JP2009206930A - Receiver, and signal equalizing apparatus and method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To estimate SNR with high accuracy without increasing overhead and to perform equalization processing in the frequency domain by MMSE.
An SNR estimator 54 averages channel frequency responses, subtracts the averaged channel frequency response and the channel frequency response to estimate a noise component, and calculates signal power from the averaged channel frequency response. Then, the noise power is calculated from the estimated noise component, and the SNR is estimated. The MMSE equalization unit 55 obtains an equalizer coefficient using the estimated SNR, and performs equalization processing in the frequency domain by MMSE.
[Selection] Figure 1

Description

  The present invention relates to a receiving apparatus suitable for use in wireless data communication, and a signal equalizing apparatus and method for equalizing a multipath fading channel.

  In wireless communication, SC-FDE (Single Carrier Frequency Domain Equalization) is a technology that has many advantages such as a low peak-to-average ratio and a strong carrier offset, and is a promising technology.

  In wireless communication, equalization processing is performed on the receiving side in order to remove intersymbol interference due to a multipath fading channel. The equalization processing includes equalization in the time domain and equalization in the frequency domain, and it is known that equalization in the frequency domain exhibits better characteristics (for example, Non-Patent Document 1). ).

Frequency domain equalization is performed by converting a time domain received signal into a frequency domain received signal by FFT (Fast Fourier Transform), and the equalized signal is again time domaind by IFFT (Inverse Fast Fourier Transform). Is returned to the received signal and output. Frequency domain equalization takes the form of both zero forcing or MMSE (Minimum Mean-Square Error), and MMSE is known to perform better than zero forcing (eg, Non-Patent Document 2).
H. Sari, G. Karam, and I. Jeanclaud, “Frequency-domain equalization of mobile radio and terrestrial broadcast channels, in Proc.IEEE GLOBECOM 194, vol. 1, pp. 1, November 1994. M. Lei, R. Kimura, C.-S. Sum, R. Funada, Y. Shoji, H. Harada, and S. Kato, “MMSE-FDE based on estimated SNR for single-carrier block transmission (SCBT) in multi-Gbps WPAN, “submitted to IEEE WCNC 2008, Sept. 2007.

  When MMSE is used, the accuracy of SNR (Signal to Noise Ratio) estimation affects the performance of MMSE-FDE. In the case of zero forcing, it is not necessary to estimate the SNR, but the equalization performance is degraded as compared with MMSE. If a special signal is inserted to estimate the SNR, a problem arises that overhead increases.

  In view of the above-described problems, the present invention is capable of estimating SNR with high accuracy without increasing overhead and performing equalization processing in the frequency domain by MMSE, as well as a signal equalizer, It aims to provide a method.

  In order to solve the above-described problems, a receiving apparatus according to the present invention converts a signal received in the time domain into a frequency domain, and performs a signal-to-noise ratio from a channel frequency response. An SNR estimation unit for estimation, and an MMSE frequency equalization unit that performs equalization in the frequency domain by obtaining an equalizer coefficient by a least mean square error method using the signal-to-noise ratio estimated by the SNR estimation unit It is characterized by.

  Preferably, the channel frequency response is obtained from a signal in the time domain. Preferably, the channel frequency response is obtained from a frequency domain signal.

  The signal equalization apparatus of the present invention converts a time domain signal into a frequency domain signal, performs equalization processing in the frequency domain, and returns to the time domain signal again for output. An SNR estimator that estimates a signal-to-noise ratio from a response, an MMSE that performs equalization in the frequency domain by obtaining an equalizer coefficient by a least mean square error method using the signal-to-noise ratio estimated by the SNR estimator And a conversion unit.

  Preferably, the SNR estimation unit averages the channel frequency response, an averaged channel frequency response, a subtraction unit that subtracts the channel frequency response to estimate a noise component, and an averaged channel frequency A signal power calculation unit that calculates signal power from the response, a noise power calculation unit that calculates noise power from the estimated noise component, the signal power obtained by the signal power calculation unit, and the noise power calculation unit And a division unit that calculates a signal-to-noise ratio from the noise power.

