JP2008511196A - Apparatus and method for reducing phase drift - Google Patents

Abstract

  An apparatus for reducing phase drift in a spectrum of a time domain signal is a converter (101) for time-frequency converting the time domain signal into a signal converted to a frequency domain, the converted signal Is a transducer (101) representing the spectrum of the time domain signal, a detector (109) for detecting the phase drift, a detector (109) for generating a control signal indicative of the phase drift; The converter (101) further pre-processes the time-domain signal before time-frequency conversion or introduces a corrected phase drift to the converted signal, depending on the control signal. The time domain signal is time-frequency converted to reduce phase drift, and the converted signal is post-processed. , The correction phase drift is to at least partially compensate for the phase drift.

Description

  The present invention relates to the field of telecommunications, specifically the field of digital signal processing.

  Multi-carrier modulation, particularly orthogonal frequency division multiplexing (OFDM), has been applied to transmission in a wide variety of digital communication systems. In an OFDM transmitter, the transmitted OFDM symbols are obtained by mapping a group of data values to a selected mapping scheme, eg, a signal space constellation point associated with QAM (Quadrature Amplitude Modulation). It contains complex values and is converted into a so-called time-domain transmission signal using, for example, an inverse Fourier transform to obtain a transmission signal for transmission. In order to remove intersymbol interference (ISI) and maintain orthogonality of the received signal at the receiver, a cyclic prefix is inserted into the guard interval (GI) of the transmitted signal. This guard interval is preferably longer than the maximum delay of the communication channel through which the transmission signal is transmitted.

  FIG. 13 is a block diagram of a conventional OFDM receiver. First, the guard interval is removed from the signal received by the antenna. After serial-parallel conversion (S / P), a Fourier transform, for example a fast Fourier transform (FFT), is performed to obtain a received version of the multicarrier signal in the frequency domain.

ここで、

はサブキャリアｉ−１とｉとの間の位相である。位相ドリフトは、２つの隣接するサブキャリア間の位相変化の平均である。多くの場合、この位相ドリフトは可能な限り小さいものでなければならない。 When transmitting an OFDM modulated signal over a multipath fading channel, the received signal will have unknown amplitude and phase changes. The channel response after OFDM demodulation is expressed by a channel transfer function (CTF) for subcarrier i, H _i = H (f = i / T). Here, 1 / T is a subcarrier interval. The observed CTF magnitude and phase snapshots are shown in FIG. This indicates that the CTF phase φ _i = arg (H _i ) experiences a phase drift defined by

here,

Is the phase between subcarriers i-1 and i. Phase drift is the average of the phase change between two adjacent subcarriers. In many cases, this phase drift must be as small as possible.

  The phase drift in the frequency domain is caused by, for example, a frame synchronization error, that is, a timing error that introduces a phase shift into the spectral domain. This phase shift increases linearly with frequency, for example.

It should be noted that imperfect timing synchronization also results in a change in phase drift E [Δφ _i ]. In fact, if the timing offset T _off does not exceed the guard interval length T _GI minus the maximum channel delay τ _max , that is, T _off ≦ T _GI −τ _max , the loss of orthogonality of the OFDM signal is Absent. However, as described in Non-Patent Document 1, T _off ≠ 0 results in different phase drifts E [Δφ _i ].

  When using a spatial frequency code or differential modulation, performance may be degraded due to phase drift. Phase drift also degrades the performance of polynomial interpolation algorithms such as linear or spline interpolation.

  FIG. 14 shows the phase and magnitude of the channel transfer function (CTF) snapshot. FIG. 15 shows a corresponding time domain snapshot of the channel impulse response (CIR).

であり、Ｔ_offはフレームオフセットである。 The non-zero phase drift E [Δφ _i ] is common for any OFDM system. This is due to the mechanism of channel impulse response (CIR), which is the inverse Fourier transform of CTF. A CIR for generating the CTF shown in FIG. 14 is shown in FIG. As a general characteristic, the CIR is non-zero only in the range [0, τ _max ]. Where τ _max is the maximum delay of the channel, ie

T _off is the frame offset.

  Since CIR is related to CTF by Fourier transform, it can be regarded as a spectrum of CTF.

  Non-zero phase drift can degrade the performance of the OFDM system, for example, in the case of differential phase coding. This is due to the fact that phase drift introduces additional phase terms that lead to phase errors when performing differential phase demodulation, resulting in information detection errors.

  Furthermore, phase drift can introduce channel estimation errors when estimating communication channels, eg, when estimating channel transfer functions, eg, to equalize multi-carrier signals received in the frequency domain.

  For example, to estimate the channel transfer function, the OFDM transmitter can introduce so-called pilot symbols known at the receiver for channel estimation. Typically, pilot symbols are used to modulate the subcarriers of the transmitted OFDM signal. In the receiver, in order to obtain a subcarrier value containing information on the channel transfer function, the pilot symbol is derived from the modulated subcarrier by means of demodulation, for example by dividing the modulated subcarrier by the corresponding pilot symbol. Removed.

  FIG. 16 is a block diagram of channel estimation based on pilot symbols for OFDM. After the received signal is transformed to the frequency domain by means of a fast Fourier transform (FFT), a transformed signal having a value associated with the subcarrier is obtained. Here, only certain subcarriers are modulated with pilot symbols for channel estimation. A demultiplexer can be used to demultiplex (DMUX) the modulated subcarriers that are subsequently provided to a channel estimator for channel estimation to extract the modulated subcarriers. The channel estimator can perform the above pilot demodulation to extract the subcarrier values. The demodulated subcarrier value is an estimate of the channel transfer function at the frequency point associated with the subcarrier modulated by the pilot symbol. In order to obtain channel estimates for all subcarriers, the channel estimator can interpolate between two subsequent channel estimates obtained from modulated subcarriers, for example, spaced by multiple subcarriers. , Interpolation can be performed to provide channel estimates for all subcarriers. A detection unit (DT) is used to detect information contained in the converted signal bypassed by the demultiplexer. The detection unit receives a channel estimate, for example, to equalize the transformed signal. However, if phase drift occurs, the channel estimate contains errors due to additional phase terms. This leads to performance degradation when detecting the information contained in the transformed signal after equalizing the transformed signal using errored channel estimation.

  In order to reduce phase drift, in the case of channel estimation, for example, phase compensation can be performed as described in Non-Patent Document 1. In particular, after extracting the DFT and subsequent pilot signals, a least square estimation of the pilot signals is performed to provide an estimated channel transfer function. In the next step, the phase change due to frame error is determined based on the subsequent value of the channel transfer function. In the next step, the estimated phase change is removed from the estimated channel transfer function, and the resulting channel transfer function is provided to the least mean square error estimator for further channel estimation. After MMSE channel estimation, the channel transfer function of the data carrier is interpolated using linear or higher order interpolation. In the next step, a post-phase compensation is performed to provide the interpolated channel transfer function including the phase change, and the previously removed phase change is recovered.

  Non-Patent Document 2 applies a phase correction coefficient in a differential demodulation process to compensate for frame synchronization errors by performing phase rotation on a subcarrier amplitude in a frequency domain for a differentially modulated signal. Is disclosed.

M.M. Hsieh and C.M. Wei, "Channel Estimate for OFDM Systems Based on Comb-Type Pilot Arrangement in Frequency Selective Fading Channels," IEEE Transtransactions. 44, pp. 217-225, Feb. 1998 S. M.M. Weinfurtner, "Frequency-Domain Frame Synchronization for Optimum Frequency-Differential De-modulation of OFDM," in Proc. IEEE Global Telecommunications Conference (GLOBECOM '99), Rio de Janeiro, Brazil, pages 857-862, 1999.

  It is an object of the present invention to provide a concept for efficient phase drift reduction.

  29. A method for reducing phase drift according to claim 1, a multi-carrier receiver according to claim 19, a method for reducing phase drift according to claim 27, and a multi-carrier signal according to claim 28. Or a computer program according to claim 29.

