new control amount. To.

The operation of the OFDM receiving apparatus and offset correction method of the present invention will be described as follows by taking the OFDM receiving apparatus as an example.
First, the SNR estimator section detects a long preamble of the OFDM signal and estimates a signal-to-noise ratio (SNR) from the detection result. The SNR is sent to the correction control unit, where a control amount for offset correction is obtained, and the offset correction unit performs correction using this control amount.

More specifically, for example, the offset correction unit includes an accumulation register, a first correction unit, an offset detection unit, a weighted addition averaging unit, and a second correction unit, and a weighting coefficient is defined as a control amount. In this case, the accumulation register sequentially accumulates the offset correction amounts. When the accumulated value of the offset correction amount used for the previous correction is read from the accumulation register and input to the first correction unit, the first correction unit reads the offset of the input signal. Correct with the accumulated value.
The correction at this time is for reproducing the offset accumulated up to the previous time, and then an offset correction amount to be newly added to this offset is obtained.
More specifically, the offset detection unit detects an offset of the signal output from the first correction unit. The detected offset is input to the weighted averaging unit together with the weighting coefficient from the correction control unit (weighting coefficient selection circuit). The weighted addition averaging unit weights and adds this offset and the offset correction amount at the time of the previous correction by using two weighting coefficients that are 1 in total. When one of the two weighting coefficients is a coefficient output from the weighting coefficient selection circuit, the other one is internally calculated by subtracting the coefficient from 1. The addition result is input as a new offset correction amount to the second correction unit, and the second correction unit corrects the signal from the first correction unit with the new offset correction amount. At the same time, a new offset correction amount is added to the accumulated value of the accumulation register, and the offset correction amount is cumulatively added here. Thus, the offset correction process for one symbol unit is completed.

The offset tracking correction is executed in the correction target packet by repeating such correction processing in symbol units while adding the new offset correction amount to the accumulated value of the accumulation register and sequentially accumulating the offset correction amount.
For example, when different packets are input, or when different values are detected as SNR even in the same packet, a control amount (weighting coefficient) corresponding to the detected value is output, and an appropriate offset correction amount is calculated accordingly. More effective correction is performed.

  According to the OFDM receiver and the offset correction method of the present invention, when the sampling offset is corrected, the control amount for correcting the received signal is adaptively changed according to the noise amount, and as a result, a flexible and accurate response is achieved. There is an advantage that offset correction with good characteristics becomes possible.

The present invention estimates a signal-to-noise ratio (SNR) from a long preamble of an OFDM signal, adaptively changes a control amount according to the estimated SNR, and executes appropriate offset correction quickly and appropriately with the control amount at that time. A system and a correction method are proposed.
Hereinafter, in phase tracking (tracking) for offset correction, an example in which an offset correction amount to be added to a previous offset is obtained by weighted averaging processing and a weighting coefficient is used as a control amount for correction will be described as an example. A form is demonstrated.

  In contrast to the schematic configuration diagram of the conventional OFDM receiving apparatus shown in FIG. 6, a schematic configuration diagram of the OFDM receiving apparatus according to the present embodiment is shown in FIG. Here, an OFDM receiver that performs sampling offset correction in the frequency domain is shown, but a similar technique can be used when performing sampling offset correction in the time domain.

  An OFDM receiver 1 shown in FIG. 1 includes an antenna 2, an RF receiver 3, an A / D converter (A / D) 4, a synchronization detection circuit (sync) 5, a fast Fourier transformer (FFT) 6, and an equalizer. (proposed equalizer) 70, demapper unit (demapper) 8, de-interleave unit (de-interleave) 9, de-puncture unit (de-puncture) 10, Viterbi decoding unit (vitabi) 11, descrambler unit (de-scrambler) 12, The SNR estimator unit 13 is included.

The RF receiver 3 has signal processing functions such as quadrature demodulation and gain adjustment. When receiving an OFDM signal from the antenna 2, the RF receiver 3 performs orthogonal demodulation on the OFDM signal using a local oscillation signal generated by a built-in local oscillator (not shown), thereby converting the OFDM signal in the radio frequency band to an intermediate frequency. Frequency conversion to a band or baseband OFDM signal. When converting to the intermediate frequency band, after performing various processes in the intermediate frequency band, the frequency band is further converted to baseband. The baseband OFDM signal is sent to the A / D converter 4.
The frequency-converted OFDM signal is converted into a digital signal by the A / D converter 4 and is synchronously detected by a synchronization detection circuit (sync) 5. The synchronization detection circuit 5 also corrects the frequency offset to some extent at the same time. Data whose synchronization detection and frequency offset have been corrected by the synchronization detection circuit 5 are input to a fast Fourier transform (FFT) unit 6 where the time domain signal is converted to a frequency domain signal.