  The signal equalization method of the present invention is a frequency domain equalization method in which a time domain signal is converted into a frequency domain signal, equalization processing is performed in the frequency domain, and the time domain signal is output again. Including a step of estimating a signal-to-noise ratio from a response and a step of obtaining an equalizer coefficient by a least mean square error method and performing equalization in a frequency domain using the estimated signal-to-noise ratio. To do.

  Preferably, estimating the signal to noise ratio comprises averaging the channel frequency response, subtracting the averaged channel frequency response and the channel frequency response to estimate the noise component, and averaging Calculating a signal power from the channel frequency response; calculating a noise power from the estimated noise component; calculating a signal-to-noise ratio from the obtained signal power and the obtained noise power; It is characterized by including.

  According to the present invention, an SNR estimator that estimates a signal-to-noise ratio from a channel frequency response in a receiving device that converts a received signal in the time domain into a frequency domain and performs equalization processing in the frequency domain, and an SNR estimator SNR estimated from the channel frequency response by including an MMSE frequency equalization unit that obtains an equalizer coefficient by a least mean square error method and performs equalization in the frequency domain using the signal-to-noise ratio estimated by Can be used to perform equalization in the frequency domain by MMSE more accurately without increasing overhead.

  According to the invention, the channel frequency response is determined from a signal in the time domain. According to the present invention, the channel frequency response is obtained from a frequency domain signal. The SNR can be estimated from the channel frequency response thus obtained.

  According to the present invention, in a frequency domain equalization apparatus that converts a time domain signal into a frequency domain signal, performs equalization processing in the frequency domain, and returns to the time domain signal again, the signal is output from the channel frequency response. An SNR estimator for estimating a noise-to-noise ratio; an MMSE equalizer for performing equalization in a frequency domain by obtaining an equalizer coefficient by a least mean square error method using the signal-to-noise ratio estimated by the SNR estimator; Therefore, using the SNR estimated from the channel frequency response, equalization processing in the frequency domain by MMSE can be performed more accurately without increasing overhead.

  According to the present invention, the SNR estimation unit averages the channel frequency response, the averaged channel frequency response, the subtraction unit that subtracts the channel frequency response to estimate the noise component, and the averaging A signal power calculation unit that calculates signal power from the channel frequency response, a noise power calculation unit that calculates noise power from the estimated noise component, a signal power obtained by the signal power calculation unit, and a noise power calculation unit Since a division unit for calculating a signal-to-noise ratio is provided from the obtained noise power, the SNR can be accurately estimated from the channel frequency response.

  According to the present invention, in a frequency domain equalization method for converting a time domain signal into a frequency domain signal, performing equalization processing in the frequency domain, and returning to the time domain signal again, the signal from the channel frequency response is output. The channel frequency response includes a step of estimating a noise-to-noise ratio and a step of performing equalization in the frequency domain by obtaining an equalizer coefficient by a least mean square error method using the estimated signal-to-noise ratio. By using the SNR estimated from the above, it is possible to perform equalization processing in the frequency domain by MMSE more accurately without increasing overhead.

  According to the invention, estimating the signal to noise ratio comprises: averaging a channel frequency response; subtracting the averaged channel frequency response; and the channel frequency response to estimate a noise component; Calculate the signal power from the averaged channel frequency response, calculate the noise power from the estimated noise component, calculate the signal power, and calculate the signal-to-noise ratio. Therefore, the SNR can be accurately estimated from the channel frequency response.

First embodiment.
Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 shows an outline of the SC-FDE system according to the first embodiment of the present invention.

  In FIG. 1, on the transmitter 1 side, a transmission bit stream is coded by a channel coding unit 11, modulated by a modulation unit 12, a guard interval is inserted by a guard interval insertion unit 13, and transmitted from a transmission unit 14. A transmission signal from the transmitter 1 is received by the receiver 2 through the transmission path of the multipath fading channel.

  On the receiver 2 side, the signal from the transmitter 1 is received by the receiver 20, the guard interval is removed by the guard interval remover 21, and then sent to the frequency domain equalizer 23.

  The frequency domain equalization unit 23 includes an FFT unit 51, a time domain channel estimation unit 52, an FFT unit 53, an SNR estimation unit 54, an MMSE equalization unit 55, and an IFFT unit 56.