  The present invention is based on the view that phase drift can be efficiently reduced when using a feedback loop phase drift reduction scheme. More specifically, to detect the phase drift, or the remaining phase drift of the signal provided by the phase drift compensation entity, and in response to the phase drift detected in the signal via the feedback loop If the detector that detects the phase drift in the signal is placed after the phase drift compensation entity to control the phase drift, the phase drift can be reduced efficiently. As a result, for example, further phase drift reduction can be realized in each compensation step.

  In accordance with the present invention, the functionality of the phase drift compensation entity is included in a converter for time frequency conversion of a time domain signal into a signal that has been converted to the frequency domain. For example, this time domain signal is a received version of a multi-carrier signal, and this received version is affected by, for example, frame synchronization errors. Thus, this transformed signal representing the spectrum of this time domain signal is subject to phase drift that affects the phase of each subcarrier.

  The detector can directly detect the phase drift in the frequency domain of the transformed signal provided by the transducer.

  Alternatively, the detector can indirectly detect phase drift in the time domain from the preprocessed time domain signal. For example, the detector can detect delays due to frame synchronization errors and calculate a delay sine, eg, phase drift caused by a Fourier transform.

  The converted signal is provided to a detector that detects the phase drift of the converted signal connected to the output of the converter. Here, the detection of the phase drift is preferably performed in the frequency domain. According to the present invention, the detector generates a control signal indicating the phase drift when detecting the phase drift. This control signal is fed back to the converter. This converter pre-processes the time domain signal in the time domain prior to the time frequency conversion or introduces a corrected phase drift to introduce a corrected phase drift into the signal converted to the frequency domain in response to the control signal. Post-process the transformed signal after time-frequency transforming the time-domain signal for direct introduction into the transformed signal. In any case, the corrected phase drift at least partially compensates for the phase drift, so that the phase drift is reduced.

  According to the present invention, the transformed signal obtained by time-frequency transforming the time domain signal, eg obtained by applying a Fourier transform to the time domain signal, can be obtained, for example, from the transformed signal. It can be post-processed in the frequency domain by manipulating the phase. As a result, corrected phase drift is introduced into the signal converted to the frequency domain. That is, the correction phase is introduced into the phase of the converted signal to reduce phase drift.

  Furthermore, the converter according to the present invention is adapted to introduce the corrected phase drift into the transformed signal obtained by transforming the time domain signal into the frequency domain after preprocessing in the time domain. It has the function of preprocessing time domain signals. In this case, for example, the time delay always introduces an additional phase shift in the spectrum of the time delay signal, so a time delay transform, for example a time delay phase shift coincidence of the Fourier transform, is used.

  As described above, the detector detects phase drift in the transformed signal, and the transformed signal transforms the time domain signal into a frequency domain signal after preprocessing the time domain signal. Or by transforming the time domain signal into the frequency domain. Therefore, the phase drift can also be reduced adaptively, where the phase drift remaining after the previous compensation step can be detected and reduced in a plurality of reduction steps.

  For example, in the first step, the detector receives a transformed signal that includes uncompensated phase drift. The detector then determines the phase drift contained in the converted signal based on the unprocessed converted signal and generates a control signal. This control signal is provided to the converter to control pre-processing and / or post-processing. For example, in response to the control signal containing information regarding the detected phase drift, the converter may pre-process or post-process, or pre-process or post-process the converted signal to reduce phase drift. Do at least one of the processes. In the next step, the detector receives a processed transformed signal containing reduced phase drift. In the next step, the detector detects the remaining phase drift contained in the (processed) converted signal provided by the converter and generates a further control signal indicating this remaining phase drift, A control signal is fed back to the converter to further reduce the remaining phase drift and the like.

  Furthermore, the phase drift can be effectively reduced when the converter applies all the processing functions according to the present invention to the reduction of the phase drift. For example, the converter coarsely reduces phase drift by means of preprocessing the time domain signal prior to time frequency conversion, and transforms the time domain signal after preprocessing to achieve accurate phase drift reduction. Thus, the resulting transformed signal can be post-processed in the frequency domain and vice versa.

  Further embodiments of the present invention will be described with reference to the following drawings.

  FIG. 1 is a block diagram of an apparatus for reducing phase drift according to the present invention according to an embodiment of the present invention.

  The apparatus shown in FIG. 1 includes a converter 101. The converter 101 has an input 103, a control input 105, and a plurality of outputs 107 connected to a plurality of inputs of the detector 109. The detector 109 has a plurality of outputs 111 and a control output 113, and the control output 113 is connected to the control input 105 of the converter 101.

  The converter 101 receives the time domain signal via the input 103. For example, if an error occurs while performing frame synchronization to obtain a time domain signal, the spectrum of the time domain signal will include a phase drift that superimposes the phase of the spectrum of the time domain signal. . For example, phase drift can be expressed as a phase ramp that increases linearly with the phase of the spectrum of the time domain signal superimposed. In this case, the phase change between two subsequent spectral values is constant.

  In order to reduce phase drift that affects the spectrum of the time domain signal that is the received version of the transmitted signal, the converter converts the time domain signal into a signal that has been converted to the frequency domain. This transformed signal represents the spectrum of the time domain signal.

  The signal converted to the frequency domain includes a plurality of spectral coefficients that are provided to detector 109 via a plurality of outputs 107 to detect the phase drift of the converted signal. The detector 109 detects the phase drift of the converted signal, generates a control signal indicative of the phase drift, and provides the control signal to the converter 101 via the control output 113. Depending on the control signal, the converter 101 either preprocesses the time domain signal before the time frequency conversion or converts the time domain signal after time frequency conversion to introduce a corrected phase drift in the converted signal. Post-process the generated signal. This corrected phase drift at least partially compensates for the phase drift to reduce the phase drift.

  In order to pre-process the time-domain signal and time-frequency convert the (pre-processed) time-domain signal to introduce the corrected phase drift into the converted signal, the converter A delay element is provided for delaying, and a correction phase shift is introduced based on the use of time delay / frequency shift correspondence.

  According to a preferred embodiment of the invention, the converter comprises a cyclic shift element as a delay element for preprocessing the time domain signal. The cyclic shift element receives a control signal from the detector 109 and cyclically shifts the time domain signal in response to the control signal to introduce a correction phase drift into the transformed signal.

  For example, the cyclic shift element cyclically shifts the time domain signal by a plurality of values, which depend on the phase drift detected by the detector. For example, the detector 109 uses the detected phase drift and determines a plurality of values to which the time domain signal is shifted by utilizing a time-frequency transform, eg, a cyclic shift or frequency shift characteristic of a Fourier transform. To do. The control signal provided to the converter can then include information about a plurality of values to which the time domain signal is shifted.

  According to a further embodiment of the invention, the detector detects a phase shift from the converted signal and provides information about the detected phase drift to the converter. The converter can determine a plurality of values to which the time domain signal is shifted in a manner similar to that described above with respect to the detector 109, depending on the received information regarding the phase drift.

  Preferably, the transformer is a Fourier transformer for transforming the time domain signal provided by the cyclic shift element into a transformed signal connected to the output of the delay element, eg the output of the cyclic shift element It has.

  In this case, the plurality of outputs 107 included in the converter 101 correspond to the plurality of outputs of the Fourier transformer, and the Fourier transformer outputs a converted signal including a plurality of spectral values. Here, the plurality of outputs 107 correspond to a plurality of spectral values. In this case, the inputs of the detector are connected to the outputs of the Fourier transformer. For example, the multiple outputs of the Fourier transformer are directly connected to the multiple inputs of the detector 109. In order to control the preprocessing of the time domain signal, the control output 113 of the detector 109 can be connected to the control input of the cyclic shift element, and the detector 109 directly controls the preprocessing of the time domain signal. To do.

  As described above, the detector 109 or the converter 101 can determine a plurality of values to which the time domain signal is shifted. For example, the plurality of values are determined based on an expected value of phase drift, and the expected value is calculated in the frequency domain by, for example, averaging the phase change between two subsequent subcarriers. To obtain a plurality of values to which the time domain signal is shifted, the expected value of phase drift is multiplied by a plurality of values contained in the transformed signal and divided by 2π. Here, the plurality of values of the converted signal correspond to, for example, a plurality of subcarriers associated with the multicarrier transmission scheme.