  Like the equalizer 7 shown in FIG. 6, the equalizer 70 shown in FIG. 1 is (1) correction of initial phase and initial amplitude by multipath, (2) correction of residual frequency offset, (3) Although the sampling offset is corrected, this embodiment is particularly characterized by a configuration (offset correction unit) relating to the sampling offset correction. Details of the configuration of the sampling correction unit and the correction method will be described later. The SNR estimator 13 that estimates the signal-to-noise ratio (SNR) as a parameter for changing the control amount at the time of offset correction estimates the SNR by estimating the SNR from the output of the synchronization detection circuit (sync) 5 in the example shown in FIG. Is provided to a proposed equalizer 70 proposed in the present embodiment.

  The data corrected by the proposed equalizer 70 is demapped by a demapper 8. That is, the original encoded data before being associated as information symbols on the orthogonal coordinates in the mapping process on the transmitter side is restored. In the subsequent deinterleaving unit 8, the data interleaved (bit rearranged) on the transmitter side is restored to the original encoded data in order to be strong against transmission path errors that occur in bursts. In the subsequent de-puncture unit (de-puncture) 10, processing to reverse the puncture processing (bit thinning-out) performed in order to increase the coding rate of convolutional coding on the transmitter side is performed, and a Viterbi decoding unit ( Vitabi) 11 executes a Viterbi composite method for returning the convolutional coding, and then descrambler 12 returns the scramble coding, thereby obtaining the original transmission coded data.

FIG. 2 shows an example of IEEE. 2 shows a packet structure of an OFDM signal defined by the 11a standard.
The OFDM signal shown in FIG. 2 includes long symbols # 1 and # 2 constituting a preamble part, a signal field (SIGNAL FIELD) constituting a header part, and a number of data symbols (1st constituting a data part). OFDM, ...). The long symbols # 1 and # 2, the signal field, and each data symbol are composed of N pieces of sampling data corresponding to the sampling frequency, and have a certain symbol length. In the head part of each symbol (left part of FIG. 2), a guard interval (a part of the back of the head formed by copying a part of the back of the head so that the desired data is within the FFT window phase even if there is a delay in the multipath environment). GI) is added. Although not particularly shown, 10 short symbols are added to the header portion before a long symbol # 1 for every predetermined number of subcarriers.

  In particular, when the packet shown in FIG. 2 is assigned to a predetermined number (for example, 52 waves in the IEEE 11a standard) of subcarriers having different frequencies at equal intervals, the subbands of 48 waves in which data is stored are stored. Pilot signals are embedded in subcarriers of four waves allocated at almost constant frequency intervals between carriers. The pilot signal has a configuration in which the data portion is replaced with a reference symbol, and is normally used as a reference signal for tracking a radio frequency shift during OFDM demodulation. In addition, short symbols used for automatic gain control (AGC) and the like are arranged with the number of subcarriers of 12 waves thinned out in total considering that the AGC operation is performed by an inexpensive local oscillator.

As described above, for example, if the sampling frequency of the A / D converter (A / D) 4 shown in FIG. 1 is slightly deviated from the sampling frequency on the transmission side, this becomes a sampling offset and the frequency or phase of the OFDM signal. Appears as a gap.
FIG. 3B is a diagram schematically illustrating how signal phase differences are accumulated when data symbols corresponding to 48 subcarriers are sequentially transmitted. The horizontal axis in FIG. 3B corresponds to the subcarrier index i, and the vertical axis represents the phase shift amount due to the sampling offset. In addition, among subcarriers to which a total of 52 effective symbols are allocated, effective symbols (K = 48 subcarriers excluding 4 pilot carriers (shaded portions excluding the vicinity of the center and the outer portion in FIG. 3A)) ( A symbol portion having a length excluding the GI length from the symbol length) is arranged, and a reference symbol is allocated to four-wave pilot carriers therebetween.

