JP5340199B2 - Multi-carrier modulation apparatus and demodulation apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce an error of an equalization coefficient by calculating high accurate frequency characteristics even though the frequency characteristics are changed by a filter etc. <P>SOLUTION: A framing unit 10 of a multicarrier modulation device 1 configures a frame by a preamble and a data symbol by disposing preamble data at its outer side in addition to a sub-channel through which real data is transmitted. A frame is subjected to orthogonal modulation after multiplexing via an FB, and then transmitted to a multicarrier demodulation device 2 as a signal of multicarrier. By interpolation, extrapolation, and weighted averaging, an equalizer 90-1 of the multicarrier demodulation device 2 calculates the frequency characteristics of an intermediate spot in an area where the frequency characteristics are varied by an HPF etc. using preamble data disposed at an outer side of the sub-channel through which the real data is transmitted. Thus, high accurate frequency characteristics can be acquired, and an error of the equalization coefficient can be reduced. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to a modulation device and a demodulation device using a multicarrier transmission system, and more particularly to a technique for obtaining an equalization coefficient using a preamble.

  Conventionally, in the field of digital broadcasting, multi-carrier transmission schemes have attracted attention as demand for large-capacity transmission increases. Multi-carrier transmission method divides high-speed signal into multiple low-speed signals, uses orthogonality to allow overlap of signals on the frequency axis, and transmits data on multiple subchannels It is. Examples of this multicarrier transmission method include OFDM (Orthogonal Frequency Division Multiplexing) and FBMC (Filter Bank based Multicarrier).

  In OFDM (orthogonal frequency division multiplexing), IFFT (Inverse Fast Fourier Transform) is used to multiplex subchannel signals, and FFT (Fast Fourier Transform) is used to separate signals. ) Is used. In addition, CP (Cyclic Prefix: GI (Guard Interval)) is used to equalize subchannel signals, and equalization is performed in the frequency region of 1 tap. Very robust against multipath interference. That is, since intersymbol interference does not occur if the delay wave is shorter than the CP period, the signal of each subchannel can be regarded as being in a flat fading state, and the level and phase are multiplied by complex coefficients. By performing correction, equalization (one-tap frequency domain equalization) can be performed.

  On the other hand, FBMC (multi-carrier scheme using a filter bank) has a merit that cannot be obtained by OFDM. The first advantage of FBMC is that a filter bank (FB) is used to multiplex and demultiplex subchannel signals, and a filter in which the attenuation in the stopband of each subchannel is arbitrarily set can be designed. It is in the point that can In OFDM, the attenuation in the stopband is 13 dB and cannot be arbitrarily designed. For this reason, in FBMC, a guard band required between other signals can be generated in a narrow range, and frequency use efficiency can be improved. When CMFB (Cosine-Modulated Filter Bank) is used as a filter bank, it is only necessary to design a prototype filter, and it is not necessary to design each subchannel filter individually (see Non-Patent Document 1).

  Further, in FBMC, for equalization, for example, the output signal of the analysis filter bank oversampled on the demodulation side (processing is performed with twice the number of samples with respect to the input signal to the synthesis filter bank provided on the modulation side) For the output signal of the analysis filter bank, an equalizer that performs equalization for each subchannel is used (see Non-Patent Document 2). The second merit of FBMC is that frequency use efficiency can be improved because CP is not required for equalization.

  As an example of an equalizer that performs equalization for each subchannel with respect to a signal having twice the number of samples as compared with the modulation side, there is an equalizer of FIR (Finite Impulse Response) type. A technique for obtaining an equalization coefficient from the viewpoint of mean square error using this FIR equalizer is disclosed (see Non-Patent Document 3). However, this method has a problem that the mounting cost increases as the number of taps increases.

  Further, a technique is disclosed in which frequency characteristics of a transmission path are obtained for a plurality of points of each subchannel, and an equalization coefficient of each subchannel is obtained from the frequency characteristics of the plurality of points (see Patent Document 1 and Non-Patent Document 4). reference). In this method, when a complex 3-tap equalization coefficient is obtained for each subchannel, a total of 3 points including 1 point with respect to the center frequency of each subchannel and 2 points with respect to a frequency at an intermediate point that is a boundary between adjacent subchannels. The equalization coefficient is obtained using the frequency characteristics of the points. According to this method, it is possible to realize equalization with less deterioration while reducing the complexity of the process for obtaining the equalization coefficient.

  However, in this method, since three-point frequency characteristics are required to obtain the equalization coefficient of each subchannel, the equalizer performs processing with a resolution higher than the frequency resolution of the filter bank. . For this reason, a means for obtaining frequency characteristics is required in addition to the means for performing equalization, and there is a problem that the configuration becomes complicated as a whole. For example, although a method for obtaining a transfer function of a transmission line in the time domain and obtaining a frequency characteristic is disclosed (see Non-Patent Document 5), there is a problem that the configuration becomes complicated as a whole and the effective cost increases. It was.

  On the other hand, an improved technique based on the above-mentioned Non-Patent Document 4 is also disclosed (see Non-Patent Document 6). In this method, in order to simplify the process of obtaining the frequency characteristics, an equalizer coefficient for performing a 1-tap complex multiplication for each subchannel is used as an MSE (Mean Squared Error: average) using a known preamble. It is obtained by a square error) standard, and this is used as the frequency characteristic of the equalizer of the center frequency in the subchannel. Then, the frequency characteristics of the equalizer at the intermediate point corresponding to the boundary of the adjacent subchannel are obtained by interpolation using the frequency characteristics of the equalizer at the center frequency of the adjacent subchannel.

(Example of conventional equalization method)
Hereinafter, an example of an equalization method in the prior art will be specifically described using an equivalent low-frequency system model. Similar to Non-Patent Document 4 described above, a multicarrier modulation apparatus including CMFB and SMFB (Sine-Modulated Filter Bank) for synthesis, and a multicarrier demodulation apparatus including CMFB and SMFB for analysis In the system configuration according to FIG. 1, as shown in FIG. 1, multicarrier transmission is performed by FBMC. Details of the configuration shown in FIG. 1 will be described later. The equalizer for each subchannel uses a complex 3-tap equalization coefficient. Further, the size M of the FB is set to 128, and the prototype filter is designed using the algorithm described in Non-Patent Document 1 with the tap length L _h = 512 and the stopband edge frequency f _S = 1.5 / M. I decided to.

(Conventional multi-carrier modulation device)
In order to secure a guard band, this equalization method uses PAM signals x _k [m], x _2M-1-k [m] (m represents a symbol number, and k represents the position of the corresponding subchannel close to DC. (0 ≦ k ≦ (M−1) = 127). 77 <k <M = 128 is a guard band, and x _k [m] = 0, x _{2M− 1-k} [m] = 0. The subchannel number is 0 to 255, 0 to 127 is represented by k, and 128 to 255 is represented by 2M-1-k.

  Such a PAM signal is processed by CMFB and SMFB and orthogonally modulated, so that it is transmitted as a multicarrier signal from the multicarrier modulation apparatus to the multicarrier demodulation apparatus via the transmission path.

  Here, the impulse response of the transmission line is a two-wave model of [1 0 0 0 0 0 0 0.3 * exp (jφ)] (that is, a delay 7 samples, a delay wave of 0.3 times level), From the method of estimating the equalization coefficient of the subchannel based on the MMSE (Minimum Mean Square Errors) standard (hereinafter referred to as mmse-ASSET) and the frequency characteristics of a plurality of points (here, 3 points) in each subchannel, Comparison is made with a method for obtaining an equalization coefficient of a subchannel (hereinafter referred to as pointwise-ASCET). Specifically, the preamble length is changed to 10, 20, and 30 symbols, and the C / N vs. BER characteristics when the respective equalization methods are used are obtained by simulation. It is assumed that a frame having a cycle of 1000 symbols including a preamble is transmitted as a multicarrier signal, and the phase φ of the reflected wave is changed by 0 to 20 degrees to select one having a poor BER characteristic. To do.

  FIG. 15 is a graph showing a simulation result (1) when the conventional technique is used. This simulation result (1) shows a case where the input data is a 16-PAM signal and is a two-wave model. In the figure, “ptws-30” represents the C / N vs. BER characteristic when the preamble length is 30 symbols and the equalization coefficient is obtained by pointwise-ASCET, and “mmse-30” represents the preamble length. The C / N vs. BER characteristic when 30 symbols are used and the equalization coefficient is obtained by mmse-ASCET is shown. “Ptws-20” and “ptws-10” represent C / N vs. BER characteristics when the preamble length is 20 and 10 symbols, respectively, and the equalization coefficient is obtained by pointwise-ASCET, and “mmse-20”. “Mmse-10” represents the C / N vs. BER characteristics when the preamble length is 20 and 10 symbols, respectively, and the equalization coefficient is obtained by mmse-ASCET.

  According to the simulation result (1) of FIG. 15, since the characteristic of “mmse-30” is close to “ptws-20”, the characteristic when the 30-symbol preamble is used by mmse-ASCET is 20 by pointwise-ASCET. It can be seen that it is obtained using the symbol preamble. Further, since the characteristics of “mmse-20” are close to “ptws-10”, the characteristics when 20-symbol preamble is used by mmse-ASCET can be obtained by using 10-symbol preamble by pointwise-ASCET. Recognize. Further, it can be seen from the characteristics of “ptws-10” and “mmse-10” that even when a 10-symbol preamble is used, the pointwise-ASCET is less deteriorated than the mmse-ASCET. Therefore, the pointwise-ASCET equalization method (method for obtaining the equalization coefficient from the frequency characteristics of a plurality of points in each subchannel) can obtain better characteristics with less preamble than mmse-ASCET. You can say that.

(Conventional multi-carrier demodulator)
On the other hand, as a multicarrier demodulator, a device using a direct conversion type receiving circuit has been put into practical use in many radio systems. The direct conversion method does not require an intermediate frequency band BPF (SAW filter or the like) necessary for the heterodyne method, and can be configured to be suitable for all semiconductor integration. However, since the reception frequency and the local oscillation frequency are the same, the leaked local oscillation wave becomes a DC offset by self-detection. For this reason, the DC offset is a big problem. Therefore, in a wideband signal system, a simple method for removing a DC offset using an HPF (High Pass Filter) is used. This technique has been put to practical use in W-CDMA receivers and the like. For example, an HPF having a cutoff frequency of about 1% of the baseband IQ signal band (a filter having the characteristics of a first-order RC filter because of AC coupling) is used (see Non-Patent Document 7).