  For example, the cyclic shift element performs a left shift or a right shift depending on the sign of the phase drift. For example, the cyclic shift element may perform a left shift to introduce a positive correction phase drift to the converted signal to compensate for the negative phase drift, or a negative shift to compensate for the positive phase drift. A right shift is performed by introducing a corrected phase drift of

  In order to cyclically shift the time domain input signal, the cyclic shift element may comprise a shift register. The shift register has an input for receiving a time domain signal and an output connected to the input, so that a cyclic shift can be performed.

  As described above, the converter can post-process the converted signal to directly introduce a corrected phase drift in the frequency domain to compensate for the phase drift. For example, the converter comprises a phase compensator for post-processing the transformed signal in the frequency domain, which phase compensator converts the phase of the transformed signal in the frequency domain to introduce corrected phase drift. Change with. For example, the phase compensator multiplies each converted signal value by a complex value, and phase-shifts the converted signal value in the frequency domain.

  According to one embodiment of the invention, the converter 101 comprises a cyclic shift element for preprocessing the time domain signal or a phase compensator for post processing the converted signal. In the latter case, the converter may comprise a Fourier transformer that performs time-frequency conversion of the time domain signal. In this case, the phase compensator is connected to one or more outputs of the Fourier transformer and the detector 109 is connected to one or more outputs of the phase compensator. Only receives the converted signal provided by the phase compensator, which receives the control signal generated by the detector 109 in detecting the phase drift of the converted signal. In order to receive the control signal, the phase compensator can comprise a control input for receiving the control signal from the detector 109. For example, the control input of the phase compensator is directly connected to the control input 105 of the converter 101.

  The compensator can add a corrected phase drift to the phase of the converted signal to compensate for the negative phase drift, or subtract the corrected phase drift from the phase of the converted signal to compensate for the positive phase drift. it can.

  According to one embodiment of the present invention, the corrected phase drift introduced by the phase compensator can be determined by the detector 109 in detecting the phase drift. Further, the corrected phase drift can be determined by the transducer 109 in response to information regarding the phase drift provided by the detector 109. For example, the corrected phase drift introduced by the phase compensator corresponds to the phase drift detected by the detector 109, except for the sign. According to a further aspect of the invention, the converter includes a cyclic shift element for preprocessing the time domain signal and a phase for post processing the converted signal to reduce phase drift in two stages. And a compensator. For example, the cyclic shift element cyclically shifts the time domain signal to coarsely reduce phase drift. The time domain signal provided by the cyclic shift element is then transformed into the frequency domain, for example by a Fourier transformer connected to the cyclic shift element, to obtain a transformed signal. The phase compensator is connected to one or more outputs of a Fourier transformer for post-processing the transformed signal, the one or more outputs of the phase compensator detecting phase drift Is connected to a detector 109. Preferably, the detector 109 controls the cyclic shift element and the phase compensator simultaneously to reduce phase drift.

  For example, the cyclic shift element receives a control signal from the detector. In this case, the control signal represents a plurality of values to which the time domain signal is shifted in order to introduce a corrected phase drift into the converted signal to roughly compensate for the phase drift. In response to the control signal, the cyclic shift element can cyclically shift the time domain signal by multiple values to roughly compensate for the phase drift.

  Thus, the phase compensator can receive an additional control signal from the detector 109, which further control phase is introduced into the phase of the converted signal to compensate for the phase drift accurately. Represents drift. Depending on the further control signal, the phase compensator can introduce further correction phase drift into the phase of the converted signal.

  It should be noted that the control signal and further control signals can be included in a common control signal generated by detector 109 based on detection of phase drift affecting the converted signal.

  As described above, multiple values to which the time domain signal is shifted, or a corrected phase drift added to the phase of the transformed signal, indicates the phase drift detected by the detector 109 or upon detection of the phase drift. It can be determined by the converter 101 when the control signal is received.

  In order to detect the phase drift, the detector 109 can determine an average of the phase change of two subsequent values of the transformed signal, the average of the phase change representing the phase drift.

  For example, the transformed signal can include all spectral values that result from transforming a time domain signal to the frequency domain. According to a further aspect of the invention, the transformed signal may consist of a series of spectral values associated with subcarriers modulated by pilot symbols for channel estimation at the transmitter. For example, the subsequent two subcarriers modulated by pilot symbols can be spaced by a plurality of subcarriers used for example for transmission of information. In this case, the detector detects the phase drift from the spectral values associated with the subcarriers modulated by the pilot symbols. In order to extract the modulated subcarriers, the converter can further comprise a selector, which is a spectrum obtained by time-frequency transforming the time domain signal of the subcarriers modulated by the pilot symbols. Select from values. In this case, a series of spectral values representing subcarriers modulated by pilot symbols constitute a transformed signal. To demodulate the modulated subcarriers, the converter can further demodulate a series of spectral values that make up the transformed signal using pilot symbols, and as an example, each spectral value associated with it. Divide by pilot symbol. Alternatively, each spectral value can be multiplied by a complex conjugate of the associated pilot symbol.

  To determine the phase drift, the detector 109 can determine the average value of the product of the two subsequent values of the transformed signal, where one of these values is a complex conjugate. Can do. For example, detector 109 may determine an average value of the product of a value associated with a subcarrier and a complex conjugate of the value associated with the subcarrier preceding this subcarrier.

  In order to determine the phase drift, the detector 109 can determine the phase of the average value, the phase of the average value representing the phase drift. If the transformed signal contains a series of values associated with spaced subcarriers, the average phase is preferably defined by the interval, which is used for the transmission of pilot symbols. A plurality of subcarriers separating the subsequent two subcarriers are shown.

  According to a further aspect of the invention, the detector determines an average of the phase change of two values of the transformed signal that are spaced apart by a plurality of values, and a phase for each value of the transformed signal. The average of the phase change can be divided by multiple values to detect the shift. This case corresponds to the above case where, for example, only a certain subcarrier is used for transmission of pilot symbols. Unlike the above embodiment, the transmitter does not include a selector at all. However, phase drift can also be derived from the spectral values of the transformed signal that are not associated with the subcarriers used to transmit the pilot symbols. In either case, complexity can be reduced because only a certain subset of the values of the transformed signal will be used to detect the phase drift.

  The present invention further provides a multi-carrier receiver comprising an apparatus for reducing the phase drift as described above.

  Preferably, the apparatus for reducing phase drift reduces phase drift of a receivable multicarrier signal in the time domain and provides a transformed signal representing the receivable multicarrier signal in the frequency domain, The converted signal has reduced phase drift. In other words, the apparatus for reducing phase drift according to the present invention is based on a multi-carrier reception scheme, for example, OFDM, to convert a multi-carrier signal receivable in the time domain to the frequency domain and simultaneously reduce the phase drift. Used.

  Furthermore, the multicarrier receiver may comprise means for extracting information contained in the converted signal.

  According to a further aspect of the invention, the multicarrier receiver may comprise a synchronizer for performing frame synchronization in the time domain. Here, the phase drift results from a frame synchronization error due to an incorrect sample time taking into account the time to determine the start of the frame in the time domain.

  In order to reduce the frame synchronization error, the detector 109 provided in the device for reducing the phase drift may generate a frame synchronization control signal indicating the phase drift and provide the frame synchronization control signal to the synchronizer. it can. The synchronizer can adjust the sample time according to the frame synchronization control signal in order to reduce frame synchronization errors. Thus, the detector can also control the operation of the synchronizer and the frame synchronization error is reduced in response to the detected phase drift, which synchronizer is a converter according to the invention for performing frame synchronization in the time domain. Placed in front of.

  In order to reduce the remaining phase drift in the frequency domain, the ability of the device to reduce phase drift can be utilized.

  According to a further aspect of the invention, the means for extracting information comprises differential demodulation for differentially demodulating the converted signal to extract information based on a change in the phase of the subsequent two values of the converted signal. Can be provided.

