JP2000295147A - Waveform equalizer, mobile radio equipment using the same, base station radio equipment and mobile communication system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a waveform equalizer with excellent equalization perfor mance even when a propagation path is rapidly fluctuated in a mobile communi cation system. SOLUTION: A forward equalizer 100A that applies waveform equalization to received data in time series sequence of reception and a reverse equalizer 100B that applies waveform equalization to the received data in a reverse sequence of the reception apply waveform equalization to the received data. An adder 117 synthesizes sets of soft decision information being equalization results in every direction by taking an equalization error into account, a discrimination device 118 discriminates the synthesized result to obtain a final equalization output.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a waveform equalizer having good equalization performance even when the propagation path fluctuates greatly, and a portable device using this waveform equalizer to eliminate the influence of frequency selective fading. The present invention relates to a mobile radio device such as a telephone, a car phone, and a private digital radio communication phone, a base station device, and a mobile communication system including the mobile radio device and the base station device.

[0002]

2. Description of the Related Art A conventional waveform equalizer will be described with reference to FIG.

In FIG. 11, an A / D converter 110
1 samples analog received data and converts it into digital data series data. This A / D converter 1
The digital data series data converted in 101 is stored in the reception data buffer 1102 for the equalization processing unit.

[0004] A forward equalizer 1100A is a waveform equalizer that performs equalization in the time series of reception.
3, a determination unit 1104, a switch (SW) 1105, an error calculation unit 1106, and a tap coefficient update unit 1107. The equalization output buffer 1108 stores the equalized output of the forward equalizer 1000, and the equalization error buffer 1109 stores the equalization error for each equalization step in the forward equalization. is there.

[0005] The backward equalizer 1100B is a waveform equalizer that performs equalization in the reverse order of the time series of reception.
11, a determination unit 1112, a switch (SW) 1113, an error calculation unit 1114, and a tap coefficient update unit 1115. The received data time-series inverter 1110 performs pre-processing for reverse equalization, which reverses the time series of the received data. The equalization output buffer 1116 stores the equalized output of the backward equalizer 1100, and the equalization error buffer 1117 stores the equalization error for each equalization step in the backward equalization.

The equalization error comparator 1118 compares the equalization error of the forward equalization stored in the equalization error buffer 1109 with the equalization error of the backward equalization stored in the equalization error buffer 1117. And outputs the equalization direction selection information. The equalization output selection switch (SW) 1119 is connected to the equalization error comparator 1
Based on the equalization direction selection information from 118, one of the equalized output by the forward equalization of the equalized output buffer 1108 and the equalized output by the backward equalization of the equalized output buffer 1116 is selected. This is output as a typical equalization output. Also, the equalized output time series inverter 1120 inverts the time series of the data in the equalized output buffer 1116, returns the equalized output of the backward equalizer 1100 to the time series of reception, and after the backward equalization. The processing is performed.

Next, the operation of the conventional waveform equalizer will be described with reference to FIGS.

FIG. 2 shows received data at time T201.
The data received from time T202 to time T202 is the training sequence 1, the data received from time T202 to time T203 is the data sequence, and the data received from time T203 to time T204 is the training sequence 2 burst.

The received analog received data is A /
The data is sampled by the D converter 1101 and converted into a digital data sequence. Then, the converted digital data sequence is stored in the reception buffer 1102 for the equalization processing unit, that is, one burst shown in FIG.

In the training period for forward equalization, FIG.
Are input to the filter unit 1103 in the order of reception time series. That is, the received data is sequentially input in the direction from time T201 to time T202. The reception data corresponding to this is taken out from the reception data buffer 1102 and input to the filter unit 1103. The filter unit 1103 performs waveform equalization,
Soft decision information is output as an equalization output of forward equalization.

In this training period, the switch 110
Since 5 is connected to the contact A side, the training information is returned to the filter unit 1103. Further, the training information and the soft decision information from the filter unit 1103 are input to an error calculation unit 1106, and an equalization error is calculated. The calculated equalization error and the tap coefficient of the filter unit 1103 are input to the tap coefficient update unit 1107, and a new tap coefficient that reduces the equalization error is calculated.
The calculated tap coefficient is set as a new tap coefficient of the filter unit 1103. The training of the forward equalization is performed by repeating the above operation.

In the tracking period of the forward equalization, FIG.
Are input to the filter unit 1103 in chronological order of reception. That is, from time T202 to time T20
The received data is sequentially input in the direction of 3. The received data corresponding to this is received data buffer 1
It is taken out from 102 and inputted to the filter unit 1103. The filter unit 1103 performs waveform equalization, and outputs soft decision information as an equalized output of forward equalization. The soft decision information output from the filter unit 1103 is
4 and the signal point of the symbol is determined.

In this tracking period, the switch 110
Since No. 5 is connected to the contact B side, the determined result is fed back to the filter unit 1103. The determined result is stored in the equalization output buffer 1108. Further, the result of the determination and the soft decision information from the filter unit 1103 are input to an error calculation unit 1106, and an equalization error is calculated. The calculated equalization error and the tap coefficient of the filter unit 1103 are input to the tap coefficient update unit 1107, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficients are stored in the filter unit 1103.
Is set as the new tap coefficient for. The equalization error for each equalization step of the forward equalization, which is output from the error calculation unit 1106, is stored in the equalization error buffer 1109. The tracking of the forward equalization is performed by repeating the above operation.

In the training period for backward equalization, FIG.
Are input to the filter unit 1111 in the reverse order of the time series of reception. That is, the received data is sequentially input in the direction from time T204 to time T203. The reception data corresponding to this is taken out from the reception data buffer 1102, and input to the filter unit 1111 through the reception data time series inverter 1110. The filter unit 1111 performs waveform equalization, and outputs soft decision information as an equalized output of backward equalization.

In this training period, the switch 111
Since 3 is connected to the contact A side, the training information is fed back to the filter unit 1111. Further, the training information and the soft decision information from the filter unit 1111 are input to an error calculation unit 1114, and an equalization error is calculated. The calculated equalization error and the tap coefficient of the filter unit 1111 are input to the tap coefficient update unit 1115, and a new tap coefficient that reduces the equalization error is calculated.
The calculated tap coefficient is set as a new tap coefficient of the filter unit 1111. The training of the backward equalization is performed by repeating the above-described operation.

In the tracking period of the backward equalization, FIG.
Are input to the filter unit 1111 in the reverse order of the time series of reception. That is, from time T203 to time T
Received data is sequentially input in the direction of 202. The corresponding reception data is taken out from the reception data buffer 1102, and the reception data time series inverter 11
The signal is input to the filter unit 1111 through 10. The filter unit 1111 performs waveform equalization, and outputs soft decision information as an equalized output of backward equalization. Filter unit 111
The soft decision information output from No. 1 is input to the determination unit 1112, and the signal point of the symbol is determined.

In this tracking period, the switch 111
Since 3 is connected to the contact B side, the determined result is fed back to the filter unit 1111. The determined result is stored in the equalization output buffer 1116. Further, the result of the determination and the soft decision information from the filter unit 1111 are input to an error calculation unit 1114, and an equalization error is calculated. The calculated equalization error and the tap coefficient of the filter unit 1111 are input to the tap coefficient update unit 1115, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficients are stored in the filter unit 1111.
Is set as the new tap coefficient for. Further, the equalization error for each equalization step of the backward equalization, which is output from the error calculator 1114, is stored in the equalization error buffer 1117. The tracking of the backward equalization is performed by repeating the above operation.

After the above-described forward equalization and backward equalization processing is completed, one of an equalized output by forward equalization and an equalized output by backward equalization is selected.

That is, the equalization error comparator 1118 includes:
An equalization error due to forward equalization from the equalization error buffer 1109 and an equalization error due to backward equalization from the equalization error buffer 1117 are input. Equalization error comparator 1118
In, some operation is performed on these equalization errors, and equalization direction selection information for selecting an equalization direction with a small equalization error is output.

This equalization direction selection information is input to the equalization output selection switch 1119. If forward equalization is specified in the equalization error selection information, the equalization output selection switch 1119 is selected.
Is set to the contact A side, an equalized output by forward equalization is extracted from the equalized output buffer 1108, and this is the final equalized output. If reverse equalization is specified in the equalization error selection information, the equalization output selection switch 1119
Is set to the contact B side, and the equalized output by the backward equalization is output from the equalized output buffer 1116 to the equalized output time series inverter 11.
20 and this is the final equalized output.

[0021]

The conventional waveform equalizer as described above compares the equalization error between the forward equalization and the backward equalization, and outputs only the equalized output in the direction with a small equalization error. It is configured to be employed. This method is effective when there is almost no change in the propagation path within a burst which is one unit of equalization processing. In addition, since training for forward equalization and training for backward equalization are performed twice, if only one of the training results is good, the equalization characteristics do not deteriorate. This is effective when the propagation path estimation before operation cannot be accurately performed.

However, if the propagation path in a burst, which is one unit of equalization processing, fluctuates greatly, the forward equalization degrades the equalization characteristics at the rear of the burst, and the backward equalization. In this case, the equalization characteristics at the front of the burst tend to deteriorate. As a result, good characteristics cannot be obtained regardless of which equalization direction is selected.

Even though two waveform equalizations of forward equalization and backward equalization are performed for one equalization processing unit, only one of them is selected. There is a problem that processing is wasted.

The present invention has been made to solve such a problem, and it is possible to obtain good equalization performance not only when there is almost no fluctuation of the propagation path in the burst but also when the fluctuation of the propagation path is severe. An object of the present invention is to provide a waveform equalizer, a mobile radio device using the same, a base station radio device, and a mobile communication system.

[0025]

SUMMARY OF THE INVENTION In order to solve the above problems, the present invention relates to a waveform equalizer and a mobile radio apparatus, a base station radio apparatus and a mobile communication system using the same.
A waveform equalizer, a forward equalizer that equalizes the waveform in the time series of reception of received data, a backward equalizer that equalizes the waveform in the reverse order of the time series of reception of the received data, and the forward equalizer. The soft decision information which is the equalization result of the forward equalization of the equalizer and the soft decision information after the time series inversion of the soft decision information which is the equalization result of the backward equalization of the backward equalizer are added. And an estimating means for obtaining an equalized output by judging the result of the addition by the adding means.