From this figure, it can be seen that the phase shift amount changes from negative to positive while the effective symbol changes in the order of the subcarrier index i. The slope α at this time can be defined as the sampling offset amount, and its value is as small as 10 ⁻⁴ order. However, for example, IEEE. In the 11a standard, a packet as long as about 1500 bytes may be sent in the packet. In this case, since the sampling offset is accumulated in the packet and cannot be ignored with respect to the noise level, in this embodiment, an equalizer is provided so that the sampling offset does not increase following the input of the received signal. 70 is corrected (tracking).
Here, with respect to the sampling offset performed by the equalizer 70, in particular, in a system such as a wireless LAN, there is no guarantee that transmission and reception are always performed with each other. Therefore, correction of the sampling offset cannot be performed by referring to past information. Every adaptation needs to be dealt with. Therefore, it is necessary to correct the sampling offset for each packet.

Next, a configuration for correcting the sampling offset and a correction method thereof will be described. First, the estimation of SNR will be described.
The SNR estimator unit 13 inputs the OFDM digital signal after synchronization detection by the synchronization detection circuit (sync) 5 shown in FIG. 1, and takes in information of two long symbols there. In this example, symbols represented by a long symbol # 1 and a long symbol # 2 shown in FIG. 2 are taken into the SNR estimator unit 13. In a virtual case where there is no noise, multipath, or frequency offset, the signal waveforms of the two symbols # 1 and # 2 are the same. In addition, since a guard interval (GI) exists before long symbol # 1, signals of two long symbols # 1 and # 2 are forwarded, that is, captured by GI unless the number of samples existing in GI is exceeded. Good.

Assuming that the long symbol # 1 signal is X ₁ (i) and the long symbol # 2 signal is X ₂ (i), the SNR estimator 13 performs the following calculation. Details of this calculation method are described in Reference 1: “S. Muller-Weinfurtner,“ OFDM for Wireless Communications: Nyquist Windowing, Peak-Power Reduction and Synchronization ”, August 2000”.

The signals of the two long symbols after performing FFT are X ₁ (i) and X ₂ (i) (0 ≦ i ≦ N−1), the respective powers are P ₁ and P ₂ , and the cross power is S. Define. P ₁ , P ₂ and S are given by the following equations.

Here, the SNR estimator ζ _SNR is defined by the following equation.

The relationship between the SNR estimator ζ _SNR and SNR can be approximated by the following equation.

Although the calculation after FFT is described in the above document 1, the SNR can be estimated before the data signal is input to the fast Fourier transform unit (FFT) 6 by a similar calculation method.
Also, the SNR estimated by the SNR estimator 13 is output to a block of a proposed equalizer 70 proposed in the present embodiment.

Next, sampling offset correction executed by the digital signal processing in the equalizer 70 will be described.
When sampling offset occurs, the signal in the frequency domain has a phase Δθi proportional to the subcarrier as shown in FIG. 3B, and when correction is not applied, the slope α increases for each symbol. Go.
To apply the correction, first, the phase gradient α is calculated from the pilot carrier. However, the inclination α is not only caused by the sampling offset, but also affected by noise. Therefore, in order to suppress the influence of noise, the averaging process is performed using the phase gradient calculated for the past symbols.

If the slope of the phase Δθm at the mth symbol is α _m , and the slope of the phase Δθm-1 after actually correcting the m−1th symbol is α ′ _m−1 , it is corrected at the mth symbol. The phase gradient α ′ _m to be obtained can be obtained by the following equation (4).

ここで、μは重み付け係数で0≦μ≦1を満たす定数である。

Here, μ is a constant satisfying 0 ≦ μ ≦ 1 as a weighting coefficient.

  The phase obtained by multiplying the slope obtained in this way by each subcarrier number is calculated. If the OFDM signal in the frequency domain to be corrected is the mth symbol, the sampling offset correction is performed by applying a phase rotation to the mth symbol data in the phase rotation component of the sum of the correction values up to the mth symbol. I do.

  If the weighting coefficient μ is constant, the tracking speed increases as the value increases, but there is a trade-off that it is more susceptible to noise. A description of a trade-off depending on the value of the weighting coefficient μ is described in Reference 2: “Heejung Yu et all,“ Residual Carrier and Sampling Frequency Offset Tracking In OFDM Wireless LAN Systems, CIC, October 2002 ”. The trade-off described in Document 2 relates to the residual frequency offset, but there is a similar trade-off regarding the sampling offset.

