KR20010011216A - smart antenna system using the structure of prebeamformer about each multipath and adaptive equalization combiner each frequency bins - Google Patents

Abstract

PURPOSE: A smart antenna system with a structure of a pre-beam forming device by a multi-path and an adaptive equalization coupler by a frequency component of a multi-path signal are provided to reduce the inter-symbol interference or the co-channel interference by using a pre-beam forming device for each multi-path signal and a delay time compensation circuit. CONSTITUTION: A smart antenna system with a structure of a pre-beam forming device by a multi-path and an adaptive equalization coupler by a frequency component of a multi-path signal comprises a synchronization portion(10), an estimation portion(20), a pre-beam forming portion(30), a delay time compensation portion(40), and an adaptive equalization portion(50a,50b). The synchronization portion(10) synchronizes the received signal. The estimation portion(20) estimates a delay time for a multi-path by using a characteristic of a pilot signal and a pre-beam forming weight value vector by using the estimated delay time. The pre-beam forming portion(30) forms an independent beam by the multi-path signals. The delay time compensation portion(40) compensates the delay time. The adaptive equalization portion(50a,50b) equalizes signals according to frequencies of signals of each path.

Description

Smart antenna system using the structure of prebeamformer about each multipath and adaptive equalization combiner each frequency bins}

The present invention relates to a smart antenna system, and more particularly, to an adaptive equalizer in a frequency domain using an antenna array using a pre-beam shaping and a delay time compensation circuit for each multipath signal.

As information becomes high-speed broadband, the system requires a high operating frequency, and delay spread due to multipath causes severe inter-symbol interference (ISI).

This is an important factor limiting the operation of the wireless communication system, and a channel equalizer is required to solve this problem.

The use of a high operating frequency gradually narrows the service radius, resulting in closer distances between adjacent base station cells using the same channel.

This causes an increase in co-channel interference (CCI) between neighboring base station cells, bringing about system efficiency and network performance degradation.

Conventional time-domain channel equalizers require a large number of Tap Delay Line (TDL) structures due to an increase in delay spread due to high-speed broadband of information.

This complexity of the channel equalizer makes it difficult to implement a real-time system.

Therefore, a channel equalizer structure in the frequency domain has been proposed.

The channel equalizer system in the frequency domain converts the wideband signal into the frequency domain and performs adaptive signal processing for each frequency component.

By doing so, the converted signal can be assumed to be a narrowband signal for each frequency component, so that the structure of the channel equalizer is very simple and less complicated than that of the TDL structure.

However, it shows slow convergence performance in the co-channel interference (CCI) environment.

Recently, smart antenna technology using an antenna array structure has been considered to be essential when implementing high-speed wireless communication systems because it provides many advantages such as extended service range, increased capacity, increased system efficiency, and ease of real-time system implementation.

In addition, since the smart antenna technology can spatially separate multipath signals to reduce the effective delay spread, the smart antenna technology has emerged as an important means for mitigating the requirements of the channel equalizer as the speed of information increases.

However, the channel equalizer in the frequency domain using the antenna array according to the related art described above has the following problems.

First, since the complexity of the incidence direction estimation and the implementation of the adaptive algorithm of the multipath signal is large, there are many difficulties in real-time implementation.

Second, it shows slow convergence performance in the same channel interference (CCI) environment.

Accordingly, the present invention has been made to solve the above problems, and in the high-speed wireless communication system of a single carrier and OFDM transmission system having a cyclic prefix frame structure, between the received signal having a cyclic prefix Estimating the delay time of multipath signals using cyclic and autocorrelation characteristics, estimating a pre-beamforming weight vector for independent beamforming for each multipath signal, and selecting the multipath signal selected from the estimated results The purpose is to effectively remove co-channel interference signals by inputting them to the equalizer through pre-beam formation and delay compensation, and to optimally combine the selected multipath signals using an adaptive equalizer in the frequency domain.

1 is a configuration diagram of a single carrier wideband smart antenna system having a cyclic prefix frame structure according to the present invention

2 is a configuration diagram of an adaptive equalizer in a frequency domain in the case of a single carrier transmission method according to the present invention

3 is a block diagram of a broadband smart antenna system employing the OFDM transmission method according to the present invention

4 is a block diagram of an adaptive equalizer in the frequency domain in the OFDM transmission method according to the present invention

5 is a schematic diagram of a prebeam former bank according to the present invention;

6 is a block diagram of a delay time and a pre-beamforming weight vector estimator of multipath signals according to the present invention;

7 is a block diagram of a complex cyclic matched filter correlator according to the present invention

8 (a) is a block diagram of a signal power combiner according to the present invention

(b) is a block diagram of a signal amplitude combiner according to the present invention

9 is a block diagram of a multipath beamforming weight vector estimator according to the present invention;