(K = 0 (0th and 255th) subchannels)
The system given as an example of the above-described equalization technique in the prior art does not use k = 0 (0th and 255th) subchannels, and is configured by applying the direct conversion system receiving circuit described above. Assume a case.

FIG. 14 is a diagram showing a data arrangement in a frame when the k = 0 (0th and 255th) subchannels are not used in the prior art. As shown in FIG. 14, a frame configured in the multicarrier modulation apparatus is composed of a preamble and a data symbol. The vertical axis indicates the direction of time (symbol), and the horizontal axis indicates the direction of frequency (subchannel). The horizontal axis is k = 0 to 127, the data of the 0th to 127th subchannels are arranged from left to right, and the 128th to 255th subchannels are arranged from right to left. Channel data is arranged. In the preamble and data symbol, the data of the subchannel closest to DC (k = 0 (0th and 255th) subchannel) is 0 (x ₀ [m] = 0, x ₂₅₅ [m] = 0). . The guard bands are arranged in the subchannels of k = 78 to 127 and 128 to 177.

  FIG. 16 is a graph showing a simulation result (2) in the prior art that does not use the k = 0 (0th and 255th) subchannels. This simulation result (2) shows a case where the input data is a 16-PAM signal and is a two-wave model. (A) shows the case where the cutoff frequency of the HPF provided in the multicarrier demodulator is 0.5 times the subchannel interval, and (b) shows the case where the cutoff frequency of the HPF is 1 times the subchannel interval. Yes. In the figure, “ptws-30”, “mmse-30”, “ptws-20”, “mmse-20”, “ptws-10”, and “mmse-10” indicate the same equalization technique and preamble length as in FIG. .

  According to the simulation result (2) in FIG. 16, the pointwise-ASCET equalization method is better when the HPF cutoff frequency is 0.5 times the subchannel interval (a) or 1 time (b). It can be seen that the BER characteristics are greatly deteriorated as compared with the equalization method by mmse-ASCET. This is because, in the pointwise-ASCET equalization method, the HPF that removes the DC offset affects the interpolation in the region closer to the DC (the intermediate point between the k = 0 subchannel and the k = 1 subchannel). The frequency characteristic is not estimated correctly by extrapolation or by substituting the frequency characteristic of the center frequency in the subchannel of k = 1), so that the error of the equalization coefficient becomes large and the DC-oriented subchannel This is because the BER characteristics of the above deteriorate.

FIG. 17 is a graph showing the simulation result (3) when C / N = 45 dB in “ptws-30” in FIG. This simulation result (3) shows that when the input data is a 16-PAM signal, the two-wave model, and the cutoff frequency of the HPF is 0.5 times the subchannel spacing, the C / N = 45 dB of “ptws-30” The error rate for each subchannel is shown. The horizontal axis is k corresponding to the subchannel number, and the vertical axis is the average symbol error rate of x _k [m], x _2M-1-k [m]. According to the simulation result (3) in FIG. 17, it can be seen that the symbol error rate is large in the subchannel region near DC with a small k (the subchannel region near the 0th and 255th).

  FIG. 18 is a graph showing a simulation result (4) in the prior art in which k = 0, 1 (0th, 1st, 254th, and 255th) subchannels are not used. This simulation result (4) shows a case where the input data is a 16-PAM signal, a two-wave model, and the cutoff frequency of the HPF is twice the subchannel interval. In the figure, “ptws-30”, “mmse-30”, “ptws-20”, “mmse-20”, “ptws-10”, and “mmse-10” have the same equalization method and preamble length as those in FIG. 15 and FIG. Show.

According to the simulation result (4) in FIG. 18, when the cutoff frequency of the HPF is twice the subchannel interval, k = 1 in addition to k = 0 (the first and 254 in addition to the 0th and 255th). Even if the sub-channel of (th) is not used, it can be seen that the BER characteristics are degraded when compared with the characteristics shown in FIG. 15 at BER = 10 ⁻⁴ .

International Publication No. 2005/091583

L Lin, et al., “Cosine-Modulated Multitone for Very-High-Speed Digital Subscriber Lines”, EURASIP Journal on Advances in Signal Processing, Volume 2006, Article ID 19329 J Alhava, et al., “Adaptive Sine-Modulated / Cosine-Modulated Filter Bank Equalizer for Transmultiplexers”, ECCTD’01-European Conference on Circuit Theory and Design, August 28-31, 2001, Espoo, Finland Hirosaki, “An analysis of automatic equalizers for orthogonally multiplexed QAM systems”, IEEE Trans. Commun., Vol. 28, pp. 73-83, January 1980 T Ihalainen, et al., “Channel Equalization in Filter Bank Based Multicarrier Modulation for Wireless Communications”, EURASIP Journal on Advances in Signal Processing Volume 2007, Article ID 49389 Y Yuan, et al., “Frequency domain equalization in single carrier transmission: filter bank approach”, EURASIP Journal on Applied Signal Processing, Vol. 2007 (2007) K Izumi, et al., “Performance Evaluation of Wavelet OFDM Using ASCET”, 11th IEEE International Symposium on Power Line Communications and its Applications (ISPLC 2007), pp. 245-250, MAR 26-28, 2007 Pisa ITALY J Ryynanen, et al., “A single-chip multimode receiver for GSM900, DCS1800, PCS1900, and WCDMA”, IEEE Journal of Solid-State Circuits Vol. 38, Issue. 4, pp.594-602, (Apr. 2003 )

  Thus, the prior art equalization method obtains the frequency characteristic of the center frequency in each subchannel from the preamble data, and obtains the frequency characteristic by interpolation such as linear interpolation at the intermediate point between the subchannels. The intermediate points that cannot be obtained by the interpolation are obtained by extrapolation or substitution, and the subchannel equalization coefficients are obtained from these frequency characteristics.

  However, since HPF or the like is used to remove the DC offset in the multicarrier demodulator, the frequency characteristics greatly change in a region close to DC due to the influence of HPF or the like. For this reason, there is a problem that the error of the equalization coefficient becomes large in the region near DC, and the BER characteristic is deteriorated.

  Therefore, the present invention has been made to solve such a problem, and its purpose is to obtain a frequency characteristic using preamble data, and when the equalization coefficient is obtained from the frequency characteristic, the frequency characteristic is changed by a filter or the like. Therefore, an object of the present invention is to provide a multicarrier modulation device and a multicarrier demodulation device capable of calculating frequency characteristics with high accuracy and reducing errors in equalization coefficients.

  In order to achieve the above object, the multi-carrier modulation apparatus according to the present invention includes a preamble in which preamble data used for calculating subchannel equalization coefficients is arranged, and data in which bit string data is arranged as actual data. A multi-carrier modulation apparatus that configures a frame using symbols and transmits the multi-carrier signal as a multi-carrier signal via a transmission line; receives the multi-carrier signal; removes a signal of a predetermined frequency by filtering; and the preamble data The frequency characteristic of the transmission path in the sub-channel is calculated based on the above, the equalization coefficient is calculated based on the frequency characteristic, the received signal is equalized based on the equalization coefficient, and is restored to the original bit string In the multicarrier modulation device of the system comprising a multicarrier demodulation device The preamble data is arranged in a subchannel where the actual data is arranged, and the preamble is assigned to a subchannel adjacent to the subchannel where the actual data is arranged among subchannels where the actual data is not arranged. A preamble generation unit that arranges data and generates a preamble; a frame configuration unit that configures a frame using the preamble generated by the preamble generation unit and the data symbol; and a sub of a frame configured by the frame configuration unit A filter bank that synthesizes signals for each channel, and an orthogonal modulation unit that orthogonally modulates a signal multiplexed by the filter bank.

  Furthermore, the multicarrier demodulator according to the present invention is a multicarrier demodulator that receives a multicarrier signal from the multicarrier modulator and restores the original bit sequence, and an orthogonal demodulator that orthogonally demodulates the multicarrier signal; A filter that removes a signal of a predetermined frequency from the signal demodulated by the orthogonal demodulator, a filter bank that separates the signal from the filter into a signal for each subchannel, and a signal separated by the filter bank, etc. An equalization unit, and the equalization unit includes the real data among the preamble data of the subchannel in which the real data is arranged and the subchannel in which the real data is not arranged. Based on the preamble data of the subchannel adjacent to the subchannel. Calculating the channel frequency characteristic of the center frequency in the subchannel in which the data is arranged, and based on the channel frequency characteristic of the center frequency in the two adjacent subchannels among the subchannels in which the preamble data is arranged, A frequency characteristic calculation unit that calculates a transmission line frequency characteristic at an intermediate point in the two adjacent subchannels, and a subchannel in which the actual data is arranged based on the transmission line frequency characteristic calculated by the frequency characteristic calculation unit An equalization coefficient calculation unit that calculates an equalization coefficient of the equalization coefficient, and an equalization processing unit that equalizes the signal separated by the filter bank based on the equalization coefficient calculated by the equalization coefficient calculation unit. It is characterized by having.

  In the multicarrier demodulator according to the present invention, the frequency characteristic calculation unit provided in the equalization unit calculates the transmission line frequency characteristic of the center frequency in the subchannel in which the preamble data is arranged based on the preamble data. The first transmission of the intermediate point in two adjacent subchannels in the predetermined area for a predetermined area that is preset to be affected by the filter among the subchannel areas in which the preamble data is arranged Determining the path frequency characteristics by interpolation, determining the second transmission path frequency characteristics at the intermediate point by extrapolation, averaging the first transmission path frequency characteristics and the second transmission path frequency characteristics by weighting, and The transmission path frequency characteristic at the intermediate point is calculated and the subchannel area where the preamble data is arranged is calculated. , The region not including the predetermined region, is calculated by interpolation of the transmission channel frequency characteristic of the intermediate points in two adjacent sub-channel in the area not including the predetermined region, characterized in that.

  In the multicarrier demodulator according to the present invention, the interpolation weight and the extrapolation weight are set in advance so that a sum of the interpolation weight and the extrapolation weight becomes a power of two. It is characterized by that.