As a result, two types of forward equalization and backward equalization can be performed.
The respective soft decision information obtained from the first and second waveform equalizations are combined by focusing on the reception level and the equalization error, and the combined result is determined to obtain the final equalized output.

Thus, according to the present invention, good equalization performance can be obtained not only when there is almost no fluctuation of the propagation path within the burst but also when the fluctuation of the propagation path is severe.

[0028]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A waveform equalizer according to a first aspect of the present invention includes a forward equalizer for equalizing a waveform in the time series of reception of received data, and a reverse equalizer for time series of reception of received data. A backward equalizer that performs waveform equalization, soft decision information that is an equalization result of forward equalization of the forward equalizer, and an equalization result of reverse equalization of the backward equalizer. An addition unit that adds the soft-decision information after time-series inversion of the soft-decision information, and a determination unit that obtains an equalized output by determining the addition result by the addition unit, includes forward equalization. By adding and combining two calculation results of the backward equalization, a diversity effect can be obtained, and there is an effect that waveform equalization can be performed with good performance even when the propagation path fluctuates greatly in a burst.

A waveform equalizer according to a second aspect of the present invention comprises:
A forward equalizer that performs waveform equalization in the time series of reception of received data, a backward equalizer that performs waveform equalization in the reverse order of the time series of reception of the received data, and data based on the training period of the received data. Receiving level detection means for detecting a reception level during a training period for forward equalization and a reception level during a training period for backward equalization; and a soft decision as an equalization result of forward equalization of the forward equalizer. First multiplying means for multiplying information by a reception level during the training period of the forward equalization, and softening information after soft-sequence inversion of soft-decision information which is an equalization result of backward equalization by the backward equalizer. A second multiplication means for multiplying the determination information by a reception level in a training period for backward equalization, an addition means for adding the multiplication results of the first and second multiplication means, and judging an addition result by the addition means By It is provided with a determination means for obtaining an equalized output, utilizing the fact that the higher the reception level during the training period, the better the equalization performance tends to be, and weighting and combining the soft decision information in each direction, and By judging the result, it is possible to perform waveform equalization with good performance even when the propagation path fluctuates greatly in the burst.

A waveform equalizer according to a third aspect of the present invention comprises:
A forward equalizer that equalizes the waveform in the time series of reception of received data, a backward equalizer that equalizes the waveform in the reverse order of the time series of reception of the received data, a forward equalizer and the like of the forward equalizer Weighting is determined for the soft decision information that is the equalization result of the forward equalization of the forward equalizer from the equalization error of each equalization step due to the equalization, and the reverse direction of the backward equalizer is determined. Weighting calculating means for determining a weight for soft decision information, which is an equalization result of backward equalization of the backward equalizer, from an equalization error for each equalization step due to equalization, First multiplication means for multiplying the soft decision information as the equalization result by the weight determined by the weight calculation means, and soft decision information after time series inversion of the soft decision information as the equalization result of the backward equalization Is multiplied by the weight determined by the weight calculating means. A multiplication means, an addition means for adding the multiplication results of the first and second multiplication means, and a judgment means for obtaining an equalized output by judging the addition result by the addition means. Paying attention to the equalization error which is also an index of, the soft decision information in each direction is weighted and synthesized by weighting calculated from this equalization error,
By obtaining a larger diversity effect and judging the synthesis result, there is an effect that waveform equalization can be performed with good performance even when the propagation path fluctuates greatly within a burst.

In the invention according to a fourth aspect of the present invention, the weighting calculating means performs weighting on the soft decision information by forward equalization at each equalization step by forward equalization at each equalization step. It is determined by the reciprocal of the equalization error, and the weight to the soft decision information by the backward equalization for each equalization step is determined by the reciprocal of the equalization error by the backward equalization for each equalization step, By directly weighting the reciprocal of the equalization error for each equalization step, there is an effect that a waveform equalizer with a small amount of calculation and high performance can be provided.

According to a fifth aspect of the present invention, the weight calculation means determines the weight of the soft decision information by the forward equalization for each equalization step by the number of times before and after each equalization step by the forward equalization. Determined by the reciprocal of the sum of the equalization errors for the symbols,
In addition, the weighting of the soft decision information by the backward equalization for each equalization step is determined by the reciprocal of the sum of the equalization errors of several symbols before and after each equalization step by the backward equalization. Not only the equalization error in the equalization step,
The weighting can be determined in consideration of the equalization error before and after that, and thereby, erroneous determination by the waveform equalizer due to the drop of the reception level for a short time can be avoided, and the waveform equalizer with good performance Has the effect of being able to provide

According to a sixth aspect of the present invention, in the weighting means, the equalization error with a forgetting coefficient λ (0 ≦ λ ≦ 1) is introduced for each equalization step of forward equalization. , And weighting the soft decision information by forward equalization at each equalization step is determined by the reciprocal of the forgetting coefficient equalization error at each equalization step by the forward equalization. For each equalization step of equalization, an equalization error with a forgetting coefficient, in which a forgetting coefficient λ (0 ≦ λ ≦ 1) is introduced, is calculated, and weighting is applied to soft decision information by backward equalization for each equalization step. This is determined by the reciprocal of the equalization error with the forgetting coefficient for each equalization step by the backward equalization. When the reception C / N becomes small by introducing the forgetting coefficient λ to the equalization error. The effect of thermal noise is reduced and a high-performance waveform equalizer can be provided. Has an action.

According to a seventh aspect of the present invention, the weight calculating means includes a C / N estimating means, and the C / N estimating means includes a C / N estimation value based on received data, an equalization error, and the like. Is calculated, and a forgetting coefficient λ is set according to the C / N estimation value. Instead of adopting a constant forgetting coefficient, when the reception quality is good, a small forgetting coefficient is used to transmit to the propagation path. In the case where the reception quality is poor, the effect of thermal noise is reduced by using a large forgetting coefficient, and the performance of the waveform equalizer can be further improved.

The invention according to claim 8 of the present invention is a mobile radio apparatus comprising a waveform equalizer for removing the influence of frequency selective fading, wherein the waveform equalizer includes a time series of reception of received data. A forward equalizer that performs waveform equalization in order, a backward equalizer that performs waveform equalization in the reverse order of the time series of reception of received data, and an equalization result of forward equalization performed by the forward equalizer. Adding means for adding the soft decision information and the soft decision information after time series inversion of the soft decision information which is the equalization result of the backward equalization of the backward equalizer, and determining the addition result by the adding means And a determination means for obtaining an equalized output by adding the two operation results of forward equalization and backward equalization, and combining them to obtain a diversity effect. Performs waveform equalization with good performance even when the road fluctuates greatly Preparative because it is possible, has the effect of mobile communication system, the transmission rate of the large zone system can provide good terminal reception performance be used in an environment where a high-speed mobile communication system like the effect of multipath fading can not be ignored.

According to a ninth aspect of the present invention, there is provided a mobile radio apparatus comprising a waveform equalizer for removing the influence of frequency selective fading, wherein the waveform equalizer includes a time series of reception of received data. A forward equalizer that performs waveform equalization in order, a backward equalizer that performs waveform equalization in the reverse order of the time series of reception of received data, and training of forward equalization based on data during a training period of the received data. Reception level detection means for detecting a reception level in a period and a reception level in a training period for backward equalization, soft decision information as an equalization result of forward equalization of the forward equalizer, and the forward equalization First multiplication means for multiplying the reception level during the training period of
Second multiplying means for multiplying the soft decision information after the time series inversion of the soft decision information, which is the equalization result of the backward equalization of the backward equalizer, by the reception level in the training period of the backward equalization, An adder for adding the multiplication results of the first and second multipliers, and a determiner for obtaining an equalized output by determining an addition result by the adder;
Taking advantage of the fact that the higher the reception level during the training period, the better the equalization performance tends to be, the soft decision information in each direction is weighted and combined, and the result is determined. Even if the fluctuations are severe, waveform equalization can be performed with good performance, so reception is possible even when using a large-zone mobile communication system or a mobile communication system with a high transmission rate in an environment where the effects of multipath fading cannot be ignored. This has the effect of providing a high-performance terminal.

According to a tenth aspect of the present invention, there is provided a mobile radio apparatus comprising a waveform equalizer for removing the influence of frequency selective fading, wherein the waveform equalizer includes a time series of reception of received data. A forward equalizer for sequentially equalizing waveforms,
A backward equalizer that performs waveform equalization in the reverse order of the time series of received data, and the forward equalizer based on an equalization error for each equalization step due to forward equalization of the forward equalizer. The weighting is determined for the soft decision information that is the equalization result of the forward equalization, and the equalization error of each equalization step by the backward equalization of the backward equalizer is used to calculate the weight of the backward equalizer. Weighting calculating means for determining a weight for the soft decision information as the equalization result of the backward equalization; and multiplying the soft decision information as the equalization result of the forward equalization by the weighting determined by the weighting calculating means. A first multiplying means for matching, and a second multiplying means for multiplying the soft decision information obtained by time-series inversion of the soft decision information as the equalization result of the backward equalization and the weight determined by the weight calculating means, Add the multiplication results of the first and second multiplication means And a determining means for obtaining an equalized output by determining the result of addition by the adding means. The soft determining information in each direction is weighted and synthesized by weighting calculated from the equalization error. so,
A larger diversity effect can be obtained, and waveform equalization can be performed with good performance even when the propagation path fluctuates greatly in a burst.Therefore, a large-zone mobile communication system or a mobile communication system with a high transmission speed can be used. This has the effect that a terminal having good reception performance can be provided even when used in an environment where the effects of multipath fading cannot be ignored.