* Description of the symbols for the main parts of the drawings *

10a, 10b: synchronizer 20: pre-beam forming unit

21, 31: cyclic fix remover 22: beam former

23: serial / parallel conversion unit 30: estimation unit

32: complex cyclic matching filter correlator 32a: symbol pattern generation unit

32b: cyclic shift register section 32c: block data buffer section

32d: cross correlator 33: signal combiner

34: beamforming weight vector estimator 34a: beamforming weight generator

34b: beamforming weight vector generator 34c: beamforming weight generator

35: delay time estimator 40: delay time compensation unit

50a, 50b: Adaptive equalizer 51, 53: FFT part

52: IFFT unit 54: weight update unit

55: vector generator 56: difference

57: equalizing unit 60: determination unit

A feature of an adaptive equalizer in a frequency domain using an antenna array according to the present invention for achieving the above object is to perform frame synchronization, timing synchronization, frequency synchronization, etc. of a signal using signals received through a plurality of antenna arrays. The synchronization unit and the signal output from the synchronization unit estimate a delay time for a multipath using characteristics of a pilot signal during a training or learning period, and calculate a pre-beam shaping weight vector based on the estimated delay time information. An estimator for estimating and a prebeam forming unit for forming an independent beam for each multipath signal in order to combine signal components for the same path by applying the pre-beam shaping weight vector output from the estimator to the signal output from the synchronizer And a delayed time of the signal for each multipath output through the pre-beam forming unit by the estimated delay time. And a delay compensator for compensating the liver to match the received time of each multipath signal, and an adaptive equalizer for equalizing the signal output from the delay time compensator in a frequency domain.

An operation according to the characteristics of the present invention is a smart antenna system for a broadband wireless communication system having a structure employing a single carrier transmission scheme having a cyclic prefix structure and a scheme employing an OFDM transmission scheme in a multipath and co-channel interference environment. As a result, it has a structure in which pre-beam formation and adaptive equalizer in frequency domain are combined, thereby effectively eliminating co-channel interference and optimal combining of multipath signals.

Other objects, features and advantages of the present invention will become apparent from the following detailed description of embodiments taken in conjunction with the accompanying drawings.

A preferred embodiment of a frequency domain adaptive equalizer using an antenna array according to the present invention will be described with reference to the accompanying drawings.

1 and 3 are overall configuration diagrams of a smart antenna for a broadband wireless communication system, FIG. 1 is a system configuration diagram of a single carrier transmission method having a cyclic prefix frame structure, and FIG. 3 is a signal transmission through a plurality of subcarriers. A system configuration diagram of an Orthogonal Frequency Division Multiplexing (OFDM) transmission scheme is shown.

1 and 3, the synchronization units 10a and 10b for synchronizing the signals received by the plurality of antennas and outputting the sampled signals corresponding to the respective antennas, and the synchronization units 10a and An estimator 30 estimating a delay time for the multipath using the signal output from 10b), and an estimator 30 estimating a pre-beam shaping weight vector based on the estimated delay time information, and an output from the estimator 30. A pre-beam forming unit 20 for forming an independent beam for each multi-path signal to apply the pre-beam forming weight vector to the signals output from the synchronization units 10a and 10b to combine signal components for the same path, and A delay time compensator 40 for compensating a delayed time of a signal for each multipath output through the pre-beam forming unit 20 by an estimated delay time and matching the received time of each multipath signal with the delay time; Bo Adaptive equalizers 50a and 50b for equalizing the signals output from the unit 40 in the frequency domain, and a determination unit for compensating a plurality of time domain signals output from the adaptive equalizers 50a and 50b. It consists of 60.

2 and 4 are diagrams showing the configuration of a frequency domain adaptive equalizer for equalizing a signal having a pre-beam formed in each multipath signal in a frequency domain, and FIG. 2 is a configuration diagram of an adaptive equalizer of a single carrier transmission method. 4 is a block diagram of an adaptive equalizer of the OFDM transmission scheme.

2, a plurality of first FFT units 51 converting a plurality of multipath time-domain signals output from the delay time compensator 40 into a plurality of frequency-domain signals;

An equalizer 57 for compensating for distortion of a plurality of frequency domain signals output from the first FFT unit 51 and a plurality of frequency domain signals output from the equalizer 57 as a plurality of time domain signals; The IFFT unit 52 converts the difference signal into a plurality of frequency domain signals using a signal obtained by determining the time domain signal and the training signal output from the IFFT unit 52 and the output time domain signal. A weight updating unit 54 combining the second FFT unit 53 and the respective frequency domain signals output from the second FFT unit 53 and the first FFT unit 51 to form an adaptive beam in the frequency domain. It consists of.