  In the multicarrier demodulator according to the present invention, the filter is a primary filter composed of a resistor and a capacitor, and the frequency characteristic calculation unit provided in the equalization unit has a subchannel region in which the preamble data is arranged. Among the predetermined regions, the adjacent regions in the predetermined region are set based on preset interpolation weights and extrapolation weights so as to approximate frequency characteristics corresponding to the frequency removal characteristics of the primary filter. The transmission path frequency characteristic of the intermediate point in the two subchannels is calculated.

  The multicarrier demodulator according to the present invention is characterized in that an equalizer frequency characteristic is used instead of the transmission line frequency characteristic.

  As described above, according to the present invention, preamble data is also arranged in a subchannel adjacent to a subchannel in which actual data is arranged. As a result, the multicarrier demodulator can calculate the frequency characteristics of the center frequency for subchannels adjacent to the subchannel where the actual data is arranged. The frequency characteristic of the point can be calculated by interpolation. Therefore, it is possible to obtain a highly accurate value by calculating the frequency characteristics of the intermediate points in all the subchannels in which the actual data is arranged using the values obtained by interpolation. That is, the error of the equalization coefficient can be reduced, and the deterioration of the BER characteristic can be reduced. Further, for example, while using the HPF for removing the DC offset as it is, the deterioration of the BER characteristic can be improved only by adding a simple process with a small amount of calculation according to the present invention.

It is the block diagram which showed the structure of the whole system containing the multicarrier modulation apparatus and multicarrier demodulation apparatus by embodiment of this invention with the equivalent low-pass system model. It is a block diagram which shows the structure of the framing part with which the multicarrier modulation apparatus was equipped. It is a block diagram which shows the structure of the equalizer with which the multicarrier demodulation apparatus was equipped. It is a figure which shows the data arrangement | positioning in a flame | frame. It is a flowchart which shows the process of a flame | frame part. It is a flowchart which shows the process of the frequency characteristic calculation part with which the equalizer was equipped. It is a figure explaining the frequency characteristic calculated by the frequency characteristic calculation part. It is a figure explaining the method of calculating the frequency characteristic of an intermediate point in the case of approximating a frequency characteristic with the formula of y = (a / x) + b. It is a graph which shows the simulation result (1) at the time of using the multicarrier modulation apparatus and multicarrier demodulation apparatus by embodiment of this invention. It is a graph which shows the simulation result (2) at the time of using the multicarrier modulation apparatus and multicarrier demodulation apparatus by embodiment of this invention. It is a graph which shows the simulation result (3) at the time of using the multicarrier modulation apparatus and multicarrier demodulation apparatus by embodiment of this invention. It is a graph which shows the simulation result (4) at the time of using the multicarrier modulation apparatus and multicarrier demodulation apparatus by embodiment of this invention. It is a graph which shows the simulation result (5) at the time of using the multicarrier modulation apparatus and multicarrier demodulation apparatus by embodiment of this invention. It is a figure which shows the data arrangement | positioning in the flame | frame in a prior art. It is a graph which shows the simulation result (1) at the time of using a prior art. It is a graph which shows the simulation result (2) at the time of using a prior art. It is a graph which shows the simulation result (3) at the time of using a prior art. It is a graph which shows the simulation result (4) at the time of using a prior art.

  The present invention relates to a multicarrier modulation apparatus that forms a frame with a preamble and data symbols, multiplexes data for each subchannel constituting the frame with FB, orthogonally modulates and transmits as a multicarrier signal, and a multicarrier modulation apparatus Multi-carrier signals are received and orthogonally demodulated, separated into data for each subchannel constituting the frame by FB, the frequency characteristics of the transmission path are obtained using the preamble data, and the equalization coefficient is obtained from the frequency characteristics A system including a multi-carrier demodulator is assumed. The multicarrier modulation apparatus arranges preamble data in subchannels outside the subchannel in which actual data is transmitted by data symbols. The multi-carrier demodulator calculates the frequency characteristics of the center frequency in the subchannel in which the actual data is transmitted using the preamble data of the subchannel, and in the subchannel outside the subchannel in which the actual data is transmitted. The frequency characteristic of the center frequency is also calculated using the preamble data of the subchannel. The frequency characteristic at the intermediate point between the subchannels in which the actual data is transmitted is calculated by interpolation such as linear interpolation using the frequency characteristic of the center frequency in the two adjacent subchannels. For the frequency characteristics at the midpoint of the subchannel in the region where the frequency characteristics change due to HPF or the like, the value calculated by interpolation using the frequency characteristics of the center frequency in two adjacent subchannels and the extrapolation A predetermined weighted average is calculated using the calculated value.

  Here, the preamble data is given to the subchannel where the actual data is transmitted in addition to the subchannel where the actual data is transmitted by the data symbol, so all the subchannels where the actual data is transmitted and the subchannels outside the subchannel. The frequency characteristics at the intermediate point can be calculated by interpolation. As a result, the frequency characteristics of the intermediate point in the subchannel in the region where the frequency characteristics change due to the influence of HPF or the like are the values obtained by the interpolation using the frequency characteristics of the center frequency in the two adjacent subchannels and the values obtained by the extrapolation. Since it can be calculated based on the above, it becomes a highly accurate value. Therefore, the error of the equalization coefficient can be reduced and the deterioration of the BER characteristic can be reduced. That is, it is possible to improve the deterioration of the BER characteristic only by adding a simple process with a small amount of calculation according to the present invention while using the HPF or the like for removing the DC offset as it is.

The best mode for carrying out the present invention will be described below with reference to the drawings.
〔system〕
First, an overall system including a multicarrier modulation apparatus and a multicarrier demodulation apparatus according to an embodiment of the present invention will be described. FIG. 1 is a block diagram showing the configuration of the entire system using an equivalent low-frequency system model. This system includes a multicarrier modulation apparatus 1 and a multicarrier demodulation apparatus 2 that implement passband transmission. The multicarrier modulation apparatus 1 framing a bit string to form a frame composed of a preamble and a data symbol, multiplexing signals for each subchannel by the FB and performing orthogonal modulation, and transmitting the multicarrier signal via the transmission path 3 Transmit to the multi-carrier demodulator 2. The multicarrier demodulator 2 receives the multicarrier signal transmitted from the multicarrier modulator 1 via the transmission path 3, performs orthogonal modulation, separates the signal into subchannel signals by FB, and performs equalization. To deframe and restore the original bit string.

  Since FIG. 1 shows an equivalent low-frequency model, the multicarrier modulation apparatus 1 converts the signal into a passband signal, and the multicarrier demodulation apparatus 2 converts the passband signal into a baseband IQ signal. Processing, frequency synchronization processing, clock synchronization processing, and the like are omitted. In the multicarrier demodulator 2, synchronization is assumed to be established. In the multicarrier modulation apparatus 1 and the multicarrier demodulation apparatus 2, the size of CMFB and SMFB is M, and M = 128. k indicates a number (corresponding to the subchannel number) obtained by counting the position of the corresponding subchannel from the side closer to DC, and m indicates a symbol number.

[Multi-carrier modulator]
Next, the configuration of multicarrier modulation apparatus 1 shown in FIG. 1 will be described in detail. The multicarrier modulation apparatus 1 includes a framing unit 10, a subtracter 20-1, an adder 20-2, multipliers 30-1 and 30-2, a combined CMFB 40-1, a combined SMFB 40-2, and an orthogonal modulator 50. I have.

The framing unit 10 receives a bit string, converts the bit string of the serial signal into a parallel signal for each data contained in one frame of data symbols, and generates a preamble and data symbols to form a frame. The framing unit 10 combines the (K + 1) th (0 ≦ k <M) PAM signals x _k [m], x _2M-1-k [m] from both ends of the 2M PAM signals. To the subtracter 20-1 and the adder 20-2. Here, when generating the preamble, the framing unit 10 arranges the preamble data in the subchannel where the actual data is transmitted by the data symbol, and arranges the preamble data in the other subchannels. Details of the framing unit 10 will be described later. Hereinafter, the frame length is 1000 symbols.

Subtractor 20-1, the framing section 10 PAM signal x _k [m], and enter the _{x 2M-1-k [m} ], PAM signal from the PAM signal _{x k [m] x 2M-} 1-k [ m] is subtracted, and the subtraction result is output to the multiplier 30-1. The multiplier 30-1 receives the subtraction result from the subtracter 20-1, multiplies the subtraction result by 1/2, and multiplies the difference ((x _k [m] −x _2M-1-k [m] ) / 2) is output to the synthesized CMFB 40-1.

The adder 20-2 receives the PAM signal x _k [m], x _2M-1-k [m] from the framing unit 10 and receives the PAM signal x _k [m] and the PAM signal x _2M-1-k [m]. m] are added, and the addition result is output to the multiplier 30-2. The multiplier 30-2 receives the addition result from the adder 20-2, multiplies the addition result by 1/2, and the sum multiplication result ((x _k [m] + x _2M-1-k [m]) / 2) is output to the combined SMFB 40-2.

The synthesis CMFB 40-1 inputs the difference multiplication result ((x _k [m] −x _2M-1-k [m]) / 2) from the multiplier 30-1, performs stepwise filtering, and inputs the result. The signals for each subchannel are combined and multiplexed. The signal multiplexed by the combined CMFB 40-1 is output to the quadrature modulator 50. The synthesis SMFB 40-2 inputs the sum multiplication result ((x _k [m] + x _2M-1-k [m]) / 2) from the multiplier 30-2, performs stepwise filtering, and inputs the result. The signals for each subchannel are combined and multiplexed. The signal multiplexed by the combined SMFB 40-2 is output to the quadrature modulator 50.

  The quadrature modulator 50 receives the signals that have passed through the FB from the combined CMFB 40-1 and the combined SMFB 40-2, performs quadrature modulation on the input signal, and transmits the multi-carrier signal after quadrature modulation via the transmission path 3. Transmit to the multi-carrier demodulator 2. The quadrature modulator 50 includes a multiplier 51 and an adder 52 when the quadrature modulation process is expressed by a complex number of an equivalent low-frequency system. The multiplier 51 multiplies the signal from the combined SMFB 40-2 by j (= √−1), and the adder 52 adds the signal resulting from the multiplication from the multiplier 51 and the signal from the combined CMFB 40-1. To do. The signal resulting from the addition by the adder 52 is transmitted as a multicarrier signal.