According to an eleventh aspect of the present invention, there is provided a base station radio apparatus provided with a waveform equalizer for removing the influence of frequency selective fading, wherein the waveform equalizer is used when receiving received data. A forward equalizer that performs waveform equalization in the sequence order, a backward equalizer that performs waveform equalization in the reverse order of the reception time series of received data, and an equalization result of the forward equalization performed by the forward equalizer. Adding means for adding certain soft decision information and soft decision information after time series inversion of soft decision information which is an equalization result of backward equalization of the backward equalizer, and an addition result by the adding means. A determination means for obtaining an equalized output by making a determination is provided. By adding and combining two operation results of forward equalization and backward equalization, a diversity effect is obtained, and a burst effect is obtained. Waveform equalization with good performance even when the propagation path fluctuates drastically Therefore, even when a large-zone mobile communication system or a mobile communication system with a high transmission rate is used in an environment where the effects of multipath fading cannot be ignored, a base station radio apparatus with good reception performance can be provided. Having.

According to a twelfth aspect of the present invention, there is provided a base station radio apparatus provided with a waveform equalizer for removing the influence of frequency selective fading, wherein the waveform equalizer is used when receiving received data. A forward equalizer that performs waveform equalization in a sequence order, a reverse equalizer that performs waveform equalization in a reverse order of a time series of reception of received data, and a forward equalization based on data of a training period of the reception data. Reception level detection means for detecting a reception level during a training period and a reception level during a training period for backward equalization; soft decision information as an equalization result of forward equalization of the forward equalizer; First multiplying means for multiplying the reception level during the training period for equalization, and the soft decision information obtained by inverting the soft decision information, which is the equalization result of the backward equalization by the backward equalizer, in time series with the soft decision information. Training period The multiplying the reception level 2
Multiplication means, addition means for adding the multiplication results of the first and second multiplication means, and judgment means for obtaining an equalized output by judging the addition result by the addition means. Taking advantage of the fact that the equalization performance tends to be better as the reception level is higher, the soft decision information in each direction is weighted and combined, and the result is determined, so that the fluctuation of the propagation path in the burst is severe. Base station with good reception performance even when using a large-zone mobile communication system or a mobile communication system with a high transmission speed in an environment where the effects of multipath fading cannot be ignored. It has an effect that a station radio apparatus can be provided.

According to a thirteenth aspect of the present invention, there is provided a base station radio apparatus provided with a waveform equalizer for eliminating the influence of frequency selective fading, wherein the waveform equalizer is used when receiving received data. A forward equalizer that performs waveform equalization in a sequence order, a backward equalizer that performs waveform equalization in a reverse order of a time series of reception of received data, and each equalization step by forward equalization of the forward equalizer Weighting is determined for the soft decision information as the equalization result of the forward equalization of the forward equalizer from each equalization error, and each equalization step is performed by the backward equalization of the backward equalizer. Weighting calculating means for determining a weight for the soft decision information which is the equalization result of the backward equalization of the backward equalizer from each equalization error; and a soft weight which is the equalization result of the forward equalization. Multiply the judgment information by the weight determined by the weight calculation means First multiplying means for multiplying the soft decision information obtained by time-series inversion of the soft decision information as the equalization result of the backward equalization and the weight determined by the weight calculating means; An adder for adding the multiplication results of the first and second multipliers, and a determiner for obtaining an equalized output by determining the result of the addition by the adder. By weighting and combining the soft decision information of the directions, a greater diversity effect can be obtained, and waveform equalization can be performed with good performance even when the propagation path fluctuates severely in a burst. Even when a mobile communication system or a mobile communication system with a high transmission rate is used in an environment where the effects of multipath fading cannot be ignored, a base station radio apparatus with good reception performance can be provided. It has the effect of.

According to a fourteenth aspect of the present invention, there is provided a mobile communication system having a base station and a mobile station, wherein at least one of the base station and the mobile station includes a waveform equalizer for removing the influence of frequency selective fading. A system, wherein the waveform equalizer is a forward equalizer that equalizes a waveform in a time series of reception of reception data, and a backward equalizer that equalizes a waveform in a reverse order of a time series of reception of reception data. And after time-series inversion of the soft decision information that is the equalization result of the forward equalization of the forward equalizer and the soft decision information that is the equalization result of the backward equalization of the backward equalizer. An addition means for adding the soft decision information; and a determination means for obtaining an equalized output by determining the result of the addition by the addition means, wherein two operation results of forward equalization and reverse equalization are obtained. And combine them to create a diversity effect. Is, such an action because it is possible to perform the performance good waveform equalization even when the variation of the transfer path in the burst intense, strong multipath fading, can provide a high-quality mobile communication system.

According to a fifteenth aspect of the present invention, there is provided a mobile communication system having a base station and a mobile station, wherein at least one of the base station and the mobile station includes a waveform equalizer for removing the influence of frequency selective fading. A system, wherein the waveform equalizer is a forward equalizer that equalizes a waveform in a time series of reception of reception data, and a backward equalizer that equalizes a waveform in a reverse order of a time series of reception of reception data. Receiving level detection means for detecting a reception level during a training period for forward equalization and a reception level during a training period for backward equalization based on the data during the training period of the received data; and First multiplication means for multiplying the soft decision information, which is the equalization result of the forward equalization, and the reception level during the training period of the forward equalization, and the equalization result of the backward equalization by the backward equalizer; Some soft A second multiplying means for multiplying the soft decision information after the time series inversion of the information and a reception level in a training period for backward equalization, an adding means for adding the multiplication results of the first and second multiplying means, It is provided with a judgment means for obtaining an equalized output by judging the addition result by the addition means, utilizing the fact that the equalization performance tends to be improved as the reception level during the training period is higher,
By weighting and combining the soft decision information in each direction and determining the result, waveform equalization can be performed with good performance even when the propagation path fluctuates drastically within the burst.
This has the effect of providing a high-quality mobile communication system that is resistant to multipath fading.

According to a sixteenth aspect of the present invention, there is provided a mobile communication system having a base station and a mobile station, wherein at least one of the base station and the mobile station includes a waveform equalizer for removing the influence of frequency selective fading. A system, wherein the waveform equalizer is a forward equalizer that equalizes a waveform in a time series of reception of reception data, and a backward equalizer that equalizes a waveform in a reverse order of a time series of reception of reception data. And weighting is determined for soft decision information as an equalization result of the forward equalization of the forward equalizer from an equalization error at each equalization step by the forward equalization of the forward equalizer. And weighting is determined for soft decision information, which is an equalization result of backward equalization of the backward equalizer, from an equalization error at each equalization step due to backward equalization of the backward equalizer. Weighting calculating means, and an equalization result of the forward equalization. First multiplying means for multiplying certain soft decision information by the weight determined by the weight calculating means; soft decision information obtained by time-series inversion of soft decision information as an equalization result of the backward equalization; and the weight calculation Second multiplying means for multiplying the weight determined by the means, and adding means for adding the multiplication results of the first and second multiplying means;
A determining unit for obtaining an equalized output by determining an addition result by the adding unit; and weighting and combining the soft-decision information in each direction by weighting calculated from the equalization error, thereby increasing the diversity. The effect is obtained, and the waveform equalization can be performed with high performance even when the propagation path fluctuates greatly in the burst, so that it is possible to provide a high-quality mobile communication system that is resistant to multipath fading.

(Embodiment 1) Embodiment 1 of the present invention will be described with reference to FIGS. 1, 2 and 3. FIG.

FIG. 1 is a block diagram showing a configuration of a waveform equalizer according to Embodiment 1 of the present invention. Note that FIG.
Is a decision feedback equalizer, and can be implemented by a waveform equalizer having any configuration such as a linear equalizer and a lattice equalizer.

In FIG. 1, a waveform equalizer includes a forward equalizer 100A, a backward equalizer 100B, and an A / D converter 10A.
1, received data buffer 102, soft decision information buffer 1
08, received data time series inverter 109, soft decision information buffer 115, soft decision information time series inverter 116, adder 1
17, a decision unit 118 is provided.

The A / D converter 101 samples analog received data and converts it into a digital data sequence. The digital data sequence converted by the A / D converter 101 is equal to the reception data buffer 10 for the equalization processing unit.
2 is stored.

The forward equalizer 100A is a waveform equalizer that performs equalization in the time series of reception, and includes a filter unit 103, a determination unit 104, a switch (SW) 105, and an error calculation unit 10.
6. It comprises a tap coefficient updating unit 107. The soft decision information buffer 108 stores the equalized output of the forward equalizer 100A.

The backward equalizer 100B is a waveform equalizer that performs equalization in the reverse order of the reception time series, and includes a filter unit 110, a determination unit 111, a switch (SW) 112, an error calculation unit 113, It comprises a tap coefficient updating unit 114. The received data time series inverter 109 inverts the time series of the received data and performs pre-processing for backward equalization. Soft decision information buffer 115 stores soft decision information which is an equalized output of inverse equalizer 100B. The soft decision information time series inverter 116 reverses the time series of the data in the soft decision information buffer 115 and returns the soft decision information, which is the equalized output of the backward equalizer 100B, to the reception time series order. Perform post-processing of equalization.

The adder 117 outputs the soft decision information obtained by the forward equalization in the forward equalizer 100A and the soft decision information obtained by the backward equalization in the backward equalizer 100B. to add. The determiner 118 determines whether the adder 11
The final result of the equalization is obtained by judging the result of addition at step 7. As described above, a bidirectional equalizer is configured by combining the results of the forward equalization and the backward equalization.

FIG. 2 shows received data at time T201.
The data received from time T202 to time T202 is the training sequence 1, the data received from time T202 to time T203 is the data sequence, and the data received from time T203 to time T204 is the training sequence 2 burst.

FIG. 3 is a diagram showing the contents of each buffer and the addition contents of the adder when the data sequence of FIG. 2 has information for n symbols. In FIG. 3, (a)
Shows the contents of the soft decision information buffer 108, (b) shows the contents of the soft decision information buffer 115, and (c) shows the contents of the output data of the adder 117.

Next, the operation of the waveform equalizer according to the first embodiment of the present invention will be described with reference to FIGS.

First, the received analog received data is sampled by the A / D converter 101 and converted into a digital data sequence. Then, the converted digital data sequence is stored in the reception data buffer 102 for an equalization processing unit, that is, one burst shown in FIG.