  According to a further aspect of the invention, the means for extracting information may comprise a low pass filter for filtering the transformed signal in the frequency domain. Preferably, the low pass filter has a real coefficient representing the real part of the filter response and the filter coefficient representing the imaginary part of the filter response is set to zero. In this case, the converter further preprocesses the receivable multi-carrier signal in the time domain or introduces a phase shift in the converted signal to shift the spectrum of the converted signal with respect to the passband of the low pass filter. In order to post-process the converted signal.

  Thus, frequency domain filtering using a real-valued filter with a symmetric double-sided filter transfer function can be performed, which further reduces the complexity of the receiver.

  According to a further aspect of the present invention, the low-pass filter provided in the means for extracting information can perform channel estimation to extract channel information, for example, a receivable multicarrier signal in the time domain It is a receivable version of the transmission signal transmitted through the network. For example, the low-pass filter is a Wiener filter for channel estimation.

  According to a further aspect of the invention, if the estimate of the channel transfer function in the frequency domain is associated with a subcarrier that is spaced apart by a plurality of subcarriers, the low pass filter may estimate the channel transfer function. It is an interpolation filter for interpolating between. In this case, the interpolation process is performed to obtain an estimate of the intermediate value of the channel transfer function.

  According to a further aspect of the invention, when spatial frequency block coding is used at the receiver, the means for extracting information comprises a spatial frequency block decoder for spatial frequency block decoding the signal transformed to the frequency domain. Yes.

  Hereinafter, further embodiments of the present invention will be described with reference to FIGS.

  FIG. 2a shows an OFDM receiver with phase compensation after Fourier transform. The receiver includes an antenna 201 connected to a guard interval removing unit 203, and the guard interval removing unit 103 is connected to a serial / parallel converter (S / P) 205. The serial parallel converter 205 has a plurality of outputs connected to a plurality of inputs of a Fourier transformer 207, which performs a fast Fourier transform (FFT). The Fourier transformer 207 has a plurality of outputs connected to a phase compensator 209, which has a plurality of inputs connected to a detector not shown in FIG. 2a. ing. It should be noted that the Fourier transformer 207 and the phase compensator 209 are provided in the above-described converter according to the present invention.

  FIG. 2 b is a block diagram of an OFDM receiver with a cyclic shift before performing a Fourier transform by the Fourier transformer 207. The cyclic shift is executed by a cyclic shift element 301 connected between the guard interval removing unit 203 and the serial / parallel converter 205. In this embodiment, the cyclic shift element 301, the serial parallel converter 205, and the Fourier transformer 207 constitute a converter according to the present invention, and a plurality of outputs of the converter 207 are connected to a detector. This is not shown in FIG. 2b. In addition, the detector can control the phase compensator 209 and the cyclic shift element 301, which are not shown in FIGS. 2a and 2b.

  The scheme as shown in FIGS. 2a and 2b is applicable to a wide range of OFDM receivers because of the structure conforming to the OFDM standard. Furthermore, if phase compensation is applied based on the embodiment of FIG. 2a, unlike the prior art approach described above, there is no need to compensate for the induced phase shift after channel estimation and interpolation.

  In particular, the negative effects of non-zero phase drift based on (1) can be compensated by the phase compensation unit after the FFT in the OFDM receiver, as shown in FIGS. 2a and 2b.

Another possibility to solve the problem with the one-sided spectrum of the frequency response is to match the filter response to the observed characteristics of the OFDM signal. This is illustrated in FIG. 5, where the passband of the channel estimation filter is selected within the range [0, τ _max ]. However, this will always result in complex-valued filter coefficients.

  In order to obtain a channel estimate with a filter having real-valued coefficients, the above-described approach according to the present invention can be applied. Here, cyclic shift at the receiver or phase compensation at the receiver can be applied. Cyclic shifts at the receiver are I. Rahman, K .; Witrisal, D.W. Prasad, O.I. Olsen and R.W. “Performance Comparison between MRC Receiver Diversity and Cyclic Delay Diversity in OFDM WLAN Systems” by Prasad, in Proc. Int. Symposium on Wireless Personal Multimedia Communications (WPMC'03), Yokosuka, Japan, Oct. 2003 and A.I. Dammann and S.M. “Standard Standard Antenna Diversity Techniques for OFDM and its Application to the DVB-T System” by Kaiser, in Proc. IEEE Global Telecommunications Conference (GLOBECOM 2001), San Antonio, TX, USA, pages 3100-3105, Nov. 2001. Here, the cyclic shift is applied on the receiver side in order to use spatial diversity. However, according to the present invention, a cyclic delay is inserted not to use spatial diversity but to reduce phase drift or to obtain a filter with real-valued filter coefficients.

FIG. 2b shows the cyclic shift of the signal received before the FFT. Cyclic shifting the signal received before the FFT by the sample −δ _cyc results in the following phase shift after the FFT.

  FIG. 3a shows the phase of the channel transfer function (CTF) after cyclically shifting the time domain signal before FFT. FIG. 3b shows the corresponding magnitude of the channel transfer function.

である。δ_cycが（２）に基づいて選択されると、受信した信号に対する効果は、ＦＦＴの後の位相補償（図２ａ）またはＦＦＴの前のサイクリックシフト（図２ｂ）のどちらを選択するかに関わらず同じである。 The effect of cyclic shift on the received signal is

It is. When δ _cyc is selected based on (2), the effect on the received signal is whether to select phase compensation after FFT (FIG. 2a) or cyclic shift before FFT (FIG. 2b). Regardless, it is the same.

を１回乗算することが必要である。他方、ＯＦＤＭ受信機全体の複雑度を考えると、サブキャリアごとにさらに１回の乗算はそれほど重大ではない場合もある。 In contrast to the cyclic shift as shown in FIG. 2b, the phase shift based on FIG. 2a is more complicated to calculate. Cyclic shift can be performed very efficiently using a shift register of sample-δ _cyc , whereas phase compensation of θ _cyc degree per subcarrier is performed for each subcarrier.

Must be multiplied once. On the other hand, considering the complexity of the entire OFDM receiver, one more multiplication per subcarrier may not be as critical.

  It should be noted that the cyclic shift only changes the phase of the CTF and the magnitude is not affected. This can be confirmed by comparing the CTF of the OFDM signal without cyclic shift shown in FIGS. 14, 3a and 3b with OFDM with cyclic shift. Since the influence of the frequency selective channel is compensated by the channel estimator, no other action is necessary.

A snapshot of the CTF and corresponding CIR magnitude and phase after cyclic shifting the received signal is shown in FIGS. 3a, 3b and 4, respectively. Phase drift E [Δφ _i ] is compensated for while there are still large variations in amplitude and phase due to frequency selective fading. The effective CIR in FIG. 4 is shifted in the direction of negative delay. The received signal has a two-sided spectrum rather than a one-sided spectrum.

  A cyclic shift on the receiver side can also be applied to channel estimation based on discrete cosine transform (DCT). Since DCT operates in a double-sided spectrum, a cyclic shift before OFDM demodulation is very useful.

が用いられ、ここで、

である。次元Ｍ_fのＦＩＲフィルタは以下の一般的な形式で表現できる。

Channel estimation using the pilot symbols (pilot-symbol aided channel estimation: PACE) For a known symbol (pilot) are inserted at regular intervals D _f subcarriers. Filtering and interpolation between pilots is necessary to reconstruct the signal. Received signal after OFDM demodulation at pilot position for channel estimation

Where:

It is. An FIR filter of dimension M _f can be expressed in the following general form.

If the symmetry of the spectrum W _m is such that the real part and the imaginary part are an even function and an odd function, respectively, the coefficients of either or both of the FIR interpolation and the smoothing filter are real values. In general, the computational cost of a real value filter is half that of a complex value filter.

  P. "Pilot-Symbol-Aided Channel Estimate in Time and Frequency" by Hoeher et al., In Proc. Communication Theory Mini-Conference (CTMC) with IEEE Global Telecommunications Conversation (Glovecom 97), Phoenix, AZ, USA, pp. 90-96, 1997 describes PACE by Wiener filtering.