(Configuration of framing unit)
Next, the framing unit 10 of the multicarrier modulation apparatus 1 shown in FIG. 1 will be described in detail. FIG. 2 is a block diagram illustrating a configuration of the framing unit 10. The framing unit 10 includes a preamble generation unit 11, a data symbol generation unit 12, and a frame configuration unit 13. As described above, the framing unit 10 receives a bit string, converts the bit string of the serial signal into a parallel signal for each frame, forms a frame including a preamble and data symbols, and generates a PAM signal x _k [m], x Output _2M-1-k [m] as a set.

  The preamble generation unit 11 of the framing unit 10 adds the preamble data to the subchannels where the actual data is transmitted by the data symbols at the position of the preset subchannel number and symbol number, and also to the subchannels outside the subchannel. In the multicarrier demodulator 2, preamble data which is valid data for calculating the frequency characteristics of the center frequency in the subchannel is arranged, and “0” which is invalid data is arranged at other positions. A guard band is generated and a preamble is generated for each frame. As the preamble data, preset data is used. Then, the preamble generation unit 11 outputs the generated preamble to the frame configuration unit 13.

  FIG. 4 is a diagram for explaining the data arrangement in the frame constituted by the framing unit 10 and corresponds to FIG. 14 showing the data arrangement in the frame in the prior art. As shown in FIG. 4, the preamble generation unit 11 arranges preamble data and generates a preamble. Specifically, the preamble generation unit 11 arranges the preamble data in the subchannels of k = 1 to 77 (1st to 77th and 178th to 254th) where the actual data is arranged in the data symbol area. Furthermore, preamble data is also arranged in k = 0 (0th and 255th) subchannels outside the subchannel in which actual data is arranged, that is, the subchannel closest to DC. Also, the preamble generation unit 11 arranges “0” in the subchannels of k = 78 to 127 (78th to 177th), and generates a guard band.

  In addition, the preamble generation unit 11 arranges the preamble data in the subchannel closest to DC, which is the k = 0 (0th and 255th) subchannel, in addition to the subchannel in which the actual data is arranged. However, this is because the HPF provided in the multicarrier demodulator 2 is assumed. In the present invention, it is not always necessary to arrange preamble data in a subchannel closer to DC, and it is only necessary to arrange preamble data in a subchannel at a necessary position according to a filter that causes a change in frequency characteristics. For example, when the multicarrier demodulator 2 includes an LPF (Low Pass Filter), the preamble generator 11 adds k = 78, 127 (78th, 127th, 128th) in addition to the subchannel in which the actual data is arranged. The preamble data may also be arranged in the (th and 177th) subchannels.

  Returning to FIG. 2, the data symbol generation unit 12 inputs a bit string, converts the serial signal of the bit string into a parallel signal, for example, gray-codes the 8-PAM signal, and sets a preset subchannel number. In addition, a Gray code string and “0” are arranged at the position of the symbol number, a guard band is generated, and a data symbol is generated for each frame. Then, the data symbol generation unit 12 outputs the generated data symbol to the frame configuration unit 13.

Referring to FIG. 4, the data symbol generation unit 12 performs Gray coding on the input bit string, and converts the Gray code string to actual data in subchannels k = 1 to 77 (1st to 77th and 178th to 254th). And “0” is arranged in the subchannel of k = 0 (0th and 255th). In addition, the data symbol generation unit 12 arranges “0” in the subchannels of k = 78 to 127 (from the 78th to the 177th), and generates a guard band. Here, as a restriction on each symbol, in the subchannel of k = 0 (0th and 255th), since there is an influence of HPF used to remove the DC offset, actual data is not transmitted. That is, for data symbols, PAM signals x ₀ [m], x ₂₅₅ [m] = 0. In order to secure a guard band, PAM signals x _k [m], x _2M-1-k [m] = 0 (77 <k <128) are set in a portion where k is large. In this calculation example, the PAM signal x _k [m], x _{2M-1- is set in} the range of 77 <k <128 in order to set the subchannel interval to 0.75 MHz and the modulation wave band to be within 120 MHz. _k [m] = 0.

Returning to FIG. 2, the frame configuration unit 13 inputs the preamble from the preamble generation unit 11 and the data symbol from the data symbol generation unit 12, and configures a frame composed of the preamble and the data symbol as shown in FIG. To do. The framing unit 10 then outputs PAM signals x _k [m] and x _2M-1-k [m] corresponding to the (k + 1) th (0 ≦ k <M) data from both ends of the 2M PAM signals. A set is output to the subtracter 20-1 and the adder 20-2.

(Processing of framing part)
Next, the processing of the framing unit 10 shown in FIGS. 1 and 2 will be described in detail. FIG. 5 is a flowchart showing processing of the framing unit 10. The data symbol generation unit 12 of the framing unit 10 receives the bit string, performs serial / parallel conversion, performs gray coding on the 8-PAM signal, and generates a gray code string (step S501). Here, the bit string of the serial signal is converted into a parallel signal every (1000-10) × 2 × (128-51) × 3 bits, which is the number of data accommodated in one frame data symbol.

  The data symbol generator 12 inserts “0” at a predetermined position of the gray code string to generate 256 units of data symbols (step S502). Specifically, the data symbol generation unit 12 arranges real data that is a Gray code string in the subchannels of k = 1 to 77 (1st to 77th and 178th to 254th), and k = 0 (0 “0” is allocated to the sub-channel closest to the DC which is the (th and 255th) subchannels (step S503), and “0” is allocated to the subchannels of k = 78 to 127 (78th to 177th). A guard band is generated (step S504). In this way, the data symbol generator 12 generates data symbols having a length of 990 symbols out of 1000 symbols constituting one frame.

  On the other hand, the preamble generation unit 11 arranges the preamble data in the k = 1 to 77 (first to 77th and 178th to 254th) subchannels in which the actual data is arranged in the data symbol area. Preamble data is also arranged in k = 0 (0th and 255th) subchannels outside the subchannel in which data is arranged, that is, in the subchannel closest to DC (step S505). Also, the preamble generator 11 places “0” in the subchannels of k = 78 to 127 (from 78th to 177th), and generates a guard band (step S506). In this way, the preamble generator 11 generates a preamble having a length of 10 symbols out of 1000 symbols constituting one frame. The preamble generated here uses the same preamble every frame.

  The frame configuration unit 13 inserts the preamble generated by the preamble generation unit 11 before the data symbol generated by the data symbol generation unit 12, and configures a frame (step S507). In this way, a frame having a symbol length of 1000 symbols is configured by the frame configuration unit 13.

  As described above, according to the multicarrier modulation device 1 according to the embodiment of the present invention, the preamble generation unit 11 of the framing unit 10 is closer to DC than the subchannel in which actual data is transmitted by data symbols. Preamble data is also arranged in the subchannels, and the frame configuration unit 13 configures a frame with data symbols in which actual data is arranged in predetermined subchannels and preambles in which the preamble data is arranged. The frame configured in this manner is multiplexed by FB for each subchannel, orthogonally modulated, and transmitted to the multicarrier demodulator 2 as a multicarrier signal. As a result, the multicarrier demodulator receives the multicarrier signal transmitted from the multicarrier modulator 1, and in addition to the preamble data of the predetermined subchannel in which the actual data is transmitted by the data symbol, is closer to the DC. By using also the preamble data arranged in the subchannels, it is possible to calculate a highly accurate frequency characteristic for a portion where the frequency characteristic changes due to the influence of HPF or the like. Details of the processing of the multicarrier demodulator 2 will be described later. Therefore, the error of the equalization coefficient can be reduced and the deterioration of the BER characteristic can be reduced.

[Multi-carrier demodulator]
Next, the configuration of the multicarrier demodulator 2 shown in FIG. 1 will be described in detail. The multicarrier demodulator 2 includes an orthogonal demodulator 60, analysis CMFBs 70-1 and 70-3, analysis SMFBs 70-2 and 70-4, subtractors 80-1 and 80-4, an adder 80-2, and an addition inverter. 80-3, equalizers 90-1 and 90-2, real part extraction units 100-1 and 100-2, and a deframing unit 110 are provided.

  The multicarrier demodulator 2 receives the multicarrier signal transmitted by the multicarrier modulator 1 via the transmission path 3. Then, the quadrature demodulator 60 receives the received multicarrier signal and converts it into a baseband IQ signal. When the orthogonal demodulation processing is expressed by an equivalent low-frequency complex number, the orthogonal demodulator 60 includes a real part extraction unit 61 (Re {·}) and an imaginary part extraction unit 62 (Im {·}). The quadrature demodulator 60 performs quadrature demodulation on the input signal, the real part extraction unit 61 determines the real part of the input signal, extracts the real part signal, and the imaginary part extraction unit 62 converts the input signal to the imaginary part. The part is determined and an imaginary part signal is extracted. That is, the signal passing through the transmission line 3 is divided into a real part signal and an imaginary part signal in the equivalent low-frequency system model.

  The DC offset is removed from the real part signal and the imaginary part signal which are the outputs of the quadrature demodulator 60 by an HPF (not shown). Here, the HPF is, for example, an IIR (Infinite Impulse Response) filter that assumes a first-order RC filter composed of a resistor and a capacitor. The real part signal is input to analysis CMFB 70-1 and analysis SMFB 70-2, and the imaginary part signal is input to analysis CMFB 70-3 and analysis SMFB 70-4.

  The analysis CMFB 70-1 receives a real part signal from the quadrature demodulator 60, performs stepwise filtering using a filter bank obtained from a prototype filter and a cosine modulation matrix, analyzes the input signal, and subchannels Separate each signal. The signals separated by the analysis CMFB 70-1 are output to the subtracters 80-1 and 80-4. The analysis SMFB 70-2 receives the real part signal from the quadrature demodulator 60, performs stepwise filtering using a filter bank obtained from the prototype filter and the sine modulation matrix, analyzes the input signal, and subchannels Separate each signal. The signal separated by the analysis SMFB 70-2 is output to the adder 80-2 and the addition inverter 80-3.

  The analysis CMFB 70-3 receives an imaginary part signal from the quadrature demodulator 60, performs stepwise filtering using a filter bank obtained from a prototype filter and a cosine modulation matrix, analyzes the input signal, and subchannels Separate each signal. The signal separated by the analysis CMFB 70-3 is output to the adder 80-2 and the addition inverter 80-3. The analysis SMFB 70-4 receives an imaginary part signal from the quadrature demodulator 60, performs stepwise filtering using a filter bank obtained from a prototype filter and a sine modulation matrix, analyzes the input signal, and subchannels Separate each signal. The signals separated by the analysis SMFB 70-4 are output to the subtracters 80-1 and 80-4.