4, an FFT unit 51 converting a plurality of multipath-specific time domain signals output from the delay time compensator 40 into a plurality of frequency domain signals, and a plurality of outputs from the FFT unit 51. An equalizer 57 for compensating for distortion of the frequency domain signal, a vector generator 55 for making a plurality of frequency domain signals output from the equalizer 57 into one vector value, and the vector generator 55 Obtaining a difference signal using a signal outputted from the signal and a training signal or a signal determined from the output signal, and combining the difference signal with each of the frequency domain signals output from the FFT unit 51 to form an adaptive beam. And a weight updater 54.

FIG. 5 is a diagram illustrating a beamforming bank configured to combine a pre-beamforming weight vector and components of the same path incident on an antenna array to form an independent beam for each multipath.

Referring to FIG. 5, a plurality of cyclic prefix cancellers 21 for removing a prefix from a plurality of sample signals output from the synchronization units 10a and 10b, and a signal output from each of the cyclic prefix removers 21 and A plurality of beam formers 22 for combining the signals weighted by the pre-beam shaping weight vector output from the estimator 30 and combining the signals for the same path signal, and the plurality of beam formers 22 It is composed of a plurality of serial / parallel converter 23 for converting the output signal of the same path in parallel.

FIG. 6 illustrates a delay time estimator for each multipath signal using signal characteristics of a learning or training signal incident through an antenna array during an initial learning or training interval for pre-beam formation and compensation for delay time for each multipath signal. The delay time estimation for each multipath signal and the pre-beam shaping weight vector estimator are used to estimate the pre-beam shaping weight vector.

6, the estimator 30 includes a plurality of cyclic prefix removers 31 for removing a prefix from a plurality of sample signals output from the synchronizers 10a and 10b, and the plurality of cyclic prefix removers. A plurality of complex cyclic matched filter correlators 32 which cross-correlate each signal output from (31) with a reference signal, and a signal combining the signals output from the complex cyclic matched filter correlator 32 into one signal. A combiner 33, a delay time estimator 35 for estimating a delay time for each multipath from the signal output from the signal combiner 33, and a signal output from the plurality of cyclic prefix cancellers 31 And a beamforming weight vector estimator 34 estimating a pre-beamforming weight vector using the delay time information estimated by the delay time estimator 35.

7 is a block diagram of a complex cyclic matched filter correlator for obtaining a cross correlation with an incident learning or training signal.

Referring to FIG. 7, a symbol pattern generator 32a outputs a reference signal, a cyclic shift register unit 32b for cyclically moving a reference signal output from the symbol pattern generator 32a, and the cyclic shift. And a cross correlation unit 32d for cross-correlating a reference signal outputted while cyclically moving through the register 32b and a block data signal outputted through the cyclic prefix remover 31 stored in the block data buffer 32c. .

The symbol pattern generator 32a output value is a value obtained by conjugating the PN signal sequence in the case of a single carrier transmission system having a cyclic prefix frame structure as shown in FIG. 1, and an OFDM transmission scheme as shown in FIG. In this case, the complex signal of the signal transformed into BPSK, QPSK, M-ary PSK, etc. having a uniform amplitude in the frequency domain is converted to the time domain through IDFT.

FIG. 8 is a signal combiner combining the cross correlation values independently obtained through the complex matched filter correlator 32, as shown in FIG. 6, in order to use the delay time of the multipath signal as the input of the estimator. ), There is a signal power combiner 33a for power combining and a signal amplitude combiner 33b for amplitude combining as shown in FIG.

9 is a diagram illustrating a pre-beamforming weight vector estimator for each multipath signal.

Referring to FIG. 9, a symbol pattern generator 32a outputs a reference signal, and a cyclic shift register unit cyclically shifts the signal output from the symbol pattern generator 32a by an estimated delay time for each multipath signal. (32b) and a plurality of beamforming weights to obtain the weighted sum of the output signals of the cyclic shift register unit 32b with the input signals from which the cyclic prefix has been removed for each antenna element to obtain the prebeam shaping weights for the multipath signals. A generator 34c and a plurality of vector generators 34b configured to obtain output beams of the plurality of beamforming weight generators 34c into one vector to obtain a prebeaming weighting vector for each multipath signal.

The signal generated by the symbol pattern generator 32a of the beamforming weight vector estimator 34 configured as described above is cyclically moved through the cyclic shift register part 32b by the estimated delay time for the multipath signal, and the cyclically shifted The signal is weighted with the input signals from which the cyclic prefix is removed for each antenna.

Here, the obtained value obtains a pre-beam shaping weight vector for each multipath signal through a vector former.

The operation of the adaptive equalizer in the frequency domain using the antenna array according to the present invention configured as described above will be described in detail with reference to the accompanying drawings.

First, signals input through a plurality of antennas are processed in block units, and signals incident through the antenna array are synchronized through the synchronization units 10a and 10b.

This results in a sampled input signal.