Therefore, in the present embodiment, the weighting coefficient μ is dynamically changed according to the magnitude of noise, and the weighting coefficient is selected as a correction control unit that outputs a control amount in the equalizer 70. A circuit and an offset correction circuit are provided.
FIG. 4A shows a weighting coefficient selection circuit, and FIG. 4B shows the configuration of the offset correction unit. The configuration shown in FIG. 4B is obtained by realizing the above-described equation (4) with a circuit.

  The weighting coefficient selection circuit 71 shown in FIG. 4A may be configured to select and output the weighting coefficient μ by calculation. Here, the weighting coefficient selection circuit 71 holds an optimum value of the weighting coefficient μ for the assumed SNR level. The weighting coefficient selection circuit 71 is composed of a table (storage circuit). In the example of FIG. 4, although not essential in the present invention, information such as a packet data length and a coding rate is also taken into the weighting coefficient selection circuit 71, and the information and the SNR are In the combination, the optimum value of the weighting coefficient μ is output.

The offset correction unit 72 shown in FIG. 4B roughly includes a storage register 73, a first correction unit 74, an offset detection unit 75, a weighted addition averaging unit 76, and a second correction unit 77.
The accumulation register 73 includes an adder 78 and a latch circuit 79 that outputs the previous addition result to one input of the adder 78, and the output of the adder 78 is input to the first correction unit 74.
When a certain symbol (m-th symbol in FIG. 4A) is input, the first correction unit 74 sets the phase of the m-th symbol to the accumulated value of the accumulation register 73, that is, the accumulated value of the offset correction amount. Based on this, the phase is reversely rotated so as to cancel a part of the offset component. The correction at this time (reverse phase rotation) is for reproducing the offset accumulated up to the previous time, and then an offset correction amount to be newly added to this offset is obtained.

More specifically, the signal after the reverse phase rotation is input from the first correction unit 74 to the offset detection unit 75, where the remaining offset α _m for the m-th symbol is detected. The detected offset α _m is input to the weighted addition averaging unit 76 together with the weighting coefficient μ from the weighting coefficient selection circuit 71.
The weighted addition averaging unit 76 has an amplifier (or multiplier) 761 that multiplies the remaining offset α _m for the _mth symbol, and an amplifier (or multiplier) 762 that multiplies the previous calculation result by (1−μ). The adder 763 adds the outputs of the two amplifiers 761 and 762, and the delay circuit 764 delays the output of the adder 763.

From the weighted addition averaging unit 76 having such a configuration, a new offset correction amount according to the above-described equation (4) is output and output to the second correction unit 77.
The second correction unit 77 further reversely rotates the phase of the input signal from the first correction unit 74 by a new offset correction amount, and as a result, the correction processing for one symbol unit (m-th symbol) is completed. To do.

Such a correction process in symbol units is repeated while adding a new offset correction amount to the accumulated value of the accumulation register 73 and sequentially accumulating the offset correction amount, thereby phase tracking the sampling offset within the correction target packet. (Tracking) correction is performed.
Also, for example, when different packets are input, or when different values are detected as SNR even in the same packet, the weighting coefficient μ corresponding to the selected value is selected and output by the weighting coefficient selection circuit 71, and an appropriate offset is set accordingly. A correction amount is calculated, and more effective correction is performed. Therefore, when the noise is low, increase the weighting coefficient μ to increase the tracking speed and execute a quick offset correction. Conversely, if the noise increases, the weighting coefficient μ is decreased to slow down the tracking, Noise tolerance can be increased so that sampling offset correction is not affected by noise. As a result, it is possible to overcome the trade-off between tracking speed and noise immunity, perform offset correction with flexibility, accuracy and good response, and realize an OFDM receiver that is resistant to demodulation errors due to phase and frequency shifts.

  Note that the optimum weighting coefficient μ differs depending on the transmission speed (encoding rate), packet data length, etc. even if the estimated noise value (SNR) is the same. Therefore, after the information included in the signal field (SIGNAL FIELD) such as the transmission rate and the packet data length is decoded, the descrambler (de-scrambler) 12 converts the information into an equalizer as shown in FIG. It is desirable to feed back to (proposed qualizer) 70 and to contribute to the selection of the optimum weighting coefficient μ together with the SNR. In that case, the response and accuracy of offset correction can be further improved.