  For each subchannel, the subtracter 80-1 subtracts the imaginary part signal subjected to the analysis SMFB processing by the analysis SMFB 70-4 from the real part signal subjected to the analysis CMFB processing by the analysis CMFB 70-1, and obtains the subtraction result I1. It outputs to the equalizer 90-1. For each subchannel, the adder 80-2 adds the real part signal analyzed by the analysis SMFB 70-2 and the imaginary part signal analyzed by the analysis CMFB 70-3, and adds the addition result Q1. It outputs to the equalizer 90-1. For each subchannel, the addition inverter 80-3 adds and inverts the real part signal analyzed by the analysis SMFB 70-2 and the imaginary part signal analyzed by the analysis CMFB 70-3. The inversion result Q2 is output to the equalizer 90-2. The subtractor 80-4 subtracts the real part signal subjected to the analysis CMFB processing by the analysis CMFB 70-1 from the imaginary part signal subjected to the analysis SMFB processing by the analysis SMFB 70-4 for each sub-channel, and obtains the subtraction result I2. It outputs to the equalizer 90-2.

  The equalizer 90-1 receives the signal I1 from the subtractor 80-1 and the signal Q1 from the adder 80-2, treats the signals I1 and Q1 as a pair of signals (I1 and Q1), and converts the preamble data. The frequency characteristic of the center frequency in the subchannel and the frequency characteristic at the intermediate point of the adjacent subchannel are calculated, and the equalization coefficient of the subchannel is calculated based on these frequency characteristics. The data signals (I1, Q1) are equalized, and the equalized signal is output to the real part extraction unit 100-1. The equalizer 90-2 receives the signal I2 from the subtractor 80-4 and the signal Q2 from the addition inverter 80-3, treats the signals I2 and Q2 as a pair of signals (I2 and Q2), and preamble data. Is used to calculate the frequency characteristic of the center frequency in the subchannel and the frequency characteristic at the intermediate point of the adjacent subchannel, calculate the equalization coefficient of the subchannel, and use the equalization coefficient to determine the actual data signal (I2, Q2) is equalized, and the equalized signal is output to the real part extraction unit 100-2. Details of the equalizers 90-1 and 90-2 will be described later.

The real part extraction unit 100-1 receives the equalization signal from the equalizer 90-1, determines the real part of the equalization signal, extracts the real part signal, and converts the extracted real part signal into the original PAM. The received signal corresponding to signal x _k [m] is output to deframer 110. The real part extraction unit 100-2 receives the equalized signal from the equalizer 90-2, determines the real part of the equalized signal, extracts the real part signal, and converts the extracted real part signal into the original PAM. The received signal corresponding to signal x _2M-1-k [m] is output to deframer 110.

Deframing section 110 receives reception signal x _k [m] from real part extraction section 100-1, and receives reception signal x _2M-1-k [m] from real section extraction section 100-2. A frame with one sub-channel signal as one symbol is deframed, divided into a preamble and a data symbol, real data is extracted from the data symbol, a gray code string of the real data is decoded, and the decoded parallel data for each frame The signal is converted into a serial signal and a bit string of the serial signal is output.

(Equalizer configuration)
Next, the equalizers 90-1 and 90-2 of the multicarrier modulation apparatus 2 shown in FIG. 1 will be described in detail. FIG. 3 is a block diagram showing a configuration of the equalizer 90-1. The equalizer 90-1 includes a frequency characteristic calculation unit 91, an equalization coefficient calculation unit 92, and an equalization processing unit 93. As described above, the equalizer 90-1 treats the input signals I1 and Q1 as a pair of signals (I1 and Q1), and uses the preamble data to determine the frequency characteristics of the center frequency in the subchannel and the adjacent subchannels. The frequency characteristic of the intermediate point is calculated, the equalization coefficient of the subchannel is calculated, and the pair of signals (I1, Q1) are equalized. The configuration of the equalizer 90-2 is the same as the configuration of the equalizer 90-1.

The frequency characteristic calculation unit 91 of the equalizer 90-1 receives the signal I1 from the subtractor 80-1 and the signal Q1 from the adder 80-2, respectively, and equalizes the equalizer that performs one-tap complex multiplication. The coefficient is obtained by the MSE (Mean Squared Error) standard using the input preamble data signal (I1, Q1) and the known preamble data, and the reciprocal of the equalization coefficient is calculated. The frequency characteristic of the frequency transmission line (transmission line frequency characteristic) η _{k, i} (i = 1) is obtained. As shown in FIG. 4, since preamble data is arranged in the subchannels of k = 0 to 77 (0th to 77th and 178th to 255th), the frequency characteristic calculation unit 91 converts the preamble data into The frequency characteristic of the center frequency in each subchannel is calculated.

Then, the frequency characteristic calculation unit 91 calculates the frequency characteristic η _{k, i} (i = 0, 2) of the intermediate point in the adjacent subchannel based on the frequency characteristic of the center frequency in the subchannel. Specifically, the frequency characteristic calculation unit 91 uses a frequency characteristic of a center frequency in two adjacent subchannels for a frequency band in which the frequency characteristic does not change so much even by HPF or the like, by interpolation such as linear interpolation. The frequency characteristic of the intermediate point in the adjacent subchannel is calculated. Further, the frequency characteristic calculation unit 91 uses a frequency characteristic of the center frequency in two adjacent subchannels for a frequency band in which the frequency characteristic changes due to HPF or the like, and a value calculated by interpolation such as linear interpolation, Based on the value calculated by the insertion, the frequency characteristic of the intermediate point in the adjacent subchannel is calculated with a predetermined weight. Details of the processing of the frequency characteristic calculation unit 91 will be described later.

  As a result, the frequency characteristic calculation unit 91 calculates the frequency characteristic of the intermediate point in the adjacent subchannel in addition to the frequency characteristic of the center frequency in the subchannel, so that the combined CMFB 40-1, the combined SMFB 40-2, the analysis The frequency characteristics of the transmission path can be calculated with a resolution higher than the frequency resolution of the CMFBs 70-1 and 70-3 and the analysis SMFBs 70-2 and 70-4.

The equalization coefficient calculation unit 92 receives the frequency characteristic η _{k, i} (i = 0, 1, 2) from the frequency characteristic calculation unit 91, that is, the frequency characteristic η _{k, i} (i = 1) of the center frequency in the subchannel. Frequency characteristics η _{k, i} (i = 0, 2) of intermediate points in adjacent subchannels are input, and an equalization coefficient e _k based on these frequency characteristics η _{k, i} (i = 0, 1, 2) Is output to the equalization processing unit 93.

The equalization processing unit 93 receives the signal I1 from the subtracter 80-1 and the signal Q1 from the adder 80-2, removes the guard band, and receives the real data signals I1 and Q1 in one subchannel. treated as a signal (I1, Q1), enter the equalization coefficient e _k from the equalization coefficient calculation unit 92 calculates the equalized signal by FIR based on the received signal (I1, Q1) and the equalization coefficient e _k, The complex data of the PAM signal x _k [m] is output to the real part extraction unit 100-1.

(Relationship between frequency characteristics and equalization coefficient)
Next, the processing of the equalization coefficient calculation unit 92 in the equalizer 90-1 shown in FIG. 3 will be described in detail. The same applies to the processing of the equalization coefficient calculation unit 92 in the equalizer 90-2.

ｚはｚ変換の変数であり、ｃ_k（−１），ｃ_k（０），ｃ_k（１）はそれぞれ等化係数ｅ_k（−１），ｅ_k（０），ｅ_k（１）の実部データ、ｓ_k（−１），ｓ_k（０），ｓ_k（１）はそれぞれ等化係数ｅ_k（−１），ｅ_k（０），ｅ_k（１）の虚部データである。カッコ内の−１，０，１は、複素３タップの等化係数におけるタップ番号に対応している。 Here, assuming that the equalization processing by the equalization processing unit 93 of the equalizer 90-1 is performed for each sub-channel with a complex 3-tap equalization coefficient, the equalizer 90 in the k-th sub-channel. The transfer function of −1 is expressed by the following equation.

z is a variable of z transformation, and c _k (−1), c _k (0), c _k (1) are equalization coefficients e _k (−1), e _k (0), e _k (1), respectively. S _k (−1), s _k (0), s _k (1) are imaginary part data of equalization coefficients e _k (−1), e _k (0), e _k (1), respectively. It is. -1, 0, 1 in parentheses corresponds to the tap number in the equalization coefficient of the complex 3 taps.

＊は複素共役を表す。ｉ＝０，１，２であり、ｉ＝１のときのχ_k,1は、ｋのサブチャネルにおける中心周波数の等化器の周波数特性である。ｉ＝０のときのχ_k,0は、ｋ−１のサブチャネルにおける中心周波数とｋのサブチャネルにおける中心周波数との間における中間地点の等化器の周波数特性である。ｉ＝２のときのχ_k,2は、ｋのサブチャネルにおける中心周波数とｋ＋１のサブチャネルにおける中心周波数との間における中間地点の等化器の周波数特性である。偶数番目のサブチャネルに対してω = 0、π/2、π、奇数のサブキャリアに対してω = π、3π/2、2πでのＥ_ｋ（ｅｘｐ（ｊω））の値が、ｉ＝０，１，２のときのχ_k,ｉの値となるように、等化係数ｅ_k（−１），ｅ_k（０），ｅ_k（１）を定めることとする。 Assuming that the transfer function of the transmission line is H _ch (z) and the spectral density of the noise of the transmission line is N ₀ , the frequency characteristics of the equalizer of the k subchannels and the equalizer at the midpoint between the adjacent subchannels The frequency characteristic (equalizer frequency characteristic) χ _{k, i} is expressed by the following expression in the MSE standard.