In the training period for forward equalization, FIG.
Are input to the filter unit 103 in the order of reception time series. That is, the received data is sequentially input in the direction from time T201 to time T202. The corresponding reception data is taken out from the reception data buffer 102 and input to the filter unit 103. The filter unit 103 performs waveform equalization, and outputs soft decision information as an equalized output of forward equalization.

In this training period, the switch 10
Since 5 is connected to the contact A side, the training information is fed back to the filter unit 103. Further, the training information and the soft decision information from the filter unit 103 are input to an error calculation unit 106, and an equalization error is calculated. And
The calculated equalization error and the tap coefficient of the filter unit 103 are input to the tap coefficient update unit 107, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficient is set as a new tap coefficient of the filter unit 103. The training of the forward equalization is performed by repeating the above operation.

In the tracking period of the forward equalization, FIG.
Are input to the filter unit 103 in chronological order of reception. That is, from time T202 to time T203
The received data is sequentially input in the direction of.
The received data corresponding to this is received data buffer 102
And input to the filter unit 103. The filter unit 103 performs waveform equalization, and outputs soft decision information as an equalized output of forward equalization. Filter unit 103
Is output from the soft decision information buffer 108.
Are stored in chronological order of reception. Also, the filter unit 10
The soft decision information output from No. 3 is input to the deciding section 104, and the signal point of the symbol is decided.

In this tracking period, the switch 10
Since 5 is connected to the contact B side, the determined result is fed back to the filter unit 103. Further, the determined result and the soft-decision information from the filter unit 103 are combined with the error calculation unit 1
06 and an equalization error is calculated. The calculated equalization error and the tap coefficient of the filter unit 103 are input to the tap coefficient update unit 107, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficient is set as a new tap coefficient of the filter unit 103. The tracking of the forward equalization is performed by repeating the above operation. Here, when the data sequence is for n symbols, the contents of the soft decision information buffer 108 are as shown in FIG.

In the training period for backward equalization, FIG.
Are input to the filter unit 110 in the reverse order of the received time series. That is, the received data is sequentially input in the direction from time T204 to time T203. The corresponding reception data is taken out from the reception data buffer 102 and input to the filter unit 110 through the reception data time series inverter 109. In the filter unit 110, waveform equalization is performed, and soft decision information is output as an equalized output of backward equalization.

During this training period, the switch 11
Since 2 is connected to the contact A side, the training information is returned to the filter unit 110. Further, the training information and the soft decision information from the filter unit 110 are input to an error calculation unit 113, and an equalization error is calculated. And
The calculated equalization error and the tap coefficient of the filter unit 110 are input to the tap coefficient update unit 114, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficients are set as new tap coefficients of the filter unit 110. The training of the backward equalization is performed by repeating the above-described operation.

In the tracking period of the backward equalization, FIG.
Are input to the filter unit 110 in the reverse order of the time series of reception. That is, from time T203 to time T2
The received data is sequentially input in the direction of 02. The received data corresponding to this is received data buffer 1
02, and is input to the filter unit 110 through the reception data time series inverter 109. Filter unit 11
At 0, waveform equalization is performed, and soft decision information is output as an equalization output for backward equalization. The soft decision information output from the filter unit 110 is stored in the soft decision information buffer 115 in the reverse order of the time series of reception. Also, the soft decision information output from filter section 110 is input to determination section 111, and the signal point of the symbol is determined.

In this tracking period, the switch 11
Since 2 is connected to the contact B side, the determined result is fed back to the filter unit 110. Further, the result of the determination and the soft decision information from the filter unit 110 are combined with the error calculation unit 1.
13 and an equalization error is calculated. The calculated equalization error and the tap coefficient of the filter unit 110 are input to the tap coefficient update unit 114, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficients are set as new tap coefficients of the filter unit 110. The tracking of the backward equalization is performed by repeating the above operation. Here, when the data sequence has n symbols, the content of the soft decision information buffer 115 becomes as shown in FIG.

After the above-described processing of forward equalization and backward equalization is completed, soft decision information by forward equalization and soft decision information by backward equalization are combined. At this time, the soft decision information buffer 108 stores the soft decision information shown in FIG. 3A, and the soft decision information buffer 115 stores the soft decision information shown in FIG. In the combination, the soft decision information corresponding to the reception time is combined.

That is, the soft decision information by the backward equalization is taken out from the soft decision information buffer 115 and input to the adder 117 through the soft decision information time series inverter 116. Soft decision information obtained by forward equalization is extracted from the soft decision information buffer 108 and input to the adder 117 as it is. In the adder 117, soft decision information based on forward equalization and soft decision information based on backward equalization are added, and the addition result is input to the determiner 118. Here, when the data sequence includes n symbols, the addition result input to the determiner 118 is as shown in FIG. The determiner 118 determines the signal point of the symbol, and obtains an equalized output as a final equalized result.

According to the first embodiment of the present invention as described above, a final equalized output is not obtained by selecting forward equalization and backward equalization as in the prior art, but in the forward direction. Since the soft decision information, which is the result of equalization and backward equalization, is combined, a diversity effect can be obtained, and waveform equalization can be performed with good performance even when the propagation path fluctuates greatly in a burst. Can be.

In the case where the equalization direction is selected as in the prior art, the calculation process which is not selected is useless. However, as in the first embodiment of the present invention, the forward equalization and the backward equalization are performed. By using the method of synthesizing the soft-decision information, there is an advantage that the calculation process can be slightly increased and the equalization performance can be improved at the same time.

(Embodiment 2) Next, Embodiment 2 of the present invention will be described with reference to FIGS. 2, 4 and 5. FIG.

FIG. 4 is a block diagram showing a configuration of a waveform equalizer according to Embodiment 2 of the present invention. Note that FIG.
Is a decision feedback equalizer, and can be implemented by a waveform equalizer having any configuration such as a linear equalizer and a lattice equalizer.

In FIG. 4, the waveform equalizer includes a forward equalizer 400A, a backward equalizer 400B, and an A / D converter 40.
1, received data buffer 402, soft decision information buffer 4
08, received data time series inverter 409, soft decision information buffer 415, soft decision information time series inverter 416, reception level detector 417, multipliers 418, 419, adder 420
And a determiner 421.

The A / D converter 401 samples analog received data and converts it into a digital data sequence. The digital data sequence converted by the A / D converter 401 is equal to the reception data buffer 40 for the equalization processing unit.
2 is stored.

The forward equalizer 400A is a waveform equalizer that performs equalization in reception time series, and includes a filter unit 403, a determination unit 404, a switch (SW) 405, and an error calculation unit 40.
6. It is composed of a tap coefficient update unit 407. The soft decision information buffer 408 stores the equalized output of the forward equalizer 400A.

The backward equalizer 400B is a waveform equalizer that performs equalization in the reverse order of the reception time series, and includes a filter unit 410, a determination unit 411, a switch (SW) 412, an error calculation unit 413, It comprises a tap coefficient updating unit 414. The received data time series inverter 409 inverts the time series of the received data, and performs pre-processing for backward equalization. Soft decision information buffer 415 stores soft decision information which is an equalized output of inverse equalizer 400B. The soft decision information time series inverter 416 inverts the time series of the data in the soft decision information buffer 415 and returns the soft decision information, which is the equalized output of the backward equalizer 400B, in the order of reception time series. Perform post-processing of equalization.

The reception level detector 417 detects the reception level while receiving the data during the training period, and the detection signal is output to the multipliers 418 and 419. Multipliers 418 and 419 weight the soft decision information by forward equalization and backward equalization according to the reception level, and output the weighted information to adder 420. The adder 420 adds each of the soft decision information weighted by the multipliers 418 and 419. The determiner 421 determines the result of addition performed by the adder 420 to obtain a final equalized output. In this way, a bidirectional equalizer is configured by combining the results of the forward equalization and the backward equalization.

FIG. 5 is a diagram showing the contents of each buffer and the contents of addition by the adder when the data sequence of FIG. 2 has information for n symbols. In FIG. 5, (a)
Shows the contents of the soft decision information buffer 408, (b) shows the contents of the soft decision information buffer 415, and (c) shows the contents of the output data of the adder 420.

Next, the operation of the waveform equalizer according to the second embodiment of the present invention will be described with reference to FIGS. 2, 4 and 5.

First, the received analog received data is sampled by the A / D converter 401 and converted into a digital data sequence. Then, the converted digital data sequence is stored in the reception data buffer 402 for the equalization processing unit, that is, for one burst shown in FIG.

In the training period for forward equalization, FIG.
Are input to the filter unit 403 in the order of reception time series. That is, the received data is sequentially input in the direction from time T201 to time T202. The corresponding reception data is taken out from the reception data buffer 402 and input to the filter unit 403. The filter unit 403 performs waveform equalization, and outputs soft decision information as an equalized output of forward equalization.

In this training period, the switch 40
Since 5 is connected to the contact A side, the training information is fed back to the filter unit 403. Further, the training information and the soft decision information from the filter unit 403 are input to an error calculation unit 406, and an equalization error is calculated. And
The calculated equalization error and the tap coefficient of the filter unit 403 are input to the tap coefficient update unit 407, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficient is set as a new tap coefficient of the filter unit 403. The training of the forward equalization is performed by repeating the above operation.

In the tracking period of the forward equalization, FIG.
Are input to the filter unit 403 in the order of reception time series. That is, from time T202 to time T203
The received data is sequentially input in the direction of.
The received data corresponding to this is received data buffer 402
And input to the filter unit 403. The filter unit 403 performs waveform equalization, and outputs soft decision information as an equalized output of forward equalization. Filter unit 403
Is output from the soft decision information buffer 408.
Are stored in chronological order of reception. Also, the filter unit 40
The soft decision information output from No. 3 is input to the determination unit 404, and the signal point of the symbol is determined.

In this tracking period, the switch 40
Since 5 is connected to the contact B side, the determined result is fed back to the filter unit 403. Further, the result of the determination and the soft-decision information from the filter unit 403 are combined with the error calculation unit 4.
06 and an equalization error is calculated. The calculated equalization error and the tap coefficient of the filter unit 403 are input to the tap coefficient update unit 407, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficient is set as a new tap coefficient of the filter unit 403. The tracking of the forward equalization is performed by repeating the above operation. Here, when the data sequence has n symbols, the contents of the soft decision information buffer 408 are as shown in FIG.