が、様々な電力遅延プロファイル（ｐｏｗｅｒ ｄｅｌａｙ ｐｒｏｆｉｌｅ）をカバーするように設計される。従って、最大遅延がＴ_Wである矩形型の電力遅延プロファイルはこの要件を満たす。［０，Ｔ_W］の範囲内で非ゼロである均一な電力遅延プロファイルのフーリエ変換により、以下の周波数相関が得られる。

According to a further aspect of the invention, when a model mismatch Wiener interpolation filter (WIF) is selected, the filter

Are designed to cover a variety of power delay profiles. Therefore, the rectangular type power delay profile of the maximum delay is T _W meet this requirement. The following frequency correlation is obtained by Fourier transform of a uniform power delay profile that is non-zero within the range [0, T _W ].

Due to the complex phase term in R _HH [Δi], the corresponding Wiener filter also has complex coefficients.

  However, the complexity of filtering or channel estimation when using a filter with complex coefficients does not increase much compared to filtering or channel estimation using filters with only real-valued coefficients.

となるように選択されると、有効ＣＩＲは、チャネル推定フィルタのパスバンド内でのみ非ゼロとなる。ローパス特性を有し、［−Ｔ_W／２，Ｔ_W／２］の範囲内で対称的な両側パスバンドを有するフィルタを前提とすると、サイクリックシフトされた信号の有効ＣＩＲは、図４に示されているように歪みのないフィルタを通過する。チャネルの最大遅延τ_maxが推定できる場合、Ｔ_WはＴ_W＝τ_max＋Δ_Wに基づいて選択できる。ここで、Δ_Wはロールオフ遅延であり、これはフィルタ応答の不完全なスロープに挿入することができる。他方、τ_maxが不明の場合、Ｔ_GIによって上限が定められ、つまりＴ_W＝Ｔ_GIとなる。 Cyclic shift without CIR is [0, τ _max] Assuming that the range of, by cyclic shifts [delta] _cyc showing a left _{shift, [- δ cyc, τ max} -δ cyc] non-zero within the range of An effective CIR is obtained. Cyclic shift

The effective CIR is non-zero only in the passband of the channel estimation filter. Assuming a filter having a low-pass characteristic and a symmetric double-side passband within the range [−T _W / 2, T _W / 2], the effective CIR of a cyclically shifted signal is shown in FIG. Pass through undistorted filter as shown. If the maximum delay τ _{max of the} channel can be estimated, T _W can be selected based on T _W = τ _max + Δ _W. Where Δ _W is the roll-off delay, which can be inserted into the incomplete slope of the filter response. On the other hand, if tau _max is unknown, the upper limit is defined by the T _GI, i.e. a T _W = T _GI.

を生成するために実数値の周波数相関関数を使用することができる。 Next, the filter coefficient

A real-valued frequency correlation function can be used to generate

Accordingly, the WIF that matches the uniform power delay profile [−T _W / 2, T _W / 2] is also a real value. This means that the calculation cost can be reduced to half.

Instead of a mismatched WIF, any FIR low-pass interpolation filter benefits from the proposed cyclic shift. As long as the filter matches the passband of [−T _W / 2, T _W / 2], the filter is a real value. For such a filter, the performance is optimal if the signal being filtered passes through an invariant filter.

  FIG. 5 shows an OFDM receiver with phase compensation after FFT, according to a further embodiment of the invention.

Unlike the embodiment of FIG. 2 a, the OFDM receiver shown in FIG. 5 includes a phase compensator 501 having a plurality of outputs connected to a detector 503. This detector 503 has a control output 505 connected to the control input of the phase compensator to provide information regarding the detected phase drift Θ _cyc .

  FIG. 6 shows an OFDM receiver with cyclic shift before FFT. Unlike the embodiment shown in FIG. 2b, the OFDM receiver shown in FIG. 6 includes a cyclic shift element 601 connected between the guard interval removing means 203 and the serial-parallel converter 205, and a Fourier And a detector 603 connected to the plurality of outputs of the converter 207. This detector 603 has one control output 605 connected to the control input of the cyclic shift element 601 in order to provide information about the multiple values to which the domain signal provided by the guard interval removal means 203 is shifted. It has.

Depending on the application, the cyclic shift can be selected to compensate for the phase drift of (1), ie:

An OFDM system with differential modulation or a spatial frequency coded OFDM system has improved performance if E [Δφ _i ] is well estimated.

In order to incorporate this phase drift compensation into the proposed receiver, an adaptive realization as shown in FIG. 5 seems attractive. In the case of the proposed receiver that performs cyclic shift before FFT, it can be realized as shown in FIG. First, the cyclic shift δ _cyc can be set to a default value in the range [0, T _GI / 2]. Then, _δcyc is estimated and can be fed back to the cyclic shift unit. It should be noted that the phase drift can be expected to change only on a long-term basis and does not require frequent updates.

  Hereinafter, the performance of the approach according to the present invention for the spatial frequency block code (SFBC) will be described.

In order to obtain a BER curve, an OFDM system using spatial frequency block coding (SFBC) with N _T = 2 transmit antennas has been realized. An OFDM system with N _T = 2 transmit antennas and N _R = 1 receive antenna is used. The system parameters for the OFDM system and channel model are shown in FIG. The channel is modeled by a tap delay line model with Q ₀ = 12 taps, and with a rapidly decaying power delay profile as shown in FIG. 8, the tap spacing is Δτ = 16 · T _spl . The total transmission power of the system is determined so that the total transmission power of the system with N _T antennas is equal to the system with a single antenna. External channel coding is not used.

For the system considered here, it can be seen that the BER floor can be reduced somewhat by optimizing the cyclic shift δ _cyc . For the parameters considered here, the optimal cyclic shift is about 44T _spl , which gives the best performance. It can be seen that the accuracy of _δcyc need not be high.

According to the present invention, the performance of a polynomial interpolator can be greatly optimized. In the case of PACE, interpolation in the frequency direction and the time direction is necessary. In the time direction, the Doppler power spectrum generally has a (at least approximately) symmetrical bilateral real-valued profile, whereas the power delay profile is real but unilateral. Hereinafter, the advantage of inserting a cyclic shift [delta] _cyc, described linear interpolator. This result also applies to higher order polynomial interpolators.

FIG. 9 shows a comparison of BER and E _b / N _{0 for} N _T = 2 and various cyclic shifts δ _cyc .

According to the present invention, the performance of a polynomial interpolator can be greatly optimized. In the case of linear interpolation, two consecutive pilot subcarriers are used to determine the channel response of subcarriers placed between these two pilots. For subcarrier i, the channel estimate is given by:

により表現できる。 It is useful to represent linear interpolation with FIR filters based on (3). Linear interpolator has filter response

Can be expressed by

を（３）に基づくＦＩＲフィルタの式に導入すると、線形補間が得られる。線形インタポレータに伴う歪みは、（８）のスペクトル、つまり以下の式

によって与えられる

の逆フーリエ変換により評価できる。

Is introduced into the FIR filter equation based on (3), linear interpolation is obtained. The distortion associated with the linear interpolator is the spectrum of (8), that is,

Given by

Can be evaluated by inverse Fourier transform.

は補間の性能を改善することになる。 The spectrum of the linear interpolator is depicted in FIG. First, a high oversampling rate is required compared to an ideal low pass filter. Second, to obtain the best performance, the spectrum of the interpolated signal must be concentrated around τ = 0. Since the CIR is zero only within the range [0, τ _max ], the cyclic shift based on (7)

Will improve the performance of interpolation.

  FIG. 10 shows a snapshot of the time domain channel impulse response after cyclic shifting the signal before FFT. FIG. 10 also shows the frequency response of the linear interpolator.

FIG. 11 shows a comparison of MSE and SNR for a polynomial interpolation algorithm with a cyclic shift of δ _cyc = 0 (dashed line) and δ _cyc = 44 (solid line).

  The performance of the approach according to the present invention for some polynomial interpolation algorithms will be described below. In FIG. 13, MSE is depicted in terms of SNR for various polynomial interpolation algorithms. It can be seen that linear interpolators with cyclic shifts have a much lower error floor. The same applies to the spline interpolator, which is close to an optimal Wiener filter when a cyclic shift is inserted.

  In the following, further advantages associated with the inventive concept will be described.