  The FFT unit 51 converts the time domain received signal into a frequency domain received signal. The time domain channel estimation unit 52 performs channel estimation in the time domain from the pilot signal. The FFT unit 53 converts the channel frequency estimated by the time domain channel estimation unit 52 into the frequency domain. The SNR estimation unit 54 estimates the SNR from the estimated channel frequency response. The MMSE equalization unit 55 obtains an equalizer coefficient from the estimated SNR and channel frequency response, and performs equalization processing in the frequency domain by MMSE. The IFFT unit 56 returns the reception signal equalized by the MMSE equalization unit 55 to the time domain.

  The reception signal equalized by the frequency domain equalization unit 23 is demodulated by the demodulation unit 24, decoded by the decoding unit 25 on the channel, and output as a received bit stream.

  As described above, in the first embodiment of the present invention, the SNR estimation unit 54 of the frequency domain equalization unit 23 estimates the SNR from the channel frequency response. The fact that the SNR can be estimated from the channel frequency response will be described below.

  Take the nth data block of the received signal as an example. When the guard interval is removed by the guard interval removing unit 21, the block of length K can be described in the following vector format.

Here, y [n, l] is the lth data symbol of the nth block.

  Further simplification, excluding the index n, the equation (1) becomes as follows.

  If the FFT unit 51 performs K-point FFT, the time-point vector y is converted to a frequency-domain vector Y by the K-point FFT.

Here, Y [k] indicates a signal on the k-th subcarrier, which can be described as follows.

(4) calculation of the equation is the processing of the FFT performed by the FFT unit 51, the process, as shown in FIG. 2, the Fourier coefficient F _k by performing the process to continue to butterfly operation.

  Now, the original transmission signal block transmitted from the transmitter 1 is

Assume that

  And the corresponding frequency domain vector is

Suppose that

  Here, the original frequency domain signal and the received frequency domain signal (kth subcarrier) have the following relationship.

  Here, H [k] and W [k] are the channel frequency response of the kth subcarrier and the noise of the subcarrier, respectively. As shown in FIG. 3, the signal model of the kth subcarrier shown in the equation (7) is multiplied by the channel frequency response H [k] by the frequency domain signal X [k] of the original kth subcarrier. This indicates that noise W [k] has been added.

  Frequency domain equalization can be realized with a K-branch linear feedback equalizer having C [k] as a complex coefficient at the k-th branch (subcarrier). This linear feedback equalizer can be described in the form of zero forcing or MMSE (least mean square error).

  When optimization based on zero forcing is performed, the equalizer coefficient C [k] is

It becomes.
When optimization based on MMSE is performed, the equalizer coefficient C [k] is

  It becomes. Here, η, *, and | H | are modules of SNR, conjugate transpose, and complex value H, respectively.

  Intense frequency selective fading, where nulls (deep fading) occur in the spectrum, the inversion of H [k] at zero forcing frequency domain equalization becomes infinite, so that these nulls (deep fading) Noise increases at the frequency of the spectrum.

  In the frequency domain equalization of MMSE, intersymbol interference (in the form of mismatch between gain and phase) and noise increase can be compromised. Therefore, the mixed effect of intersymbol interference and noise is minimized. This is particularly suitable for equalizing frequency selective fading channels. In the SC-FDE system, the frequency domain equalization of MMSE is better than the frequency domain equalization of zero forcing.

  However, in the frequency domain equalization of MMSE, it is necessary to know SNR (η) as shown in Equation (9). The accuracy of SNR estimation affects the performance of MMSE in frequency domain equalization.

  In the embodiment of the present invention, the SNR is estimated based on the channel frequency response as follows. The channel frequency response is the frequency domain channel gain, which can be obtained by time domain or frequency domain channel estimation.

  Assume an M channel estimation sequence in the preamble for channel estimation. Each channel estimation sequence can simply generate an estimate of the channel frequency response vector.

  Here, H [m, k] is the channel frequency response in the k-th subcarrier estimation of the channel frequency response vector.

  Averaging the M estimates of the channel frequency response with the k th subcarrier,

It becomes.
Noise averages to zero. Therefore, the noise on the kth subcarrier in the mth channel frequency response is

It becomes.
The signal power can be estimated as follows:

The noise power can be estimated as follows.