  Further, the SNR may be estimated in the frequency domain. In that case, the SNR is estimated from the signal output from the fast Fourier transform (FFT) unit 6, but at this time, there may be a case where there is not much time margin for offset correction by the equalizer 70. In that sense, it is desirable to estimate the SNR in the time domain as shown in FIG.

On the other hand, a method of changing the A / D clock frequency by providing a frequency control circuit can also be adopted. FIG. 5 shows the configuration of the offset correction circuit when this method is employed.
The offset correction unit 72A illustrated in FIG. 5 includes a storage register unit 73 and a weighted addition averaging unit 76, as in the case illustrated in FIG. Moreover, it has the phase rotation part 74A of the same function as the 1st correction | amendment part 74 shown to FIG. 4 (B). When the m-th symbol is input, the phase rotation unit 74A reversely rotates the phase so as to cancel a part of the offset component based on the accumulated value of the accumulation register 73, that is, the accumulated value of the offset correction amount. Let However, since the phase of the symbol is not actually corrected by this phase rotation but is used to estimate a frequency shift, it is referred to as a “phase rotation unit” instead of a “correction unit”. An offset of the signal after phase rotation is detected by the offset detection unit 75, and a weighted addition average using the detected offset and the weighting coefficient μ is executed by the weighted addition average unit 76.

  The offset correction unit 72 </ b> A illustrated in FIG. 5 includes a frequency control unit 80. The frequency control unit 80 calculates a frequency for correcting the offset based on the offset amount output from the weighted addition averaging unit 76, and sets the oscillation frequency of the clock used for reception processing, for example, A / D, so that the calculated frequency is obtained. Control.

In the present embodiment, as described above, signal field information other than the OFDM signal and SNR is input to the SNR estimator section (SNR estimator) 13. As shown in FIG. 2, the signal field is a symbol after the long symbol, and this signal field includes information such as a transmission rate (rate) and a data length (data length) in the packet. . Since the noise immunity varies depending on the transmission speed, and the number of symbols varies depending on the data length, the tracking speed that minimizes the error rate also varies.
In this embodiment, when the values of SNR, transmission rate, and data length are given, for example, a weighting coefficient μ for sampling offset correction that minimizes the PER (packet error rate) is set at the receiver design stage. In FIG. 4A, a weighting coefficient selection circuit 71, which is a block that outputs an optimum weighting coefficient μ when each signal is input, is provided in an equalizer 70. Incorporate into. As a result, when the A / D clock frequency is changed as shown in FIG. 5, the optimum clock frequency shift to be corrected is offset, and when the digital processing shown in FIG. Can be output as a quantity.

  As described above, according to the present embodiment, in the OFDM receiver, the frequency of the oscillator that supplies the clock to the D / A on the transmitter side and the frequency of the oscillator that supplies the clock to the A / D on the receiver side are Even if they do not match, the receiver can reduce the error rate and reliability by dynamically changing the weighting coefficient μ used to correct the sampling offset in consideration of the trade-off between tracking speed and noise effects on a packet basis. High OFDM signal communication can be realized. At this time, determination of the weighting coefficient μ used for correcting the sampling offset can be realized with an easy circuit configuration.

It is a schematic block diagram of the OFDM receiver which concerns on this Embodiment. IEEE. It is a schematic diagram which shows the packet structure of the OFDM signal prescribed | regulated by 11a standard. (A) is a schematic diagram showing the arrangement of subcarriers, and (B) is a schematic diagram showing how signal phase differences are accumulated when data symbols corresponding to 48 subcarriers are sequentially transmitted. FIG. (A) is a figure which shows a weighting coefficient selection circuit, (B) is a figure which shows the structure of an offset correction | amendment part. It is a figure which shows the structural example of the offset correction circuit in the case of correct | amending an offset by changing the clock frequency of A / D. It is a figure which shows schematic structure of the conventional OFDM receiver.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... OFDM receiver, 2 ... Antenna, 3 ... RF receiver, 4 ... A / D converter, 5 ... Synchronization detection circuit, 6 ... Fast Fourier transform part, 7, 70 ... Equalizer, 8 ... Demapper part, DESCRIPTION OF SYMBOLS 9 ... Deinterleave part, 10 ... Depuncture part, 11 ... Viterbi decoding part, 12 ... Descrambler part, 13 ... SNR estimator part, 71 ... Weighting coefficient selection circuit, 72, 72A ... Offset correction part, 73 ... Accumulation register, 74: First correction unit, 74A: Transfer rotation unit, 75: Offset detection unit, 76: Weighted addition averaging unit, 77: Second correction unit, 80: Frequency control unit, μ: Weighting coefficient