* Represents a complex conjugate. χ _{k, 1} when i = 0, 1, 2 and i = 1 is the frequency characteristic of the equalizer of the center frequency in k subchannels. χ _{k, 0} when i = 0 is the frequency characteristic of the equalizer at the intermediate point between the center frequency in the k−1 subchannel and the center frequency in the k subchannel. χ _{k, 2} when i = 2 is the frequency characteristic of the equalizer at the intermediate point between the center frequency in the k subchannels and the center frequency in the k + 1 subchannels. The value of E _k (exp (jω)) at ω = 0, π / 2, π for even-numbered subchannels and ω = π, 3π / 2, 2π for odd-numbered subcarriers is i = The equalization coefficients e _k (−1), e _k (0), and e _k (1) are determined so as to be the value of χ _{k, i} when 0, 1, and 2.

また、奇数番目のサブチャネルの場合、以下の式で表される。

In the k sub-channels, between the equalization coefficients e _k (−1), e _k (0), e _k (1) and the frequency characteristics χ _{k, i} (i = 0, 1, 2) of the equalizer In the case of an even-numbered subchannel, this relationship is expressed by the following equation.

In the case of an odd-numbered subchannel, it is expressed by the following equation.

以上の関係から、等化係数算出部９２は、ｋのサブチャネルにおける中心周波数の周波数特性及びｋのサブチャネル前後における中間地点の周波数特性、すなわち伝送路の周波数特性η_k,i（ｉ＝０，１，２）を用いて、ｋの等化係数ｅ_k（−１），ｅ_k（０），ｅ_k（１）を算出することができ、周波数特性η_k,i（ｉ＝０，１，２）は、周波数特性算出部９１により算出される。以下、周波数特性算出部９１が周波数特性を算出する処理について説明する。 Here, the frequency characteristic η _{k, 1} of the transmission line is the frequency characteristic of the center frequency in the k subchannels, and η _{k, 0} is the frequency characteristic η _k−1,1 of the center frequency in the k−1 subchannel. frequency characteristics of the intermediate points in between the frequency characteristic eta _{k, 1} of the center frequencies in the sub-channel, eta _{k, 2} a center frequency in the frequency characteristic eta _{k, 1} and k + 1 sub-channel of the center frequency in subchannels k A frequency characteristic at an intermediate point between the frequency characteristic η _{k + 1,1} is used. The frequency characteristic η _{k, i} (i = 0, 1, 2) of the transmission line and the frequency characteristic χ _{k, i} (i = 0, 1 _, 2) of the equalizer are approximated by the following equations.

From the above relationship, the equalization coefficient calculation unit 92 performs the frequency characteristic of the center frequency in the k subchannels and the frequency characteristic of the intermediate point before and after the k subchannels, that is, the frequency characteristic η _{k, i} (i = 0) of the transmission line. , 1, 2), the equalization coefficients e _k (−1), e _k (0), e _k (1) of _k can be calculated, and the frequency characteristic η _{k, i} (i = 0, 1, 2) is calculated by the frequency characteristic calculator 91. Hereinafter, a process in which the frequency characteristic calculation unit 91 calculates the frequency characteristic will be described.

(Processing of the frequency characteristic calculation unit provided in the equalizer)
The processing of the frequency characteristic calculation unit 91 in the equalizer 90-1 shown in FIG. 3 will be described in detail. FIG. 6 is a flowchart showing the processing of the frequency characteristic calculation unit 91, and FIG. 7 is a diagram for explaining the frequency characteristic calculated by the frequency characteristic calculation unit 91. The processing of the frequency characteristic calculation unit 91 in the equalizer 90-2 is the same. Referring to FIG. 6, first, frequency characteristic calculation unit 91 calculates the frequency characteristic η _{k, i} (i = 1) of the center frequency in the subchannel, using the preamble data arranged in the subchannel (step 1). S601). Specifically, the frequency characteristic calculation unit 91 performs k sub-reactions according to a procedure for obtaining an equalization coefficient by an equalizer using a 1-tap equalization coefficient described in Non-Patent Document 2 based on the preamble data. Using all the preamble data arranged in the channel, an equalization coefficient in k subchannels is obtained by MMSE, and this value is obtained as a frequency characteristic χ _{k, i} = 1 / η corresponding to the reciprocal of the transfer function of the transmission path. An approximate value of _{k, i} (i = 1) is assumed. Thus, the frequency characteristic η _{k, i} (i = 1) of the center frequency in the k subchannels is calculated. Similarly, the frequency characteristic calculation unit 91 also performs frequency characteristics η _{k−1, i} , η _{k + 1, i} (i = 1) of the center frequency of the adjacent k−1 subchannels and k + 1 subchannels. The approximate value of is calculated. In this way, for the subchannels in which the preamble data is arranged (k = 0 to 77 (0th to 77th and 178th to 255th) subchannels, see FIG. 4), the black circles in FIG. As indicated by the mark, the frequency characteristic of the center frequency is calculated.

  When calculating the frequency characteristic of the intermediate point in the adjacent sub-channel, the frequency characteristic calculation unit 91 determines whether or not the intermediate point is a region where the frequency characteristic changes due to the influence of HPF or the like (step) S602). Specifically, a frequency region in which the frequency characteristic changes under the influence of HPF or the like is set in advance, and the frequency characteristic calculation unit 91 calculates an intermediate point for calculating the frequency characteristic based on the preset region. When the frequency is included in a preset area, it is determined that the frequency characteristic is changed, and when the frequency at the intermediate point is not included in the preset area, the frequency characteristic is not changed. judge.

  In FIG. 7, the region set in advance for determining whether or not the frequency characteristic is changed under the influence of HPF or the like is a frequency band of k <2. FIG. 7 shows that the influence of HPF or the like that removes the DC offset appears in the frequency band in the range of k <2, and the frequency characteristics are lower than those in the other frequency bands.

Returning to FIG. 6, when it is determined in step S602 that the frequency characteristic is not a region where the frequency characteristic changes (step S602: N), as shown in FIG. Using the frequency characteristic η _{k, 1} and the frequency characteristics η _k−1,1 , η _{k + 1,1} of the center frequency in the adjacent k−1, k + 1 subchannels, interpolation is performed according to the following equation: The frequency characteristics η _{k, 0} and η _{k, 2} at the intermediate point are calculated (step S603).

そして、周波数特性算出部９１は、内挿により算出した中間地点の周波数特性と、外挿により算出した中間地点の周波数特性とを用いて、予め設定された内挿の重み及び外挿の重みにより重み付き平均を算出し、中間地点の周波数特性η_k,0，η_k,2を求める（ステップＳ６０４）。このようにして、図７に示すように、中間地点の周波数特性η_k,0，η_k,2が、図７の黒抜きの丸印が示す内挿により算出した中間地点の周波数特性と、バツ印が示す外挿により算出した中間地点の周波数特性とを用いて、重み付き平均により算出される。 On the other hand, when the frequency characteristic calculation unit 91 determines in step S602 that the frequency characteristic changes (step S602: Y), the frequency characteristic η _{k, 1} of the center frequency in the k subchannels and the adjacent frequency characteristic are adjacent to each other. Using the frequency characteristics η _k−1,1 , η _{k + 1,1} of the center frequencies in the k−1, k + 1 subchannels, interpolation is performed according to the above equation (6), and the frequency characteristics are determined by interpolation at the intermediate point. Is calculated, and extrapolation is performed according to the following formula to calculate the frequency characteristic by extrapolation of the intermediate point.

Then, the frequency characteristic calculation unit 91 uses the frequency characteristic of the intermediate point calculated by the interpolation and the frequency characteristic of the intermediate point calculated by the extrapolation, by the weights of interpolation and extrapolation set in advance. The weighted average is calculated, and the frequency characteristics η _{k, 0} , η _{k, 2} at the intermediate points are _obtained (step S604). In this way, as shown in FIG. 7, the frequency characteristics η _{k, 0} , η _{k, 2} of the intermediate points are calculated by the interpolation shown by the black circles in FIG. The weighted average is calculated using the frequency characteristic of the intermediate point calculated by extrapolation indicated by the cross mark.

Returning to FIG. 6, the frequency characteristic calculation unit 91 outputs the frequency characteristic η _{k, i} (i = 0, 1, 2) calculated in steps S601 to S604 to the equalization coefficient calculation unit 92 (step S605). . The equalization coefficient calculation unit 92 calculates the equalization coefficient using the frequency characteristic η _{k, i} (i = 0, 1, 2), and the equalization processing unit 93 converts the equalization signal using the equalization coefficient. calculate.

ｙは周波数特性であり、ｘはサブチャネル番号に対応した番号でありｋに相当する。ａ，ｂは定数とする。 (Weight of interpolation and extrapolation)
Here, the weight when calculating the frequency characteristic of the intermediate point using the frequency characteristic calculated by the interpolation and the frequency characteristic calculated by the extrapolation will be described. FIG. 8 is a diagram for explaining a method for calculating the frequency characteristic at the intermediate point when the frequency characteristic is approximated by the equation y = (a / x) + b. The frequency characteristic of the predetermined region near the DC changes due to the influence of the HPF having the characteristic of the primary RC filter. It is assumed that the frequency characteristic that changes in a predetermined region close to DC can be approximated by a curve represented by the following equation as shown in FIG. That is, it is assumed that the frequency rejection characteristic by the first-order RC filter corresponds to the curve shown in the following equation.

y is a frequency characteristic, and x is a number corresponding to a subchannel number and corresponds to k. a and b are constants.

The frequency characteristics of k−1, k, and k + 1 are y ₁ , y ₂ , and y ₃ , respectively, and the frequency characteristic of an intermediate point in the adjacent k−1 and k subchannels is y ₄ . Further, the frequency characteristic calculated by interpolation using the frequency characteristics y ₁ and y ₂ is z ₁ (black circle), and the frequency characteristic calculated by extrapolation using the frequency characteristics y ₂ and y ₃ is z. ₂ (X mark). These data are represented by the following equations.

式（１３）が０になるようにＡを決定するものとする。ｋ＝０，１のサブチャネルにおける中間地点の周波数特性を算出する場合は、（ｋ＝０のときの周波数）：（ｋ＝１のときの周波数）＝１：３であるから、ｘ：Δｘ＝３：２となる。したがって、式（１３）が０になるのはＡ＝３／８のときである。また、ｋ＝１，２のサブチャネルにおける中間地点の周波数特性を算出する場合は、（ｋ＝１のときの周波数）：（ｋ＝２のときの周波数）＝３：５であるから、ｘ：Δｘ＝５：２となる。したがって、式（１３）が０になるのはＡ＝９／１６のときである。 The frequency characteristic y ₄ at the intermediate point indicated by the expression (10) is calculated from the frequency characteristic z ₁ calculated by the interpolation indicated by the expression (11) and the frequency characteristic z ₂ calculated by the extrapolation indicated by the expression (12). Ask. Weighted interpolation for extrapolation A (0 ≦ A ≦ 1) , i.e., the weight of interpolation A, the weight of extrapolation as 1-A, the following equations taking the difference between the y _4.