In the training period for backward equalization, FIG.
Are input to the filter unit 410 in the reverse order of the received time series. That is, the received data is sequentially input in the direction from time T204 to time T203. The reception data corresponding to this is taken out from the reception data buffer 402 and input to the filter unit 410 through the reception data time-series inverter 409. The filter unit 410 performs waveform equalization, and outputs soft decision information as an equalized output of backward equalization.

In this training period, the switch 412
Since is connected to the contact A side, the training information is fed back to the filter unit 410. Further, the training information and the soft decision information from the filter unit 410 are input to an error calculation unit 413, and an equalization error is calculated. Then, the calculated equalization error and the tap coefficient of filter section 410 are input to tap coefficient update section 414, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficient is set as a new tap coefficient of the filter unit 410. Training for backward equalization is performed by repeating the above-described operation.

In the tracking period of the backward equalization, FIG.
Are input to the filter unit 410 in the reverse order of the time series of reception. That is, starting from time T203, T
Received data is sequentially input in the direction of 202. The corresponding reception data is taken out from the reception data buffer 402 and the reception data time-series inverter 409.
Through the filter section 410. The filter unit 410 performs waveform equalization, and outputs soft decision information as an equalized output of backward equalization. The soft decision information output from filter section 410 is stored in soft decision information buffer 415 in chronological order of reception. Also, the soft decision information output from filter section 410 is input to determination section 411, and the signal point of the symbol is determined.

In this tracking period, the switch 41
Since 2 is connected to the contact B side, the determined result is fed back to the filter unit 410. Further, the result of the determination and the soft decision information from the filter unit 410 are combined with the error calculation unit 4.
13 and an equalization error is calculated. The calculated tap error of the filter unit 410 and the calculated equalization error are input to the tap coefficient update unit 414, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficient is set as a new tap coefficient of the filter unit 410. The tracking of the backward equalization is performed by repeating the above operation. Here, when the data sequence has n symbols, the content of the soft decision information buffer 415 at the end of the tracking in the backward equalization is as shown in FIG. 5B.

On the other hand, the received data is received by the reception level detector 41
7 and the reception level during the training period is calculated. That is, the reception level of the part of the training sequence 1 shown in FIG. 2 is calculated, and this is input to the multiplier 418 as a forward training period level. Further, the reception level of the part of the training sequence 2 shown in FIG. 2 is calculated, and this is input to the multiplier 419 as the backward training period level.

After the above-described forward equalization and backward equalization processes are completed, soft decision information based on forward equalization and soft decision information based on backward equalization are combined. At this time, the soft decision information buffer 408 stores the soft decision information shown in FIG. 5A, and the soft decision information buffer 415 stores the soft decision information shown in FIG. In the combination, the soft decision information corresponding to the reception time is combined.

That is, soft decision information by backward equalization is taken out from soft decision information buffer 415 and input to multiplier 419 through soft decision information time series inverter 416. Soft decision information obtained by forward equalization is extracted from the soft decision information buffer 408 and directly input to the multiplier 418.

The multiplier 418 multiplies the soft decision information by forward equalization by the level of the forward training period, and the result of the multiplication is input to the adder 420. Then, the multiplier 419 multiplies the soft decision information by the backward equalization by the backward training period level, and inputs the result of the multiplication to the adder 420. This multiplication process operation is a weighting process before combining soft decision information by forward equalization and soft decision information by backward equalization, and the higher the reception level during the training period, the better the equalization performance tends to be. It is intended to improve the performance by taking advantage of the fact.

The adder 420 adds the soft decision information by forward equalization and the soft decision information by backward equalization weighted by the reception level during the training period, and the addition result is input to the decision unit 421. You. Here, when the data sequence includes n symbols, the addition result input to the determiner 421 is as shown in FIG. Judge 4
At 21, the signal point of the symbol is determined, and an equalized output as a final equalized result is obtained. In FIG. 5,
The reception level during the training period for forward equalization is denoted by a, and the reception level during the training period for backward equalization is denoted by b.

According to the second embodiment of the present invention as described above, the same effects as those of the first embodiment can be obtained.
The soft decision information obtained by the forward equalization and the soft decision information obtained by the backward equalization are added, and a final equalized output is obtained from the addition result. An equalizer can be realized.

(Embodiment 3) Next, Embodiment 3 of the present invention will be described with reference to FIGS. 2, 6, and 7. FIG.

FIG. 6 is a block diagram showing a configuration of a waveform equalizer according to Embodiment 3 of the present invention. Note that FIG.
Is a decision feedback equalizer, and can be implemented by a waveform equalizer having any configuration such as a linear equalizer and a lattice equalizer.

In FIG. 6, the waveform equalizer includes a forward equalizer 600A, a backward equalizer 600B, and an A / D converter 60.
1, received data buffer 602, soft decision information buffer 6
08, an equalization error buffer 609, a received data time series inverter 610, a soft decision information buffer 616, an equalization error buffer 617, an equalization error time series inverter 618, a weight calculation means 619, a weighting buffer 620, and a soft decision information Sequence inverter 621, multipliers 622 and 623, adder 62
4, a determination unit 625 and a C / N estimation unit 626 are provided.

The A / D converter 601 samples analog received data and converts it into a digital data sequence. The digital data sequence converted by the A / D converter 601 is equal to the reception data buffer 60 for the equalization processing unit.
2 is stored.

The forward equalizer 600A is a waveform equalizer that performs equalization in the time series of reception.
Determination unit 604, switch (SW) 605, error calculation unit 6
06, and a tap coefficient updating unit 607. The soft decision information buffer 608 stores the equalized output of the forward equalizer 600A. The equalization error buffer 609 stores an equalization error for each equalization step in forward equalization.

The backward equalizer 600B is a waveform equalizer that performs equalization in the reverse order of the time series of reception, and includes a filter unit 61B.
1, a determination unit 612, a switch (SW) 613, an error calculation unit 614, and a tap coefficient update unit 615. The reception data time series inverter 610 inverts the time series of the reception data, and performs pre-processing for backward equalization. The soft decision information buffer 616 stores the soft decision information which is the equalized output of the inverse equalizer 600B. The equalization error buffer 617 stores an equalization error for each equalization step in backward equalization.

The equalization error time series inverter 618 performs a process of inverting the time series of the data in the equalization error buffer 617 and restoring the time series order of the equalization error corresponding to the equalization output of each step. . The weight calculation unit 619 determines the weight for combining the soft decision information that is the equalized output of the forward equalization and the soft decision information that is the equalized output of the backward equalization from the equalization error in each direction. Further, the weighting buffer 620 stores the weighting determined by the weighting calculating means 619. The soft-decision information time-series inverter 621 includes a soft-decision information buffer 6
Invert the time series of 16 data, and apply the inverse equalizer 600B
The soft-decision information, which is the equalized output of (i), is returned in chronological order of reception.

Multipliers 622 and 623 multiply the weight for forward equalization or backward equalization stored in weighting buffer 620 by soft decision information for forward equalization or soft decision information for backward equalization. , To the adder 624. The adder 624 adds each of the soft decision information weighted by the multipliers 622 and 623. The determiner 625 determines the result of the addition performed by the adder 624 to obtain a final equalized output. In this way, a bidirectional equalizer is configured by combining the results of the forward equalization and the backward equalization.

FIG. 7 is a diagram showing the contents of each buffer and the addition contents of the adder when the data sequence of FIG. 2 has information for n symbols. In FIG. 7, (a)
(B) shows the contents of the equalization error buffer 609, (c) shows the contents of the soft decision information buffer 616, (d) shows the contents of the equalization error buffer 617, (e) and (f) show the contents of the weighting buffer 620, and (g) show the contents of the output data of the adder 624, respectively.

Next, the operation of the waveform equalizer according to the third embodiment of the present invention will be described with reference to FIGS. 2, 6, and 7.

First, the received analog received data is sampled by the A / D converter 601 and converted into a digital data sequence. Then, the converted digital data sequence is stored in the reception data buffer 602 for the equalization processing unit, that is, for one burst shown in FIG.

In the training period for forward equalization, FIG.
Are input to the filter unit 603 in the order of reception time series. That is, the received data is sequentially input in the direction from time T201 to time T202. The corresponding reception data is taken out from the reception data buffer 602 and input to the filter unit 603. The filter unit 603 performs waveform equalization, and outputs soft decision information as an equalized output of forward equalization.

In this training period, the switch 60
Since 5 is connected to the contact A side, the training information is fed back to the filter unit 603. Further, the training information and the soft decision information from the filter unit 603 are input to an error calculation unit 606, and an equalization error is calculated. And
The calculated equalization error and the tap coefficient of the filter unit 603 are input to the tap coefficient update unit 607, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficient is set as a new tap coefficient of the filter unit 603. The training of the forward equalization is performed by repeating the above operation.

In the tracking period of the forward equalization, FIG.
Are input to the filter unit 603 in chronological order of reception. That is, from time T202 to time T203
The received data is sequentially input in the direction of.
The received data corresponding to this is received data buffer 602.
And input to the filter unit 603. The filter unit 603 performs waveform equalization, and outputs soft decision information as an equalized output of forward equalization. Filter section 603
Is output from the soft decision information buffer 608.
Are stored in chronological order of reception. Also, the filter unit 60
The soft decision information output from No. 3 is input to the determination unit 604, and the signal point of the symbol is determined.

In this tracking period, the switch 605
Is connected to the contact B side, the determined result is fed back to the filter unit 603. Further, the result of the determination and the soft decision information from the filter
6 and an equalization error is calculated. Then, the calculated equalization error and the tap coefficient of the filter unit 603 are input to the tap coefficient update unit 607, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficient is set as a new tap coefficient of the filter unit 603. The tracking of the forward equalization is performed by repeating the above operation. Here, when the data sequence is for n symbols, the content of the soft decision information buffer 608 is shown in FIG.
(A), and the equalization error buffer 60
The content of 9 is as shown in FIG.