Therefore, the estimated value of SNR is obtained from the equations (13) and (14).

  Can be obtained.

  FIG. 4 shows the configuration of the SNR estimation unit 54 that estimates the SNR from the channel frequency response as described above. In FIG. 4, the averaging unit 71 obtains M average values of the channel frequency response based on the equation (11). The subtracting unit 72 subtracts the channel frequency response and the averaged channel frequency response based on the equation (12) to obtain a noise component. The noise power calculation unit 73 obtains the noise power from the noise component from the subtractor 102 based on the equation (14). The signal power calculation unit 74 obtains the signal power from the averaged channel frequency response based on the equation (13). The dividing unit 75 obtains the SNR from the ratio of the noise power from the noise power calculation unit 73 and the signal power from the signal power calculation unit 74 based on the equation (15).

  FIG. 5 shows an MMSE equalization unit 55 that performs frequency domain equalization of MMSE. In FIG. 5, the MMSE equalization unit 55 obtains an equalizer coefficient of MMSE from the estimated values of H [k] and η based on the equation (16).

  Then, the MMSE equalization unit 55 multiplies the signal of each subcarrier from the FFT unit 51 by the equalizer coefficient obtained as described above to perform equalization processing. The IFFT unit 56 performs conversion from the equalized frequency domain signal to the time domain signal,

An equalized signal such as
The equalized time domain signal block is

  It becomes.

  6 and 7 show the evaluation in the environment of IEEE802.15.3c (60 GHz WPAN (Wireless Personal Area Network)). The multipath fading channel used in this simulation is required by IEEE 802.15.3c (60 GHz WPAN), and its specification is shown in FIG.

  The symbol rate (Nyquist bandwidth), roll-off factor, and channel width come from the channelization plan. The modulation is QPSK and the channel coding is Reed-Solomon code (RS (255, 239, 8)).

  The lengths of the FFT and the guide interval (GI) are set to 512 and 128, respectively. It is assumed that synchronization is ideal. The power amplification back-off (OBO) output is 3.0 dB and the phase noise is -93 dBc / Hz at 1 MHz. Each data packet has a data capacity of 2048 bytes.

  6 and 7, ZF indicates a case where zero-forcing frequency domain equalization is used, and MMSE indicates a case where MMSE frequency domain equalization is used. PA and PN indicate power amplification and phase noise, respectively. M is the number of channel equalization iterations, which is used for channel frequency response and SNR estimation.

  FIG. 6 shows the BER (Bit Error Rate) performance of MMSE frequency domain equalization based on SNR estimation. The effects of power amplification and phase noise are not considered at this point (PA-OFF, PN-OFF). When the channel is fully known, the MMSE frequency domain equalization process performs better than the zero forcing frequency domain equalization.

The performance of MMSE frequency domain equalization is degraded by the channel frequency response (H [k] and SNR (η)) used for estimation. At M = 2, the BER degradation of 10 ⁻⁶ is 2.4 dB, and when M = 5, it decreases to 0.8 dB. Even using the estimated channel frequency response (H [k]) and SNR (η), MMSE frequency domain equalization has better performance than zero-forcing frequency domain equalization of complete channel information.

  FIG. 7 shows BER performance considering the effects of power amplification and phase maintenance (PA-ON, PR-ON). Also in this case, it can be seen that the frequency domain equalization of MMSE has higher performance than the frequency domain equalization of the zero four signal. Performance degradation caused by estimated channel frequency response (H [k]) and SNR (η) compared to MMSE frequency domain equalization with full channel estimation under RF defects (nonlinear distortion, phase noise) Are 2.6 dB and 0.8 dB when M = 2 and M = 5, respectively.

  Simulation results suggest that MMSE frequency domain equalization based on SNR estimation is very effective even under RF defects. Furthermore, SC-FDE having such frequency domain equalization of MMSE is suitable for supporting multi-gigabit in 60 GHz WPAN (IEEE802.15.3c).