The application of the cyclic shift proposed in the OFDM receiver is, for example, as follows.
The first is to provide real value filter coefficients in the case of PACE. This reduces the number of multiplications required per channel estimation by half while maintaining the same performance.
The second is to improve the performance of the polynomial interpolation algorithm. The error floor due to the interpolation error of the polynomial interpolation algorithm can be greatly improved.
The third is to improve the performance of differentially modulated signals and spatial frequency codes.

  Compared with phase shift insertion, which requires one multiplication of a complex number for each subcarrier, it is very easy to perform a cyclic shift. Furthermore, post processing following either or both of channel estimation and demodulation is not necessary. For example, M.M. Hsieh and C.I. Wei's "Channel Establishment for OFDM Systems Based on Comb-Type Pilot Arrangement in Frequency Selective Fading Channels", IEEE Transacton Cosns. 44, pp. 217-225, Feb. The scheme described in 1998 requires further multiplication for each subcarrier after interpolation.

  Hereinafter, orthogonal frequency division multiplexing (OFDM) will be described.

ここで、ｎ（ｔ）は加算性ホワイトガウスノイズ（ａｄｄｉｔｉｖｅ ｗｈｉｔｅ Ｇａｕｓｓｉａｎ ｎｏｉｓｅ）であり、Ｎ_sym＝Ｎ_FFT＋Ｎ_GIは、ＯＦＤＭシンボルごとのサンプル数を表している。受信機においてガードインターバルが除去され、信号サンプルの受信ブロックに対してＤＦＴを実行することにより情報を回復し、ＯＦＤＭ復調の出力Ｙ_l,iを得る。ＯＦＤＭ復調後の受信した信号は以下の式により与えられる。

ここで、Ｘ_l,iおよびＨ_l,iはそれぞれ、第ｌ番目のＯＦＤＭシンボルのサブキャリアｉにおける送信された情報シンボルとチャネル伝達関数（ＣＴＦ）を表している。項Ｎ_l,iは、ゼロ平均および分散Ｎ₀を有する加算性ホワイトガウスノイズ（ＡＷＧＮ）を表している。送信信号は、各々がＮ_C個のサブキャリアを有するＬ個のＯＦＤＭシンボルからなると想定できる。 For OFDM, the signal stream is typically divided into N _c parallel substreams for any multi-carrier modulation scheme. The i-th substream of the l-th symbol block called an OFDM symbol, generally called a subcarrier, is represented by X _{l, i} . An inverse DFT with N _FFT points is performed for each block, after which a guard interval with N _GI samples is inserted to obtain _{xl, n} . After D / A conversion, the signal x (t) is transmitted via the mobile radio channel having the response h (t, τ). _Assuming perfect synchronization, the received signal of the equalized baseband system at the sampling instant t = [n + _{1N sym} ] T _spl is of the form

Here, n (t) is additive white Gaussian noise, and N _sym = N _FFT + N _GI represents the number of samples for each OFDM symbol. The guard interval is removed at the receiver, and information is recovered by performing DFT on the received block of signal samples to obtain an OFDM demodulated output Y _{l, i} . The received signal after OFDM demodulation is given by:

Here, X _{l, i} and H _{l, i} represent the transmitted information symbol and channel transfer function (CTF) on the subcarrier i of the l-th OFDM symbol, respectively. The term N _{l, i} represents additive white Gaussian noise (AWGN) with zero mean and variance N ₀ . It can be assumed that the transmitted signal consists of L OFDM symbols each having N _C subcarriers.

  The following focuses on channel estimation in the frequency direction (subcarrier index i). Therefore, the index l representing the OFDM symbol can be reduced as follows.

であると一般的に想定できる。 Consider a time-variant frequency selective fading channel modeled by a tap delay line with Q ₀ non- _zero taps. It can generally be assumed that the channel is time limited by the maximum delay τ _max . The channel impulse response (CIR) h (t, τ) is zero outside the range of [0, τ _max ]. The channel impulse response (CIR) is almost constant over one OFDM symbol, so that the time dependence of CIR within one OFDM symbol can be reduced, ie

It can generally be assumed that

The channel transfer function (CTF) of (11) is the Fourier transform of CIR h ⁽ μ ⁾ (τ). Sampling the result at frequency f = i / T, the CTF on subcarrier i is defined by H _i = H (f = i / T). Here, H (f) is an analog CTF, and T _sym = (N _FFT + N _GI ) T _spl and T = N _FFT T _spl are an OFDM symbol period having a guard interval and an OFDM symbol period having no guard interval. Respectively.

である場合、ＯＦＤＭ復調後の受信機における直交性は維持され、（１１）に基づく受信した信号が得られる。 If the guard interval is longer than the maximum delay of the channel, that is

, The orthogonality in the receiver after OFDM demodulation is maintained and the received signal based on (11) is obtained.

  Hereinafter, an OFDM receiver having a cyclic shift on the receiver side will be described.

In this section, the effective channel model after cyclic shift before FFT is considered. After the cyclic prefix is removed, the received signal based on (10) is cyclic delayed by the sample -δ _cyc .

の等化なＣＩＲを表す有効ＣＩＲを定義することが有用である。 Cyclic shifted signal

It is useful to define an effective CIR that represents the equivalent CIR of

Whereas the maximum delay tau _max of the channel does not change, effective CIR h ^[cyc] (τ) _{is, [- δ cyc, τ max} -δ cyc] is non-zero in the range of.

The FFT converts the cyclic delay into a phase shift. The effective CTF of the cyclic receiver is expressed by the following equation.

Next, the cyclic shifted received signal after OFDM demodulation can be expressed by the following equation.

  Hereinafter, the principle of channel estimation using pilot symbols for OFDM will be described.

を有するパイロット

のみを含む受信した信号のシーケンスのサブセットを定義することが有用である。ゆえに、パイロットシーケンスは周波数方向にＤ_f倍低いレート

で送信される。（一般的な慣例として、パイロットシンボルを表す変数は以下、〜とマークする。）パイロット

がＰＳＫコンステレーションから選択され、つまり

と想定できる。 In the case of channel estimation (PACE) using pilot symbols, a known symbol (pilot) is multiplexed into the data stream, which is used as side information for estimating the channel. PACE is first introduced for single carrier systems and requires a flat fading channel. To explain channel estimation using pilot symbols,

Pilot with

It is useful to define a subset of the sequence of received signals that contains only Therefore, the pilot sequence is D _f times lower in the frequency direction

Sent by (As a general convention, the variables representing pilot symbols are marked with.

Is selected from the PSK constellation, ie

Can be assumed.

を得る。ここでＧは、パイロットを含んだＯＦＤＭフレームのサブセットである。 After OFDM demodulation, the received signal Y _{i of} (11) is obtained. For channel estimation, the received signal at the pilot position is demultiplexed based on the data stream and the received pilot sequence

Get. Here, G is a subset of the OFDM frame including the pilot.

  Hereinafter, OFDM channel estimation by FIR filtering will be described.

、ただし、

におけるＣＴＦの初期推定を提供する。 The first step in the channel estimation process is to remove the pilot symbol modulation, which is the pilot position.

However,

Provides an initial estimate of the CTF at.

を使用して、チャネル推定

を得る。ここで、Ｍ_fはフィルタ次数、つまりＦＩＲフィルタＷ_mの係数の数を表している。 The channel estimator is a demodulated pilot based on (17)

Use to estimate the channel

Get. Here, M _f represents the filter order, that is, the number of coefficients of the FIR filter W _m .

は例えば、ローパス補間フィルタ、多項式インタポレータまたはウィーナー補間フィルタとして実現できる。ウィーナー補間フィルタは、所望のレスポンスＨ_iと観察されたものとの間、つまり受信したパイロットシンボル間の平均二乗誤差（ＭＳＥ）を最小化する。これは、チャネル統計に関する情報が必要であることを意味している。対照的に、ローパス補間フィルタおよび多項式インタポレータは、チャネル統計に関するなんらの情報も前提としていない。 FIR filter

Can be implemented as, for example, a low-pass interpolation filter, a polynomial interpolator, or a Wiener interpolation filter. The Wiener interpolation filter minimizes the mean square error (MSE) between the desired response _Hi and what is observed, i.e., between received pilot symbols. This means that information about channel statistics is needed. In contrast, low-pass interpolation filters and polynomial interpolators do not assume any information about channel statistics.