Assume that A is determined so that Equation (13) becomes zero. When calculating the frequency characteristic of the intermediate point in the subchannel of k = 0, 1, since (frequency when k = 0) :( frequency when k = 1) = 1: 3, x: Δx = 3: 2. Therefore, Equation (13) becomes 0 when A = 3/8. Further, when calculating the frequency characteristic of the intermediate point in the subchannel of k = 1, 2, since (frequency when k = 1) :( frequency when k = 2) = 3: 5, x : Δx = 5: 2. Therefore, equation (13) becomes 0 when A = 9/16.

  As described above, assuming that the frequency characteristic in the region close to DC, which changes due to the influence of the HPF having the characteristic of the first-order RC filter, can be approximated by the curve of Expression (8), the intermediate in the subchannel of k = 0, 1 When calculating the frequency characteristic of a point, it is desirable that the weight is set to interpolation weight: extrapolation weight = 3: 5. Further, when calculating the frequency characteristic of the intermediate point in the subchannels of k = 1, 2, it is desirable that the weight is set to interpolation weight: extrapolation weight = 9: 7. In the present invention, the interpolation weight and the extrapolation weight are not limited to the above values, but may be other values.

  As described above, according to the multicarrier demodulation apparatus 2 according to the embodiment of the present invention, a multicarrier signal in which preamble data is arranged in a subchannel closer to DC than a subchannel in which actual data is transmitted by data symbols. Are received from the multicarrier modulation apparatus 1. Then, the frequency characteristic calculation unit 91 of the equalizers 90-1 and 90-2 calculates the frequency characteristic of the center frequency in the subchannel in which the actual data is transmitted and the subchannel outside the subchannel using the preamble data. Further, the frequency characteristic calculation unit 91 calculates the frequency characteristic at the midpoint of the subchannel in the region not affected by HPF or the like by interpolation using the frequency characteristic of the center frequency in the subchannel. Further, when the frequency characteristic calculation unit 91 calculates the frequency characteristic at the midpoint of the subchannel in the region affected by HPF or the like, the frequency characteristic calculation unit 91 calculates the frequency characteristic of the center frequency in the subchannel by interpolation and extrapolation. Then, the result of interpolation and the result of extrapolation are calculated by a weighted average. Here, since the preamble data is arranged not only in the subchannel in which the actual data is transmitted by the data symbols but also in the subchannels outside the subchannel, the frequency characteristics of the intermediate points in all these subchannels are internal. It can be calculated by insertion. As a result, the frequency characteristic at the intermediate point of the subchannel where the frequency characteristic changes due to the influence of HPF or the like is based on the value obtained by interpolation using the frequency characteristic of the center frequency in the adjacent subchannel and the value obtained by extrapolation. Since it is calculated, the value is more accurate than the conventional technique obtained by extrapolation or substitution. Therefore, the error of the equalization coefficient can be reduced and the deterioration of the BER characteristic can be reduced.

  Further, according to the multicarrier demodulator 2 according to the embodiment of the present invention, the calculation of the frequency characteristic at the intermediate point in the frequency band in which the frequency characteristic changes due to the influence of HPF or the like is performed by the interpolation process according to the equation (6) and The extrapolation process according to the equation (7) is performed. Here, the interpolation process according to Expression (6) and the extrapolation process according to Expression (7) can be performed by simple addition / subtraction, multiplication and division. As a result, it is possible to improve the deterioration of the BER characteristic only by adding a simple process with a small amount of calculation while using a simple HPF or the like for removing the DC offset as it is.

  Further, according to the multicarrier demodulator 2 according to the present invention, when calculating the frequency characteristic of the intermediate point in the frequency band where the frequency characteristic changes due to the influence of HPF or the like, k is used as the weight of interpolation and extrapolation. When calculating the frequency characteristics of the intermediate point in the subchannels of 0 and 1, interpolation weight: extrapolation weight = 3: 5 is used. Further, when calculating the frequency characteristic of the intermediate point in the subchannels of k = 1, 2, the interpolation weight: extrapolation weight = 9: 7 is used. In this case, when the frequency characteristic in the DC-closed region that changes due to the influence of the HPF having the characteristic of the first-order RC filter can be approximated by the curve represented by the equation (8), the calculated frequency characteristic of the intermediate point is the above equation. It becomes a value close to (8). Thereby, the frequency characteristic of the calculated intermediate point becomes a highly accurate value. Therefore, the error of the equalization coefficient can be further reduced, and the deterioration of the BER characteristic can be further alleviated.

  Also, according to the multicarrier demodulator 2 of the present invention, as described above, when calculating the frequency characteristics of the intermediate point in the subchannel of k = 0, 1 as the interpolation and extrapolation weights, the interpolation weights : Extrapolation weight = 3: 5 is used, and when calculating the frequency characteristic of the intermediate point in the subchannel of k = 1, 2, the interpolation weight: extrapolation weight = 9: 7 is used. did. Since the total of the interpolation weight and the extrapolation weight is 8, 16, each is a power of 2. Therefore, the interpolation process and the extrapolation process can be performed by addition / subtraction, constant multiplication, and bit shift (calculation of power multiplication of (1/2)). That is, by adding the interpolation weight and the extrapolation weight to a power of 2, a simple HPF or the like that removes the DC offset is used as it is, and simple processing with a smaller amount of computation is added. Thus, it is possible to improve the deterioration of the BER characteristics.

(simulation result)
Hereinafter, simulation results when the multicarrier modulation apparatus 1 and the multicarrier demodulation apparatus 2 according to the embodiment of the present invention shown in FIG. 1 are used will be described. The simulation results (1) to (5) of FIGS. 9 to 13 shown below are described in Non-Patent Document 1 with the tap length L _h = 512 of the prototype filter and the stopband edge frequency f _S = 1.5 / M. It is obtained by the algorithm. Further, in the simulation results (1) to (3) shown in FIGS. 9 to 11 shown below, the equalizer 90-1 and the equalizer 90-2 of the multicarrier demodulator 2 according to the embodiment of the present invention are provided. When the weighted average of interpolation and extrapolation is calculated by the frequency characteristic calculation unit 91, the frequency characteristic of the intermediate point in the subchannel of k = 0, 1 is the weight of interpolation: the weight of extrapolation = 1. : Calculated at 4. Further, the frequency characteristics of the intermediate points in the subchannels of k = 1, 2 and k = 2, 3 are calculated with equality of interpolation weight: extrapolation weight = 1: 1, and other than that The frequency characteristic of the intermediate point is calculated by interpolation from the frequency characteristics of the center frequency in the subchannels on both sides.

  FIG. 9 is a graph showing the simulation result (1). This simulation result (1) shows that the input data is a 16-PAM signal and the impulse response of the transmission path is [1 0 0 0 0 0 0 0.3 * exp (jφ)] (that is, delay 7 samples, level 0. This shows a case of a two-wave model (with a triple delay wave), and HPF cutoff frequencies of 0.5 and 1 times the subchannel spacing. The phase φ of the reflected wave is changed by 0 to 20 degrees, and the one with poor BER characteristics is selected. (A) shows the case of preamble length 10, and (b) shows the case of preamble length 30. In the figure, “prop.-fc: 0.5” represents the C / N vs. BER characteristic when the cutoff frequency of the HPF is 0.5 times the subchannel interval when the embodiment of the present invention is used. ing. “Conv.-fc: 0.5” represents the C / N vs. BER characteristic when the cutoff frequency of the HPF is 0.5 times the subchannel interval in the case where the conventional technique is used. “Prop.-fc: 1” represents the C / N vs. BER characteristics when the cutoff frequency of the HPF is one time the subchannel interval when the embodiment of the present invention is used. “Conv.-fc: 1” represents the C / N vs. BER characteristic when the cutoff frequency of the HPF is one time the subchannel interval in the case where the conventional technique is used.

According to the simulation result (1) of FIG. 9, by using the embodiment of the present invention, the BER of the error floor portion is about 2/3 regardless of whether the preamble length is 10 or (b) 30. It can be seen that Also, depending on the type of error correction to be combined, for example, if the required BER before error correction is 10 ⁻⁴ , when the HPF cutoff frequency is 0.5 times the subchannel interval, (a) the preamble length of 10 It can be seen that the required C / N can be improved by 4 dB or more in the case of (b) when the preamble length is 30 and about 0.5 dB.

FIG. 10 is a graph showing a simulation result (2) at C / N = 45 dB of “prop.-fc: 0.5” in the preamble length 30 of FIG. 9B. The graph of FIG. 10 shows the C / N of “prop.-fc: 0.5” when the input data is a 16-PAM signal, the two-wave model, and the cutoff frequency of the HPF is 0.5 times the subchannel interval. = Shows the error rate for each subchannel at 45 dB. The horizontal axis is k corresponding to the subchannel number, and the vertical axis is the average symbol error rate of x _k [m], x _2M-1-k [m].

  According to the simulation result (2) of FIG. 10, it can be seen that the symbol error rate can be improved to about 2/3 as compared with the simulation result (3) of the prior art shown in FIG.

  FIG. 11 is a graph showing the simulation result (3). This simulation result (3) shows a case where the input data is an 8-PAM signal, a two-wave model, the cutoff frequency of the HPF is one subchannel interval, and the preamble length is 10. In the figure, “prop.-fc: 1” represents the C / N vs. BER characteristic when the cutoff frequency of the HPF is one time the subchannel interval when the embodiment of the present invention is used. “Conv.-fc: 1” represents the C / N vs. BER characteristic when the cutoff frequency of the HPF is one time the subchannel interval in the case where the conventional technique is used.

According to the simulation result (3) of FIG. 11, depending on the type of error correction to be combined, for example, if the required BER before error correction is 10 ⁻⁴ , the required C / N of about 0.5 dB can be improved. Recognize.