In the training period for backward equalization, FIG.
Are input to the filter unit 611 in the reverse order of the time series of reception. That is, the received data is sequentially input in the direction from time T204 to time T203. The corresponding reception data is taken out from the reception data buffer 602 and input to the filter unit 611 through the reception data time series inverter 610. The filter unit 611 performs waveform equalization, and outputs soft decision information as an equalized output of backward equalization.

In the training period, the switch 61
Since 3 is connected to the contact A side, the training information is fed back to the filter unit 611. Further, the training information and the soft decision information from the filter unit 611 are input to an error calculation unit 614, and an equalization error is calculated. And
The calculated equalization error and the tap coefficient of the filter unit 611 are input to the tap coefficient update unit 615, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficient is set as a new tap coefficient of the filter unit 611. The training of the backward equalization is performed by repeating the above-described operation.

In the tracking period of the backward equalization, FIG.
Are input to the filter unit 611 in the reverse order of the time series of reception. That is, from time T203 to time T2
The received data is sequentially input in the direction of 02. The received data corresponding to this is received data buffer 6
02, and is input to the filter unit 611 through the received data time series inverter 610. Filter section 61
In step 1, waveform equalization is performed, and soft decision information is output as an equalization output for backward equalization. The soft decision information output from the filter unit 611 is stored in the soft decision information buffer 616 in the reverse order of the reception time series. Also, the soft decision information output from filter section 611 is input to determination section 612, and the signal point of the symbol is determined.

In this tracking period, the switch 613
Is connected to the contact B side, the determined result is fed back to the filter unit 611. Further, the result of the determination and the soft decision information from the filter unit 611 are combined with the error calculation unit 61.
4 and an equalization error is calculated. Then, the calculated equalization error and the tap coefficient of the filter unit 611 are input to the tap coefficient update unit 615, and a new tap coefficient that reduces the equalization error is calculated. The calculated tap coefficient is set as a new tap coefficient of the filter unit 611. Further, the equalization error for each equalization step of the backward equalization, which is the output of the error calculation unit 614, is stored in the equalization error buffer 617 in the reverse order of the time series corresponding to the data storage order of the soft decision information buffer 616. Is stored in The tracking of the backward equalization is performed by repeating the above operation. Here, when the data sequence has n symbols, the content of the soft decision information buffer 616 becomes as shown in FIG.
The contents of the equalization error buffer 617 are as shown in FIG.

After the processing of the forward equalization and the backward equalization described above is completed, each of the soft equalization information based on the forward equalization and the soft equalization information based on the backward equalization is used as information for synthesizing the soft decision information. The weight for the determination information is determined by the weight calculation means 619.

At this time, the equalization error at each equalization step due to forward equalization is stored in the equalization error buffer 609 in FIG.
They are stored in chronological order as shown in (b).
In the equalization error buffer 617, equalization errors for each equalization step by backward equalization are stored in reverse order of time series as shown in FIG. 7D. Also, the weighting is determined by the equalization errors corresponding to the equalization output of each equalization step as the reception time. Therefore, the equalization error due to the backward equalization is taken out from the equalization error buffer 617 and passed through the equalization error time series inverter 618 to calculate the weighting means 619.
Is output to Further, the equalization error due to forward equalization is taken out from the equalization error buffer 609 and output to the weight calculation means 619 as it is.

The weighting calculation means 619 uses the equalization error for each equalization step by forward equalization and backward equalization to obtain soft decision information by forward equalization and soft decision information by reverse equalization. Is determined for each symbol and stored in the weighting buffer 620 in chronological order. Here, when the data sequence corresponds to n symbols, the content of the weighting data stored in the weighting buffer 620 becomes as shown in FIG. 7E when the soft decision information is obtained by the forward equalization, and in the reverse order. FIG. 7F shows the case of soft decision information by directional equalization.

When the weights for the combination are determined, the soft decision information based on the forward equalization and the soft decision information based on the backward forward equalization are combined based on the weighting data stored in the weighting buffer 620. At this time, the soft decision information buffer 60
8 stores soft decision information by forward equalization shown in FIG. 7A, and soft decision information buffer 616 stores soft decision information by reverse forward equalization shown in FIG. 7C. At the time of combining, soft decision information corresponding to the reception time is combined.

Therefore, at the time of synthesis, soft decision information by backward equalization is extracted from the soft decision information buffer 616,
The soft decision information is input to the multiplier 623 through the time series inverter 621. Further, soft decision information obtained by forward equalization is extracted from the soft decision information buffer 608 and directly input to the multiplier 622. Then, weighting data for soft decision information by forward equalization is input from multiplier 622 to weighting buffer 620, and weighting data for soft decision information by backward equalization is input to multiplier 623 from weighting buffer 620. .

The multiplier 622 multiplies the soft-decision information for each data symbol by forward equalization by the corresponding weight, and inputs the result to the adder 624. The multiplier 623 multiplies the soft decision information for each data symbol by backward equalization with the corresponding weight, and inputs the result to the adder 624. The adder 624 performs an addition process on the soft decision information obtained by weighted forward equalization and the soft decision information obtained by weighted backward equalization.
Is input to Here, when the data sequence includes n symbols, the addition result input to the decision unit 625 is as shown in FIG.
(G). The decision unit 625 decides the signal point of the symbol and obtains an equalized output as a final equalization result.

Next, a method of determining the weight by the weight calculating means 619 will be described.

Here, the weight calculation means that the data stored in the equalization error buffer 609 as shown in FIG. 7B and the data stored in the equalization error buffer 617 as shown in FIG. It means that weighting for combining soft decision information by forward equalization and soft decision information by backward equalization is calculated for each equalization step from data.
The calculation result is stored in the weighting buffer 620 as shown in FIGS.

First, the method of determining the weight by the weight calculating means 619 will be described with reference to FIG. FIG.
FIG. 3 is a diagram showing the contents of each buffer when the data sequence shown in FIG. 2 has information for n symbols. 8, (a) shows equalization error data obtained by forward equalization of the equalization error buffer 609, (b) shows weight data for soft decision information obtained by forward equalization of the weighting buffer 620, (c) ) Shows weight data for soft decision information by backward equalization of the weighting buffer 620, and (d) shows equalization error data by backward equalization of the equalization error buffer 617.

In this case, the weight to the soft decision information by the forward equalization shown in FIG. 8B is obtained by comparing the equalization error data (see FIG. 8A) in the equalization error buffer 609 for the forward equalization. The reciprocal is calculated by the weight calculation means 619 for each equalization step. That is, the equalization error data of the forward equalization is expressed as Ef
_i (i = 0, 1,..., n−1), weights a _i (i = 0, 1,...) for soft decision information by forward equalization
·, N-1) is calculated by a _i = 1 / Ef _i.

Similarly, the weight to the soft decision information by the backward equalization shown in FIG. 8C is determined by the equalization error data of the equalization error buffer 617 for the backward equalization (see FIG. 8D). Is calculated by the weight calculation means 619 for each equalization step. That is, the equalization error data of the backward equalization is expressed as E
Assuming that b _i (i = 0, 1,..., n−1), weights b _i (i = 0, 1,...) for soft decision information by backward equalization
·, N-1) is calculated by b _i = 1 / Eb _i.

The weight calculation means 619 is realized by executing the above-described calculation. If the signal level is not related to the determination by the determiner 625, the weighting of the soft-decision information by forward equalization is performed. The weight of the soft decision information by the backward equalization only needs to be the correct ratio. in this case,
Weights _ai (i = 0,
1, ···, n-1) is a a _i = Eb _i, weight b _{i (i} = 0,1 to soft decision information by the backward equalization, ...
·, N-1) may be b _i = Ef _i, it is not necessary to obtain the inverse, it is effective in reducing the amount of calculation.

Next, another embodiment of the weight determining method by the weight calculating means 619 will be described with reference to FIG.

FIG. 9 shows the stored data relating to the symbol near the m-th symbol in the data sequence. In FIG. 9, (a) shows equalization error data obtained by forward equalization of the equalization error buffer 609, and (b) shows a weighting buffer 620.
Weight data to soft decision information by forward equalization of
(C) shows weight data for soft decision information by backward equalization of the weighting buffer 620, and (d) shows equalization error data by backward equalization of the equalization error buffer 617.

In this case, the weight on the soft decision information by the forward equalization shown in FIG. 9B is equal to the equalization error data for several symbols before and after each equalization step by the forward equalization (FIG. 9B).
(See (a)) as the reciprocal of the sum of
9 is calculated for each equalization step.

[0125] For example, when calculating the weight in consideration of the two symbols prior to equalization error data, the weight a _m to soft decision information on forward _{equalization, a m = 1 / (Ef} m -2 + E
f _{m -1} + Ef _m ).

Similarly, the weight to the soft decision information by the backward equalization shown in FIG. 9C is equalized to the equalization error data (FIG. 9) for several symbols before and after each equalization step by the backward equalization.
(D) as the reciprocal of the sum of
9 is calculated for each equalization step.

For example, when the weight is calculated in consideration of the previous two symbols of the equalization error data, the weight b _m for the soft decision information by backward equalization is b _m = 1 / (Eb _{m −2} + E
b _{m -1} + Eb _m ).

The weight calculation means 619 is realized by executing the above-described calculation. If the signal level is not related to the determination by the determiner 625, the weighting of the soft-decision information by forward equalization is performed. The weight of the soft decision information by the backward equalization only needs to be the correct ratio.

Next, still another embodiment of the weight determining method by the weight calculating means 619 will be described with reference to FIG.

FIG. 10 shows the stored data relating to the symbol near the m-th symbol in the data sequence. 10A shows equalization error data obtained by forward equalization of the equalization error buffer 609, and FIG.
Weighting data for soft decision information by forward equalization of 20
(C) shows weight data for soft decision information by backward equalization of the weighting buffer 620, and (d) shows equalization error data by backward equalization of the equalization error buffer 617.

In this case, the weights for the soft decision information obtained by the forward equalization shown in FIG. 10B are obtained by using the equalization error data obtained by the forward equalization (see FIG. 10A). For each step, an equalization error with a forgetting coefficient, in which a forgetting coefficient λ (0 ≦ λ ≦ 1) is introduced, is calculated, and given as the reciprocal thereof.