Second Embodiment
FIG. 9 shows a second embodiment of the present invention. The difference between the first embodiment and the second embodiment is that the time domain channel estimation unit 52 performs channel estimation in the time domain in the first embodiment, whereas the frequency domain channel in the second embodiment. Channel estimation is performed in the frequency domain by the estimation unit 152. Either method can generate a channel frequency response vector. About another point, 1st Embodiment and 2nd Embodiment are the same, The description is abbreviate | omitted.

  As described above, in the embodiment of the present invention, the SNR can be estimated from the channel frequency response with a simple configuration, and thus, frequency domain equalization of MMSE can be performed.

  The present invention is not limited to the above-described embodiments, and various modifications and applications can be made without departing from the gist of the present invention.

  The present invention can be widely used in communication systems that perform transmission using multipath fading channels in addition to wireless communication such as WPAN.

It is a block diagram used for description of the communication system of 1st Embodiment of this invention. It is a block diagram used for description of the process of FFT in the communication system of 1st Embodiment of this invention. It is a block diagram used for description of the transmission channel in the communication system of 1st Embodiment of this invention. It is a block diagram used for description of calculation of SNR in the communication system of 1st Embodiment of this invention. It is a block diagram used for description of the signal equalization in the communication system of 1st Embodiment of this invention. It is a graph for demonstrating the effect of 1st Embodiment of this invention. It is a graph for demonstrating the effect of 1st Embodiment of this invention. It is explanatory drawing of the item of the simulation for demonstrating the effect of 1st Embodiment of this invention. It is a block diagram used for description of the communication system of 2nd Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Transmitter 2 Receiver 11 Channel coding part 12 Modulation part 13 Guard interval insertion part 14 Transmission part 20 Reception part 21 Guard interval removal part 23 Frequency domain equalization part 24 Demodulation part 25 Decoding part 51 FFT part 52 Time domain channel estimation Unit 53 FFT unit 54 SNR estimation unit 55 MMSE equalization unit 56 IFFT unit 71 averaging unit 72 subtraction unit 73 noise power calculation unit 74 signal power calculation unit 75 division unit 152 time domain channel estimation unit

Claims (7)

In a receiving device that converts a received signal in the time domain into a frequency domain and performs equalization processing in the frequency domain,
An SNR estimator for estimating a signal to noise ratio from a channel frequency response;
A receiving apparatus comprising: an MMSE equalization unit that obtains an equalizer coefficient by a least mean square error method using the signal-to-noise ratio estimated by the SNR estimation unit and performs equalization in a frequency domain.   The receiving apparatus according to claim 1, wherein the channel frequency response is obtained from a signal in a time domain.   The receiving apparatus according to claim 1, wherein the channel frequency response is obtained from a signal in a frequency domain. In a signal equalization apparatus that converts a time domain signal into a frequency domain signal, performs equalization processing in the frequency domain, and returns to the time domain signal again and outputs it.
An SNR estimator for estimating a signal to noise ratio from a channel frequency response;
A signal equalization comprising: an MMSE equalization unit that obtains an equalizer coefficient by a least mean square error method using the signal-to-noise ratio estimated by the SNR estimation unit and performs equalization in a frequency domain apparatus. The SNR estimator is
An averaging unit that averages the channel frequency response;
A subtraction unit that subtracts the averaged channel frequency response and the channel frequency response to estimate a noise component;
A signal power calculator for calculating signal power from the averaged channel frequency response;
A noise power calculator for calculating noise power from the estimated noise component;
The division part which calculates a signal-to-noise ratio from the signal power calculated | required by the said signal power calculating part and the noise power calculated | required by the said noise power calculating part is provided. Signal equalizer. In the frequency domain equalization method of converting a time domain signal to a frequency domain signal, performing equalization processing in the frequency domain, and returning to a time domain signal again,
Estimating a signal to noise ratio from the channel frequency response;
Using the estimated signal-to-noise ratio to obtain an equalizer coefficient by a least mean square error method and performing equalization in the frequency domain. Estimating the signal to noise ratio comprises:
Averaging the channel frequency response;
Subtracting the averaged channel frequency response from the channel frequency response to estimate a noise component;
Calculating signal power from the averaged channel frequency response;
Calculating noise power from the estimated noise component;
The signal equalization method according to claim 6, further comprising: calculating a signal-to-noise ratio from the obtained signal power and the obtained noise power.