  For multicarrier systems, the observed channels are generally correlated in two dimensions, frequency and time. Furthermore, it can be extended to two-dimensional PACE.

  Hereinafter, Wiener filtering will be described.

を解くことによって得られる。 The Wiener interpolation filter (WIF) can be realized by an FIR filter having M _f taps based on (18). WIF W '[Δi] is the Wiener Hoff equation

Is obtained by solving

In order to generate a WIF, information about the autocorrelation matrix and the cross-correlation vector is required.

The components of the mth row and the nth column of the CTF autocorrelation matrix at the pilot position are given by the following equations.

Assuming that the channel can be expressed by a tap delay line, the frequency correlation R ′ _HH [(mn) D _f ] between subcarriers having an interval of Δf = Δi / T Hz is as follows.

の第ｍ番目のエントリは以下のように表すことができる。

Cross correlation vector

The mth entry of can be expressed as:

  The mismatch estimator according to the present invention will be described below.

の周波数相関関数を提供する。ミスマッチ推定器は、（６）に基づく

を（２１）および（２３）へ代入することによって決定できる。そしてウィーナーホッフ方程式（１９）は一度だけ決定できればよい。ミスマッチ推定器を使用することによって、フィルタ係数を事前に計算し、記憶することができる。 For WIF, autocorrelation and cross-correlation functions need to be estimated at the receiver. The filter coefficient may not be estimated during real-time operation. Alternatively, a robust estimator of model mismatch can be selected. The filter W is designed to cover various power delay profiles. For example, a rectangular power delay profile with a maximum delay T _W meets this requirement. This assumption is based on the mismatch estimator based on (6)

Provides a frequency correlation function of The mismatch estimator is based on (6)

Can be determined by substituting (21) and (23). The Wiener Hoff equation (19) need only be determined once. By using a mismatch estimator, the filter coefficients can be pre-calculated and stored.

でなければならない。チャネル推定器を決定するためには、Ｔ_W、Ｆ_Wおよびγ_Wのみでよい。チャネルの最大遅延τ_maxが不明であれば、ガードインターバル期間Ｔ_GIによって上限を設定することができる。フィルタはサンプリング定理も満たさなければならないため、フィルタパスバンドは以下の範囲内で選択できる。

It is important to note that the robust estimator parameters must always be greater than or equal to the worst-case channel conditions, ie, the mobile user's propagation delay and maximum predicted rate. Furthermore, the average SNR at the filter input γ _W used to generate the filter coefficients must be greater than or equal to the actual average SNR. That is

Must. Only T _W , F _W and γ _W are required to determine the channel estimator. If the maximum channel delay τ _max is unknown, the upper limit can be set by the guard interval period T _GI . Since the filter must also satisfy the sampling theorem, the filter passband can be selected within the following range.

  Hereinafter, estimation of phase drift will be described.

であり、これは十分に正確な推定を提供する。Ｅ［Δφ_i］を推定するためのより高度なアルゴリズムは、Ｓ．Ｋａｙによる「Ａ Ｆａｓｔ ａｎｄ Ａｃｃｕｒａｔｅ Ｓｉｎｇｌｅ Ｆｒｅｑｕｅｎｃｙ Ｅｓｔｉｍａｔｏｒ」、ＩＥＥＥ Ｔｒａｎｓａｔｉｏｎｓ ｉｎ Ａｃｏｕｓｔｉｃｓ，Ｓｐｅｅｃｈ ａｎｄ Ｓｉｇｎａｌ Ｐｒｏｃｅｓｓｉｎｇ」、ｖｏｌ．３７，ｐｐ．１９８７〜１９９０，Ｄｅｃｅｍｂｅｒ １９８９に記載されている。 The phase drift E [Δφ _i ] determines the average of the phase change between two adjacent subcarriers defined in (1). One possibility to estimate phase drift is

This provides a sufficiently accurate estimate. A more sophisticated algorithm for estimating E [Δφ _i ] is S.E. “A Fast and Accurate Single Frequency Estimator” by Kay, IEEE Transactions in Acoustics, Speech and Signal Processing, vol. 37, pp. 1987-1990, December 1989.

によって容易に生成できる。ここで、

はＸ_iの硬判定である。（２５）の精度は、ノイズおよび判定フィードバックの効果により低下する。幸い、

は特別正確である必要はない。 Clearly, CTF H _i can not be used at the receiver. Instead, the noise estimation of H _i (noisy estimate) is

Can be easily generated. here,

Is a hard decision of X _i . The accuracy of (25) decreases due to the effects of noise and determination feedback. Fortunately,

Need not be particularly accurate.

ここで、

は（１７）で定義されたものである。パイロットはＤ_f個のサブキャリアの間隔を持つため、つまりＤ_f倍の位相ドリフトが推定されるため、Ｄ_fによる分割が必要になる。 For example, if pilot symbols are transmitted for channel estimation, the phase drift can be estimated by the following equation:

here,

Is defined in (17). Since the pilot has an interval of D _f subcarriers, that is, a phase drift of D _f times is estimated, division by D _f is necessary.

  FIG. 12 shows a further embodiment of an OFDM receiver according to the present invention.

  Unlike the embodiment of FIG. 2a, the OFDM receiver shown in FIG. 12 comprises a demultiplexer 1201 for demultiplexing subcarriers modulated by pilot symbols for channel estimation. Demultiplexer 1201 (DMUX pilot) can be configured to demodulate subcarriers modulated by pilot symbols. The demultiplexer has an output connected to a device 1203 for reducing phase drift according to the present invention. The apparatus 1203 for reducing phase drift is connected to a channel estimator 1205 that estimates the channel transfer function in the frequency domain. The channel estimator is connected to a means 1207 for introducing a compensated phase shift, so that the influence of all channels can be taken into account. Phase post-compensation means 1207 is connected to means 1209 for extracting information contained in the converted signal provided by FFT 207. In other words, the means 1209 for extracting information determines the amount of information contained in the signal provided by the demultiplexer 1201.

  According to a further aspect of the present invention, the device for reducing phase drift (phase drift compensator) 1203 can provide a signal containing information relating to phase drift to the phase post-compensation means 1207, by means of the device 1203. Depending on the provided signal, post-phase compensation can be performed.

  Furthermore, the present invention provides concepts related to filtering and interpolation for OFDM. In general, the frequency response of a received signal after OFDM demodulation has a one-sided spectrum. By cyclically shifting the signal received before FFT, the one-sided spectrum can be converted into a symmetric two-sided spectrum. In general, the performance of standard interpolation algorithms such as linear or spline interpolation can be improved if the cyclic shift is properly selected. Furthermore, the performance of the spatial frequency code and differential modulation can also be improved. The symmetry of the two-sided spectrum is such that if the real and imaginary parts are even and odd functions respectively, the coefficients of either or both of the FIR interpolation and the smoothing filter are real values, which are calculated Cut costs in half.

  Furthermore, depending on the implementation requirements of the method according to the invention, the method according to the invention can be implemented in hardware or software. This can be done using a digital storage medium, in particular a disc or CD storing electronically readable control signals, which is a computer system that can be programmed so that the method according to the invention can be carried out. Can work in concert. Accordingly, the present invention is generally a computer program product having a program code stored on a machine-readable carrier, which program code performs at least one of the methods according to the invention on a computer. Execute. Therefore, in other words, the method according to the present invention is a computer program having program code for executing the method according to the present invention on a computer.