[0132] Now, to calculate the weight a _m to soft decision information on forward equalization. At this time, since the equalization error with the forgetting coefficient for the symbol m-1 is given as the reciprocal of the weight for the symbol m-1, it becomes 1 / am _-1 . Using this, the equalization error with the forgetting factor for the symbol m is λ /
a _m-1 + Ef _m becomes, the weight a _m is the reciprocal of, a
_m = 1 / (λ / a _m-1 + Ef _m ).

Similarly, the weights for the soft decision information obtained by the backward equalization shown in FIG. 10C are calculated with respect to the equalization error data obtained by the backward equalization (see FIG. 10D). For each step, an equalization error with a forgetting coefficient, in which a forgetting coefficient λ (0 ≦ λ ≦ 1) is introduced, is calculated, and given as the reciprocal thereof.

Now, a weight b _m for soft decision information by backward equalization is calculated. At this time, since the equalization error with the forgetting coefficient for the symbol m-1 is given as the reciprocal of the weight of the symbol m-1, it becomes 1 / b _m-1 . Using this, the equalization error with the forgetting factor for the symbol m is λ /
b _m-1 + Ef _m becomes, the weight b _m is the reciprocal, b
_m = 1 / (λ / b _m-1 + Ef _m ).

The weight calculating means 619 is realized by executing the above-described calculation. If the signal level is not related to the determination by the determiner 625, the weighting of the soft-decision information by forward equalization is performed. The weight of the soft decision information by the backward equalization only needs to be the correct ratio.

Next, a case will be described in which a weighting calculation is performed by adding a C / N (average carrier power to noise power ratio) estimating means 626 to the weighting calculating means 619 as shown in FIG.

When the C / N ratio of the received data is good, a small value is set for the forgetting coefficient λ (0 ≦ λ ≦ 1) so that the equalization performance is not deteriorated and the fluctuation of the propagation path is quickly followed. A waveform equalizer that performs Also, the C / N of the received data
Is not good, by setting a large value for the forgetting coefficient λ, the influence of thermal noise can be minimized, which contributes to the improvement of the equalization performance.

Accordingly, the C / N estimating means 626 calculates the C / N estimated value by inputting the received data, the equalization error and the like, and weights this C / N estimated value to the weighting calculating means 619.
To enter. The weighting calculating means 619 sets a small forgetting coefficient λ when the estimated C / N value is large, and sets the C / N
When the estimated value is small, the weight is calculated by setting a large forgetting coefficient λ.

As described above, in the third embodiment, in combining soft decision information by forward equalization and soft decision information by backward equalization, the weight is changed for each data symbol. Therefore, as compared with the second embodiment in which constant weighting is performed for each equalization processing unit, such as one entire burst, the waveform equalizer is more resistant to channel fluctuations and has better equalization performance. Can be realized.

The degree of improvement in the performance of the waveform equalizer changes depending on the weighting method used by the weighting calculating means 619. When this is realized in, for example, a mobile communication system, the desired performance of the entire system, the condition of the propagation path, Alternatively, the selection may be made according to the margin of the amount of calculation of the LSI mounted on the mobile wireless device.

(Embodiment 4) In a large-zone mobile communication system or a mobile communication system with a high transmission rate, if a plurality of radio wave propagation paths exist between a base station and a mobile station, these paths may be used. The effect of multipath fading, such as the occurrence of intersymbol interference due to the difference, cannot be ignored.

However, in the present invention, the above-mentioned problem can be solved by using the waveform equalizer having the configuration shown in FIGS. 1, 4 and 6 described above. Therefore, the mobile radio apparatus and the base station radio apparatus using the waveform equalizer of the present invention become a terminal and a base station infrastructure having good reception performance, and furthermore, a mobile communication system constructed by a combination of them becomes a multipath fading. A high-quality system that is strong against

[0143]

As described above, according to the waveform equalizer of the present invention, a diversity effect can be obtained by adding and combining two operation results of forward equalization and backward equalization. Waveform equalization can be performed with good performance even when the propagation path fluctuates greatly within a burst.

According to the waveform equalizer of the present invention, the soft decision information in each direction is weighted by utilizing the fact that the higher the reception level during the training period, the higher the equalization performance tends to be. By combining and judging the result, even when the propagation path fluctuates drastically within the burst, it is possible to follow up with good performance and perform waveform equalization with better performance.

Further, according to the waveform equalizer of the present invention, attention is paid to the equalization error which is also an index of the equalization performance, and the soft decision information in each direction is weighted and synthesized by weighting calculated from the equalization error. In this way, a greater diversity effect can be obtained, and by determining the synthesis result, waveform equalization can be performed with good performance even when the propagation path fluctuates drastically within a burst.

Further, according to the waveform equalizer of the present invention, by using the reciprocal of the equalization error for each equalization step as it is, the waveform equalizer with a small amount of calculation and high performance can be provided. .

Further, according to the waveform equalizer of the present invention, it is possible to determine the weight by taking into account not only the equalization error in a certain equalization step but also the equalization errors before and after the equalization step. An erroneous determination in the waveform equalizer due to a drop in the reception level for a short time can be avoided, and a high-performance waveform equalizer can be provided.

Further, according to the waveform equalizer of the present invention, the reception C / C
The effect of thermal noise when N becomes small is reduced, and a waveform equalizer with good performance can be provided.

According to the waveform equalizer of the present invention, C /
By setting the forgetting factor according to the N estimation value, the following speed is improved by using a small forgetting factor when the reception quality is good, and by using a large forgetting factor when the reception quality is bad. The effect of thermal noise can be reduced, and the performance of the waveform equalizer can be further improved.

Further, according to the present invention, in a large zone mobile communication system or a mobile communication system having a high transmission rate, a mobile radio terminal or base station used in an environment where the effects of multipath fading cannot be ignored. Also in the infrastructure, the reception performance can be improved in response to various propagation path conditions.

According to the present invention, a high-quality mobile communication system resistant to multipath fading can be provided, and a large-zone mobile communication system and a high-speed mobile communication system can be easily constructed. Has an effect.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a waveform equalizer according to Embodiment 1 of the present invention.

FIG. 2 is an explanatory diagram of received data.

FIG. 3 is an explanatory diagram showing the contents of each buffer when a data sequence has information for n symbols in Embodiment 1 of the present invention.

FIG. 4 is a configuration diagram of a waveform equalizer according to a second embodiment of the present invention.

FIG. 5 is an explanatory diagram showing contents of each buffer and addition contents of an adder when a data sequence has information for n symbols in the second embodiment of the present invention.

FIG. 6 is a configuration diagram of a waveform equalizer according to Embodiment 3 of the present invention.

FIG. 7 is an explanatory diagram showing contents of each buffer and addition contents of an adder when a data sequence has information for n symbols in Embodiment 3 of the present invention.

FIG. 8 is an explanatory diagram showing the contents of each buffer when calculating weights according to the present invention;

FIG. 9 is an explanatory diagram showing the contents of each buffer when calculating weights according to the present invention;

FIG. 10 is an explanatory diagram showing the contents of each buffer when calculating weights according to the present invention;

FIG. 11 is a configuration diagram of a conventional waveform equalizer.

[Explanation of symbols]

 100A, 400A, 600A Forward equalizer 100B, 400B, 600B Reverse equalizer 101, 401, 601 A / D converter 102, 402, 602 Receive data buffer 103, 403, 603 Filter section 104, 404, 604 Judgment unit 105, 405, 605 Switch 106, 406, 606 Error calculation unit 107, 407, 607 Tap coefficient update unit 108, 408, 608 Soft decision information buffer 109, 409, 610 Received data time series inverter 110, 410, 611 Filter section 111, 411, 612 Judgment section 112, 412, 613 Switch 113, 413, 614 Error calculation section 114, 414, 615 Tap coefficient update section 115, 415, 616 Soft decision information buffer 116, 416, 621 For soft decision information Series inverter 17, 420, 624 Adder 118, 421, 625 Judge 417 Reception level detector 418, 419, 622, 623 Multiplier 609 Equalization error buffer 617 Equalization error buffer 618 Equalization error time series inverter 619 Weight calculation means 620 Weighting buffer 626 C / N estimation means

 ──────────────────────────────────────────────────続 き Continuing from the front page (72) Inventor Kazuyuki Aota 4-3-1 Tsunashima Higashi, Kohoku-ku, Yokohama-shi, Kanagawa Prefecture Inside Matsushita Communication Industrial Co., Ltd. No.3-1, Matsushita Communication Industrial Co., Ltd. F-term (reference)

Claims (16)