Claims (9)

An OFDM receiver that corrects an offset that is a frequency or phase shift of a received orthogonal frequency division multiplexing (OFDM) signal,
An SNR estimator that estimates the signal-to-noise ratio (SNR) from the long preamble of the OFDM signal;
A correction control unit that outputs a control amount of offset correction adapted to the signal-to-noise ratio;
An offset correction unit that performs the correction according to a control amount adapted to the signal-to-noise ratio;
When the signal-to-noise ratio output from the SNR estimator unit is changed, the correction control unit outputs a control amount corresponding to the changed signal-to-noise ratio, and the offset correction unit is based on the output control amount. An OFDM receiver that performs the correction again. The offset correction unit
A storage register that stores the offset correction amount;
A first correction unit that reads an accumulated value of an offset correction amount used for the previous correction in the correction target packet from the accumulation register, and corrects an offset of an input signal by the read accumulated value;
An offset detection unit for detecting an offset from the signal corrected by the first correction unit;
A weighted addition averaging unit that calculates the new offset correction amount by adding the offset output from the offset detection unit and the offset correction amount at the time of the previous correction by weighting and adding the two weighting coefficients with a total of 1.
A second correction unit that corrects the phase of the output signal of the first correction unit with a new offset correction amount;
While adding a new offset correction amount to the accumulated value of the accumulation register and sequentially accumulating the offset correction amount, offset correction by the first correction unit, detection of the offset, calculation by the weighted addition averaging unit, and the first The OFDM receiver according to claim 1, wherein the offset correction by the two correction units is repeated for each symbol in the correction target packet. The OFDM receiver according to claim 2, wherein the correction control unit includes a weighting coefficient selection circuit that selects and outputs a weighting coefficient value adapted to the input signal-to-noise ratio for at least one of the two weighting coefficients. . The OFDM receiver according to claim 2, wherein the offset detection unit extracts a plurality of pilot carriers in symbol units in the frequency domain, and calculates an offset change amount by calculating a value of the extracted pilot carriers. The OFDM receiver according to claim 2, wherein the offset detection unit calculates a frequency or phase shift of an OFDM signal by calculating a correlation using a guard interval in a symbol in a time domain, and outputs the offset as the offset. The SNR estimator unit detects two identical reference symbols from an input OFDM signal, and calculates an SNR by multiplying and dividing the values of the two reference symbols in the time domain or frequency domain. The OFDM receiver according to the description. The offset correction unit
A storage register that stores the offset amount;
A phase rotation unit that reads the accumulated value of the offset amount from the accumulation register and rotates the phase of the input signal according to the read accumulated value;
An offset detector for detecting an offset from the signal after phase rotation;
Weighted addition for calculating and outputting a new offset amount by adding the offset output from the offset detection unit and the offset amount already calculated before the phase rotation by adding the two weighting coefficients that are 1 in total. Average part,
A frequency control unit that calculates a frequency for correcting an offset based on an offset amount output from the weighted averaging unit and controls an oscillation frequency of a clock used for reception processing so as to be the calculated frequency. The OFDM receiver according to 1. The OFDM receiver according to claim 7, wherein the correction control unit includes a weighting coefficient selection circuit that selects and outputs a weighting coefficient value adapted to the input signal-to-noise ratio for at least one of the two weighting coefficients. . An OFDM received signal offset correction method for correcting an offset that is a frequency or phase shift of a received orthogonal frequency division multiplexing (OFDM) signal, comprising:
An SNR calculation step of calculating a signal to noise ratio (SNR) from a long preamble of the OFDM signal;
A correction control step for obtaining an offset correction control amount adapted to the signal-to-noise ratio;
An offset correction step for performing the correction according to a control amount adapted to the signal to noise ratio,
When the noise level of the OFDM signal is changed and the signal-to-noise ratio obtained in the SNR calculating step is changed, the correction control step is executed again to obtain a new control amount corresponding to the changed signal-to-noise ratio, An offset correction method for an OFDM reception signal, wherein the offset correction step is executed again based on the new control amount.

Images