1 is a block diagram of an apparatus for reducing phase drift according to an embodiment of the present invention. It is a figure which shows the post process which concerns on this invention. It is a figure which shows the pre-process which concerns on this invention. It is a figure which shows the phase of a channel transfer function. It is a figure which shows the magnitude of a channel transfer function. FIG. 4 shows the corresponding spectrum of the channel transfer function of FIGS. 3a and 3b. FIG. 6 is a block diagram of an apparatus for reducing phase drift according to a further embodiment of the present invention. FIG. 6 is a block diagram of an apparatus for reducing phase drift according to a further embodiment of the present invention. It is a figure which shows an OFDM system parameter. It is a figure which shows the power delay profile (power delay profile) of a channel. It is a figure which shows the performance of the approach which concerns on this invention. It is a figure which shows the effective channel impulse response produced when cyclically shifting the received signal in a time domain. 2 shows the performance of the approach according to the invention. FIG. 6 is a block diagram of an apparatus for reducing phase shift according to the present invention based on further embodiments of the present invention. 2 is a block diagram of an OFDM receiver. FIG. It is a figure which shows the phase and magnitude of a channel transfer function. FIG. 6 shows a corresponding time domain channel impulse response. It is a figure which shows the approach of channel estimation.

Claims (29)

An apparatus for reducing phase drift in a spectrum of a time domain signal, comprising:
A converter (101) for time-frequency converting the time-domain signal into a signal converted to a frequency domain, wherein the converted signal represents a spectrum of the time-domain signal; 101)
A detector (109) for detecting the phase drift and generating a control signal indicative of the phase drift;
The converter (101) is further adapted to pre-process the time domain signal prior to time-frequency conversion or correct phase drift for the converted signal to reduce phase drift, depending on the control signal. To post-process the transformed signal after time-frequency transforming the time domain signal to introduce
The corrected phase drift is at least partially compensated for the phase drift;
apparatus.   The detector (109) detects phase drift in the converted signal provided by the converter (101) or detects phase drift from the time domain signal after preprocessing. The apparatus of claim 1. The converter (101) comprises a cyclic shift element (301) for preprocessing the time domain signal;
The cyclic shift element (301) receives the control signal, and cyclically shifts the time domain signal in response to the control signal to introduce the correction phase drift into the converted signal. is there,
The apparatus according to claim 1 or 2.   The detector (107) according to any one of claims 1 to 3, wherein the detector (107) is connected to the output of the converter (101) to detect the phase drift in the converted signal. Equipment.   The converter (101) comprises a Fourier transformer connected to the output of the cyclic shift element for converting the time domain signal provided by the cyclic shift element into the transformed signal. The device according to claim 3 or 4, wherein the device is.   The apparatus of claim 5, wherein the detector is connected to an output of the converter (101) to detect a phase drift in the converted signal. The detector (109) has a control output;
The control output is connected to the control input of the cyclic shift element;
Apparatus according to any one of claims 3-6. The cyclic shift element cyclically shifts the time domain signal by a plurality of values;
The detector (109) has the following formula:

E [Δφ _i ] represents an expected value of the phase drift Δφ _i detected for the i-th carrier among the N _FFT subcarriers. Is,
Apparatus according to any one of claims 3-7.   The cyclic shift element performs a left shift to introduce a corrected phase drift into the transformed signal to compensate for negative phase drift, or a negative correction to compensate for positive phase drift 9. Apparatus according to any one of claims 3 to 8, which performs a right shift to introduce a phase drift into the transformed signal. The cyclic shift element comprises a shift register;
The shift register has an input and an output connected to the input.
Apparatus according to any one of claims 3-9. The converter (101) comprises a phase compensator for post-processing the converted signal in the frequency domain;
The phase compensator changes the phase of the transformed signal in the frequency domain to introduce the corrected phase drift;
The device according to claim 1. The converter (101) comprises a Fourier transformer for time-frequency transforming the time domain signal,
The phase compensator is connected to the output of the Fourier transformer;
The detector (109) is connected to the output of the phase compensator;
The phase compensator receives the control signal from the detector (109).
The apparatus of claim 11.   13. Apparatus according to claim 12, wherein the phase compensator comprises a control input for receiving the control signal from the detector (109).   The phase compensator adds the corrected phase drift to the phase of the transformed signal to compensate for negative phase drift, or the transformed phase drift is transformed to compensate for positive phase drift. 14. A device according to any one of claims 11 to 13, wherein the device is subtracted from the phase of the signal. The detector (109) determines the average of the phase change between two subsequent values of the transformed signal;
The average of the phase change represents the phase drift,
The device according to claim 1. The detector determines an average value of a product of two subsequent values of the converted signal and determines a phase of the average value;
The phase of the average value represents the phase drift,
The device according to claim 1.   The detector (109) determines an average of a phase change between two values spaced by a plurality of values of the converted signal, and detects the phase drift for each value of the converted signal. 17. Apparatus according to any one of the preceding claims, wherein the average of the phase change is divided by the plurality of values to do so. The converter is connected to a cyclic shift element that preprocesses the time domain signal and to the cyclic shift element for time frequency conversion of the time domain signal cyclically shifted by the cyclic shift element. And a phase compensator connected to the output of the Fourier transformer for post-processing the transformed signal provided by the Fourier transformer,
The output of the phase compensator is connected to the detector (109);
The cyclic shift element receives a control signal from the detector (109);
The control signal introduces a correction phase drift into the transformed signal to roughly compensate for the phase drift, and the time domain signal is shifted to cyclically shift the time domain signal by the plurality of values. Represents multiple values,
The phase compensator receives a further control signal representing a further corrected phase drift introduced into the phase of the converted signal to accurately compensate for the phase drift, and the further corrected phase drift is converted into the converted signal. Is introduced into the phase of
The device according to claim 1. 17. Apparatus according to any one of the preceding claims for reducing phase drift in a multi-carrier signal receivable in the time domain, providing a transformed signal representing the multi-carrier signal receivable in the frequency domain. The converted signal has reduced phase drift; and
A multi-carrier receiver comprising: means for extracting information contained in the converted signal. The multi-carrier receiver further includes a synchronizer that performs frame synchronization;
The phase drift results from frame synchronization errors due to erroneous sample time instants;
The detector (109) generates a frame synchronization control signal representing the phase drift,
The synchronizer adjusts a sample time instant according to the frame synchronization control signal in order to reduce the frame synchronization error.
The multicarrier receiver according to claim 19.   The means for extracting the information comprises a differential demodulator for differentially demodulating the converted signal to extract information based on a phase change between two subsequent values of the converted signal. The multicarrier receiver according to claim 19 or 20, comprising: The means for extracting the information comprises a low pass filter for filtering the transformed signal in the frequency domain,
The low pass filter has a real valued coefficient representing the real part of the filter response;
The filter coefficient representing the imaginary part of the filter response is set to zero;
The converter further pre-processes the receivable multi-carrier signal in the time domain or introduces a phase shift into the converted signal to convert the spectrum of the converted signal to the passband of the low-pass filter. To post-process the transformed signal in the frequency domain to shift relative to
The multicarrier receiver according to claim 19 or 20. The receivable multi-carrier signal in the time domain is a receivable version of a transmitted signal transmitted over a communication channel;
The low pass filter is configured for channel estimation;
The multicarrier receiver according to claim 22.   The multi-carrier receiver according to claim 23, wherein the low-pass filter is a Wiener filter for channel estimation.   The multi-carrier receiver according to claim 23 or 24, wherein the low-pass filter is an interpolation filter.   The multicarrier receiver according to any one of claims 19 to 25, wherein the means for extracting the information comprises a spatial frequency block decoder. A method for reducing phase drift in a spectrum of a time domain signal, comprising:
Transforming the time domain signal into a frequency domain transformed signal, wherein the transformed signal represents a spectrum;
Detecting the phase drift;
Generating a control signal representative of the phase drift;
Depending on the control signal, the time domain signal is preprocessed before time frequency conversion, or the time domain signal is time frequency converted to introduce a corrected phase drift into the converted signal. Post-processing the processed signal, wherein the corrected phase drift is at least partially compensated for the phase drift to reduce phase shift. 28. The method of claim 27, wherein the transformed signal represents a multi-carrier signal that is receivable in the frequency domain and reduces a phase drift of the multi-carrier signal that is receivable in the time domain. Having reduced phase drift, and
Extracting the information contained in the converted signal. A method of receiving a multicarrier signal.   A computer program having program code for performing the method of claim 27 or 28 on a computer.

Images