[Claims] 1. A forward equalizer for equalizing a waveform in a time series of reception of received data, a backward equalizer for equalizing a waveform in a reverse order of time series of reception of the received data, and the forward equalization The soft decision information which is the equalization result of the forward equalization of the equalizer and the soft decision information after the time series inversion of the soft decision information which is the equalization result of the backward equalization of the backward equalizer are added. A waveform equalizer, comprising: an adding unit; and a determining unit that obtains an equalized output by determining an addition result by the adding unit. 2. A forward equalizer for equalizing a waveform in a time series of reception of received data, a backward equalizer for equalizing a waveform in a reverse order of time series of reception of the received data, and training of the received data. Reception level detection means for detecting a reception level during a training period for forward equalization and a reception level during a training period for backward equalization based on period data, and equalization for forward equalization of the forward equalizer First multiplying means for multiplying the soft decision information as a result by the reception level during the training period of the forward equalization, and a time series of soft decision information as an equalization result of the backward equalization of the backward equalizer Second multiplying means for multiplying the inverted soft decision information by the reception level during the training period for backward equalization, adding means for adding the multiplication results of the first and second multiplying means, and addition by the adding means Judge the result A waveform equalizer comprising: a determination unit that obtains an equalized output by performing the above operation. 3. A forward equalizer for equalizing a waveform in a time series of reception of reception data, a backward equalizer for equalizing a waveform in a reverse order of reception time of reception data, and the forward equalization. Weighting is determined for soft decision information as an equalization result of forward equalization of the forward equalizer from the equalization error at each equalization step due to forward equalization of the equalizer, and the backward equalization is performed. Weighting calculating means for determining a weight for soft decision information that is an equalization result of backward equalization of the backward equalizer from an equalization error for each equalization step due to backward equalization of the unit, First multiplication means for multiplying the soft decision information as the equalization result of the forward equalization by the weight determined by the weight calculation means, and time-series inversion of the soft decision information as the equalization result of the backward equalization After the soft decision information and the weight determined by the weight calculation means, A second multiplying means for multiplying; an adding means for adding the multiplication results of the first and second multiplying means; and a judging means for obtaining an equalized output by judging the addition result by the adding means. Waveform equalizer. 4. The weighting calculating means determines weighting of the soft decision information by forward equalization for each equalization step by a reciprocal of an equalization error by forward equalization for each equalization step,
4. The waveform according to claim 3, wherein weighting of soft decision information by inverse equalization at each equalization step is determined by a reciprocal of an equalization error by inverse equalization at each equalization step. Chemist. 5. The weighting calculating means calculates the weighting of the soft decision information by forward equalization for each equalization step as the reciprocal of the sum of the equalization errors for several symbols before and after each equalization step by forward equalization. And the weighting of the soft decision information by the backward equalization for each equalization step is determined by the reciprocal of the sum of the equalization errors of several symbols before and after each equalization step by the backward equalization. The waveform equalizer according to claim 3, wherein: 6. A weighting calculating means calculates an equalization error with a forgetting coefficient in which a forgetting coefficient λ (0 ≦ λ ≦ 1) is introduced for each equalizing step of forward equalization, and calculates an equalizing error for each equalizing step. The weighting of the soft decision information by forward equalization is determined by the reciprocal of the equalization error with the forgetting coefficient for each equalization step by the forward equalization, and the forgetting factor is set for each equalization step of the backward equalization. Calculates an equalization error with a forgetting coefficient in which λ (0 ≦ λ ≦ 1) is introduced, and weights soft decision information by backward equalization in each equalization step for each equalization step by the backward equalization. 4. The waveform equalizer according to claim 3, wherein the waveform equalizer is determined by a reciprocal of an equalization error with a forgetting coefficient. 7. The weighting calculating means includes a C / N estimating means, and the C / N estimating means calculates a C / N estimated value based on received data, an equalization error, and the like, and calculates the C / N estimated value. 7. The waveform equalizer according to claim 6, wherein the forgetting coefficient λ is set according to the following equation. 8. A mobile radio apparatus comprising a waveform equalizer for removing the influence of frequency-selective fading, wherein the waveform equalizer performs a waveform equalization in a time-series order of reception of received data. And a backward equalizer that performs waveform equalization in the reverse order of the time series of reception of received data, soft decision information that is an equalization result of forward equalization of the forward equalizer, and the reverse equalizer Adding means for adding the soft-decision information, which is the equalization result of the backward equalization, after time-series inversion, to soft-decision information, and determining means for determining an addition result by the adding means to obtain an equalized output A mobile wireless device comprising: 9. A mobile radio apparatus comprising a waveform equalizer for removing the influence of frequency-selective fading, wherein the waveform equalizer performs a waveform equalization in a time series of reception of received data. And a backward equalizer that performs waveform equalization in the reverse order of the time series of reception of received data, and a reception level and a backward equalization for a training period of forward equalization based on data of the training period of the received data. A reception level detection unit for detecting a reception level in a training period, and a soft level information that is a result of equalization of the forward equalization of the forward equalizer multiplied by a reception level in the training period for the forward equalization. Multiplying the soft decision information after time series inversion of the soft decision information, which is the equalization result of the backward equalization by the backward equalizer, with the reception level in the training period of the backward equalization; Squared Means and, an adding means for adding the multiplication result of the first and second multiplying means, the mobile radio apparatus characterized by comprising determination means for obtaining an equalized output by determining the addition result by the adding means. 10. A mobile radio apparatus comprising a waveform equalizer for removing the influence of frequency selective fading, wherein the waveform equalizer performs a waveform equalization in a time series order of reception of received data. And a backward equalizer that performs waveform equalization in the reverse order of the time series of reception of the received data, and the forward equalization based on an equalization error for each equalization step due to forward equalization of the forward equalizer. The weighting is determined for the soft decision information which is the equalization result of the forward equalization of the equalizer, and the backward equalization is performed from the equalization error at each equalization step by the backward equalization of the backward equalizer. Weighting calculating means for determining a weight for soft decision information which is an equalization result of backward equalization of the device,
First multiplying means for multiplying the soft decision information as the equalization result of the forward equalization by the weight determined by the weight calculation means, and time-series inversion of the soft decision information as the equalization result of the backward equalization Second multiplication means for multiplying the soft decision information after the weighting and the weight determined by the weight calculation means,
A mobile radio apparatus comprising: an adding unit that adds the multiplication results of the first and second multiplication units; and a determining unit that obtains an equalized output by determining a result of the addition by the adding unit. 11. A base station radio apparatus comprising a waveform equalizer for removing the influence of frequency selective fading, wherein the waveform equalizer performs forward equalization for equalizing a waveform in a time series of reception of received data. And a backward equalizer that performs waveform equalization in the reverse order of the time series of reception of received data, soft decision information that is a result of forward equalization performed by the forward equalizer, and the reverse equalization. Adding means for adding the soft-decision information, which is the equalization result of the inverse equalization by the time series, to the soft-decision information after time-series inversion, and determining that an equalized output is obtained by determining the addition result by the adding means. A base station radio apparatus comprising means. 12. A base station radio apparatus comprising a waveform equalizer for removing the influence of frequency-selective fading, wherein the waveform equalizer performs forward equalization for equalizing a waveform in a time-series order of reception of received data. An equalizer for performing waveform equalization in the reverse order of the time series of reception of received data, and a reception level and a backward equalization for a training period for forward equalization based on the data for the training period for the received data. Receiving level detecting means for detecting the receiving level during the training period, and multiplying the soft decision information as the equalization result of the forward equalization by the forward equalizer with the receiving level during the training period for the forward equalization. The first multiplication means multiplies the soft decision information obtained by time-series inversion of the soft decision information, which is the equalization result of the backward equalization by the backward equalizer, by a reception level in a training period of the backward equalization. No. A base station radio comprising: a multiplication unit; an addition unit for adding the multiplication results of the first and second multiplication units; and a judgment unit for obtaining an equalized output by judging the addition result by the addition unit. apparatus. 13. A base station radio apparatus comprising a waveform equalizer for removing the influence of frequency-selective fading, wherein the waveform equalizer performs forward equalization for equalizing a waveform in a time series of reception of received data. An equalizer, a backward equalizer that performs waveform equalization in the reverse order of the time series of reception of received data, and a forward equalizer that calculates an equalization error for each equalization step due to forward equalization of the forward equalizer. The weighting is determined for the soft decision information as the equalization result of the forward equalization of the equalizer, and the equalization error for each equalization step due to the equalization of the backward equalizer by the backward equalization is used to calculate the backward equalization. Weight calculation means for determining weights for soft decision information that is an equalization result of the backward equalization of the equalizer,
First multiplying means for multiplying the soft decision information as the equalization result of the forward equalization by the weight determined by the weight calculation means, and time-series inversion of the soft decision information as the equalization result of the backward equalization Second multiplying means for multiplying the soft decision information after the weighting and the weight determined by the weight calculating means,
A base station radio apparatus comprising: an adding unit that adds the multiplication results of the first and second multiplication units; and a determining unit that determines an addition result by the adding unit to obtain an equalized output. 14. A mobile communication system including a base station and a mobile station, wherein at least one of the base station and the mobile station includes a waveform equalizer for removing an influence of frequency selective fading, wherein the waveform equalizer Is a forward equalizer that performs waveform equalization in the order of reception data reception time series, a backward equalizer that performs waveform equalization in the reverse order of reception data reception time series, and a forward equalizer. Adding means for adding the soft decision information that is the equalization result of the direction equalization and the soft decision information after the time series inversion of the soft decision information that is the equalization result of the backward equalization of the backward equalizer. A mobile communication system comprising: a determination unit that obtains an equalized output by determining a result of addition by the addition unit. 15. A mobile communication system including a base station and a mobile station, wherein at least one of the base station and the mobile station includes a waveform equalizer that removes the influence of frequency selective fading, wherein the waveform equalizer Is a forward equalizer that equalizes the waveform in the time series of reception of received data, a backward equalizer that equalizes the waveform in the reverse order of the time series of reception of the received data, and data during a training period of the received data. And a reception level detecting means for detecting a reception level during a training period for forward equalization and a reception level during a training period for backward equalization based on the above, and an equalization result of forward equalization of the forward equalizer. First multiplying means for multiplying the soft decision information by the reception level during the training period of the forward equalization, and after time-series inversion of the soft decision information as the equalization result of the backward equalization by the backward equalizer Soft A second multiplying means for multiplying the information by a reception level in a training period for backward equalization, an adding means for adding the multiplication results of the first and second multiplying means, and determining the addition result by the adding means A mobile communication system comprising a determination unit for obtaining an equalized output. 16. A mobile communication system including a base station and a mobile station, wherein at least one of the base station and the mobile station includes a waveform equalizer for removing an influence of frequency selective fading, wherein the waveform equalizer Is a forward equalizer that performs waveform equalization in the order of reception data reception time series, a backward equalizer that performs waveform equalization in the reverse order of reception data reception time series, and a forward equalizer. From the equalization error for each equalization step due to direction equalization, weighting is determined for soft decision information that is the result of forward equalization of the forward equalizer, and the inverse of the inverse equalizer is determined. Weight calculation means for determining a weight for soft decision information that is an equalization result of backward equalization of the backward equalizer from the equalization error for each equalization step by direction equalization,
First multiplying means for multiplying the soft decision information, which is the equalization result of the forward equalization, with the weight determined by the weight calculating means; and time-series inversion of the soft decision information, which is the equalization result of the backward equalization, Second multiplying means for multiplying the soft decision information after the weighting and the weight determined by the weight calculating means,
A mobile communication system comprising: an adding unit that adds the multiplication results of the first and second multiplication units; and a determining unit that determines an addition result by the adding unit to obtain an equalized output.