  Further, in the simulation results (4) and (5) shown in FIGS. 12 and 13 shown below, the frequencies provided in the equalizer 90-1 and the equalizer 90-2 of the multicarrier demodulator 2 according to the embodiment of the present invention. When the characteristic calculation unit 91 calculates the weighted average of interpolation and extrapolation, the frequency characteristic of the intermediate point in the subchannel of k = 0, 1 is the weight of interpolation: the weight of extrapolation = 3: 5 is calculated. Further, the frequency characteristic of the intermediate point in the subchannel of k = 1, 2 is calculated by the weight of interpolation: weight of extrapolation = 9: 7, and the intermediate frequency in the subchannel of k = 2,3. The frequency characteristics of the points are calculated with an equality of interpolation weight: extrapolation weight = 1: 1. In addition, the frequency characteristics at other intermediate points are calculated by interpolation from the frequency characteristics of the center frequencies in the subchannels on both sides.

  FIG. 12 is a graph showing the simulation result (4). This simulation result (4) shows that when the cutoff frequency of the HPF is designed to be 1% of the baseband band and the cutoff frequency is shifted to 1.3% due to an RC error (see “ Watanabe and 4 others, “High-speed cut-off frequency auto-tuning suitable for WCDMA direct conversion receiver”, IEICE Technical Report ICD 2006-148, (Dec 2006) ” , When the RC variation is assumed to be ± 30%). This simulation result (4) shows a case where the input data is an 8-PAM signal, a two-wave model, and a preamble length of 10. (A) shows a case where the reflection delay is 7 samples and the level is −10 dB, and (b) shows a case where the reflection delay is 46 samples and the level is −26 dB. In the figure, “prop.−1.3%” represents the C / N vs. BER characteristic when the cutoff frequency of the HPF is 1.3% of the baseband band when the embodiment of the present invention is used. Yes. “Conv.—1.3%” represents the C / N vs. BER characteristics when the cutoff frequency of the HPF is 1.3% of the baseband band when the conventional technique is used. “Prop.−1%” represents the C / N vs. BER characteristics when the cutoff frequency of the HPF is 1% of the baseband band when the embodiment of the present invention is used. “Conv.−1%” represents the C / N vs. BER characteristics when the cutoff frequency of the HPF is 1% of the baseband in the case where the conventional technique is used.

According to the simulation result (4) in FIG. 12, depending on the type of error correction to be combined, for example, if the required BER before error correction is 10 ⁻⁴ , the cutoff frequency of the HPF is the baseband band due to the RC error. Even if it becomes 1.3%, it can be seen that the required C / N of 0.5 dB or more in the case of (a) and 3 dB or more in the case of (b) can be improved.

  FIG. 13 is a graph showing the simulation result (5). This simulation result (5) shows the case where the input data is an 8-PAM signal, a two-wave model, and a preamble length of 10, as in the simulation result (4) shown in FIG. FIG. 12A corresponds to FIG. 12A and shows a case where the reflection delay is 7 samples and the level is −10 dB. FIG. 12B corresponds to FIG. 12B, and shows a case where the reflection delay is 46 samples and the level is −26 dB. The horizontal axis represents the cutoff frequency of HPF normalized by the baseband bandwidth, and the vertical axis represents the fixed deterioration amount.

  According to the simulation result (5) in FIG. 13, it can be seen that the deterioration amount when the cutoff frequency of the HPF is shifted is improved in both cases (a) and (b).

  Note that the frequency characteristic calculation unit 91 of the equalizer 90-1 shown in FIG. 3 uses the frequency characteristic of the transmission path at the center frequency of the subchannel for the frequency band near the DC affected by the HPF or the like. Although the frequency characteristics of the transmission path at the intermediate point of the subchannel are obtained by performing the insertion process, extrapolation process, and weighting process, the frequency characteristics of the equalizer may be used. That is, the frequency characteristic calculation unit 91 obtains an equalization coefficient of an equalizer that performs 1-tap complex multiplication for each subchannel by using the input preamble data and the known preamble data, based on the MSE standard. Is the frequency characteristic of the equalizer at the center frequency of the subchannel. Then, with respect to a frequency band closer to DC affected by HPF or the like, the frequency characteristics of the equalizer at the center frequency of the subchannel are used to perform interpolation processing, extrapolation processing, and weighting processing similar to the frequency characteristics of the transmission path. I do. Also, the frequency characteristics of the equalizer at the midpoint of the subchannel are obtained by performing interpolation processing for the frequency band not affected by HPF or the like. Then, the equalization coefficient is derived from the relationship of equations (3) and (4).

  As described above, the frequency characteristics of the transmission line gradually decrease under the influence of HPF or the like in the frequency band close to DC as shown in FIG. On the other hand, the frequency characteristic of the equalizer has a reciprocal relationship with the frequency characteristic of the transmission line as shown in the equation (5), and therefore gradually increases in the frequency band close to DC. In this gradually increasing characteristic region, interpolation processing, extrapolation processing, and weighting processing are performed. As a result, the frequency characteristic can be obtained with high accuracy as in the case of obtaining the frequency characteristic of the transmission line. In addition, the error of the equalization coefficient can be reduced, and the deterioration of the BER characteristic can be reduced.

DESCRIPTION OF SYMBOLS 1 Multicarrier modulation apparatus 2 Multicarrier demodulation apparatus 3 Transmission path 10 Framing part 11 Preamble generation part 12 Data symbol generation part 13 Frame structure part 20-1,80-1,80-4 Subtractor 20-2,52,80 -2 Adder 30-1, 30-2, 51 Multiplier 40-1 Synthesis CMFB
40-2 Synthetic SMFB
50 Quadrature Modulator 60 Quadrature Demodulator 61, 100-1, 100-2 Real Part Extractor 62 Imaginary Part Extractor 70-1, 70-3 Analysis CMFB
70-2, 70-4 Analysis SMFB
80-3 addition inverter 90-1, 90-2 equalizer 91 frequency characteristic calculation unit 92 equalization coefficient calculation unit 93 equalization processing unit 110 deframe conversion unit

Claims (6)

A frame is configured using a preamble in which preamble data used for calculating subchannel equalization coefficients is arranged, and a data symbol in which bit string data is arranged as actual data, and a transmission path as a multicarrier signal. A multi-carrier modulation apparatus that transmits the signal, receives the multi-carrier signal, removes a signal of a predetermined frequency by filtering, calculates a frequency characteristic of a transmission path in a sub-channel based on the preamble data, and In the multicarrier modulation apparatus of the system, comprising: a multicarrier demodulation apparatus that calculates an equalization coefficient based on characteristics, equalizes the received signal based on the equalization coefficient, and restores the original signal to an original bit string ,
The preamble data is arranged in a subchannel where the actual data is arranged, and the preamble is assigned to a subchannel adjacent to the subchannel where the actual data is arranged among subchannels where the actual data is not arranged. A preamble generator for arranging data and generating a preamble;
A frame configuration unit that configures a frame using the preamble generated by the preamble generation unit and the data symbol;
A filter bank that synthesizes signals for each sub-channel of the frame configured by the frame configuration unit;
An orthogonal modulation unit that orthogonally modulates a signal multiplexed by the filter bank;
A multi-carrier modulation apparatus comprising: A multicarrier demodulator that receives a multicarrier signal from the multicarrier modulator of claim 1 and restores the original bit sequence.
An orthogonal demodulator for orthogonally demodulating the multicarrier signal;
A filter for removing a signal of a predetermined frequency from the signal demodulated by the orthogonal demodulation unit;
A filter bank that separates the signal from the filter into signals for each subchannel;
An equalization unit for equalizing the signal separated by the filter bank,
The equalization unit
Based on the preamble data of the subchannel where the actual data is arranged, and the preamble data of the subchannel adjacent to the subchannel where the actual data is arranged among the subchannels where the actual data is not arranged, Calculate the transmission line frequency characteristics of the center frequency in the subchannel where the preamble data is arranged,
Of the subchannels in which the preamble data is arranged, a frequency characteristic for calculating a transmission line frequency characteristic at an intermediate point in the two adjacent subchannels based on a transmission line frequency characteristic of a center frequency in the two adjacent subchannels A calculation unit;
An equalization coefficient calculation unit that calculates an equalization coefficient of a subchannel in which the actual data is arranged based on the transmission path frequency characteristic calculated by the frequency characteristic calculation unit;
A multi-carrier demodulating apparatus comprising: an equalization processing unit that equalizes a signal separated by the filter bank based on the equalization coefficient calculated by the equalization coefficient calculation unit. The multicarrier demodulator according to claim 2,
The frequency characteristic calculation unit provided in the equalization unit,
Based on the preamble data, the transmission channel frequency characteristic of the center frequency in the subchannel in which the preamble data is arranged is calculated,
The first transmission path frequency at the intermediate point in two adjacent subchannels in the predetermined area for a predetermined area that is preset to be affected by the filter among the subchannel areas in which the preamble data is arranged The characteristic is obtained by interpolation, the second transmission line frequency characteristic at the intermediate point is obtained by extrapolation, the first transmission line frequency characteristic and the second transmission line frequency characteristic are averaged by weighting, and the intermediate point is obtained. Calculate the transmission line frequency characteristics of
Of the sub-channel regions where the preamble data is arranged, for the region not including the predetermined region, the transmission path frequency characteristics of the intermediate points in two adjacent sub-channels in the region not including the predetermined region are interpolated. A multi-carrier demodulator characterized by calculating. The multicarrier demodulator according to claim 3, wherein
The interpolating weight and the extrapolating weight are set in advance so that the sum of the interpolating weight and the extrapolating weight becomes a power of 2 value. . The multicarrier demodulator according to claim 3, wherein
The filter is a primary filter composed of a resistor and a capacitor,
The frequency characteristic calculation unit provided in the equalization unit,
Preset interpolation weights and extrapolation are made so as to approximate the frequency characteristics corresponding to the frequency removal characteristics of the primary filter for the predetermined area among the subchannel areas in which the preamble data is arranged. A multi-carrier demodulator that calculates a transmission path frequency characteristic at an intermediate point in two adjacent subchannels in the predetermined area based on the weight of the predetermined frequency. In the multicarrier demodulation device according to any one of claims 2 to 5,
A multicarrier demodulating apparatus characterized in that an equalizing unit frequency characteristic is used instead of the transmission line frequency characteristic.

Images