Subsequently, as shown in FIG. 6, the cyclic prefix is removed through the complex cyclic matched filter correlator 32 of each antenna while cyclically shifting the signal output from the symbol pattern generator 32a as shown in FIG. 7. Measure the correlation with the block.

As such, the output values are combined by the signal power combiner 33a as shown in FIG. 8 (a) through the signal combiner 33, or by the signal amplitude combiner 33b as shown in FIG. 8 (b). Make amplitude coupling.

The signal coupled through the signal combiner 33 estimates the delay time of the multipath signal from the degree of circular shift of the reference signal sequence, where the multipath delay time estimator 35 is above a certain threshold correlation value.

Using this, the pre-beamforming weight vector is estimated by multiplying the input signal with the cyclically used reference signal sequence.

In order to form an independent beam for each of the multipath signals from the above information, the beamforming weight vector and the same path-specific components incident on the antenna array are combined through the beamforming bank as shown in FIG. 5.

After compensation by the delay time for each multipath signal estimated by the delay time compensator 40, and enters the input of the adaptive equalizer (50a, 50b).

The input signals of the adaptive equalizers 50a and 50b are converted into signals in the frequency domain through the FFT unit 51 as shown in FIGS. 2 and 4.

The converted signal is hunted down to a desired mean square error (MSE) through an adaptive algorithm for each frequency component.

That is, during the initial learning or training period, the adaptive equalizers 50a and 50b are adapted to reduce the MSE with the known reference signal.

After the learning or training period has passed, the output signal is determined to use the signal as a reference signal, and the adaptive equalizers 50a and 50b are continuously adapted.

First embodiment

1 and 2 show an overall system configuration of a single carrier transmission method using a cyclic prefix frame structure.

In the multipath and co-channel interference environment, the signal incident through the antenna array performs frame synchronization, timing synchronization, frequency synchronization, etc. through the synchronization unit 10a, and inputs the signal through the sampling and A / D converters for each antenna. Get

As shown in FIGS. 5 and 6, the sampled input signal is subjected to a signal processing process by removing the cyclic prefix through the cyclic prefix remover 21 and 31.

Input signal vector sampled through any a-th antenna Signal Vectors with and Cyclic Prefixes Removed Is expressed by Equation 1 below.

, a = 1, , A

In Equation 1, m represents the m-th transmitted block, N represents the size of one block.

First, delay time for each multipath signal is estimated. As shown in FIG. 7, delay time of the multipath signal is estimated.

Referring to FIG. 7, a complex that obtains a cross correlation with a sampled input signal while cyclically moving an output value obtained by conjugating the reference PN code signal sequence output from the symbol pattern generator 32a through the circular shift register unit 32b. A cyclic matched filter correlator 32 is shown.

The cross correlation output value of the complex cyclic matched filter correlator 32 is combined through the signal combiner 33 as shown in FIG.

The signal combiner 33 performs power combining by the signal power combiner 33a as shown in FIG. 8 (a) or amplitude combining by the signal amplitude combiner 33b as shown in FIG. 8 (b).

,

In Equation 2, d denotes a reference PN code signal, and * denotes a conjugate complex.

The delay time of the multipath signal is determined by the correlation value R (τ) When exceeding, by selecting the degree of cyclic shift of the cyclic shift register 32b, the delay times of the multipath signal can be expressed as in Equation (3).

,

L in Equation 3 is the number of selected multipaths silver The first quarter.

And said The number of branches of the next adaptive equalizer 50a is determined according to the number.

Next, the estimated delay time for each multipath signal Using Equation 3, the pre-beam shaping weight vector for each multipath signal for beamforming is estimated.

As shown in FIG. 9, a signal obtained by conjugate conjugate of the reference PN code signal sequence is generated through the symbol pattern generation unit 32a, and the cyclic shift register unit 32b of the beamforming weight vector unit 34a for each multipath signal is generated. It is read as

The cyclic shift register section 32b of the first branch delays the multipath signal. Cyclically shift the output and take the weighted sum of the input signals sampled at each antenna.

The weight values thus obtained generate a pre-beam forming weight vector via the vector former 34b.

The estimated pre-beamforming weight vector is expressed as in Equation 4.

In Equation 4, silver The pre-beam shaping weight vector for the second multipath signal is normalized to the input signal.

Beamforming weight vector determined in this way As shown in FIG. 5, the signal components of the same paths incident on the antenna array are combined.

And the signal weighted by the pre-beam formation is expressed as in Equation 5.

H stands for conjugate transpose.

The weighted output signal obtained by Equation 5 is estimated for each branch before entering the adaptive equalizer 50a. Compensating the delay time by cyclically shifting as much as possible, and enters the input of the adaptive equalizer 50a.

Therefore, the final input of the adaptive equalizer 50a is expressed by Equation 6 below.

Referring to FIGS. 1 and 2, the signal obtained through the pre-beam formation is converted into the frequency domain through the first FFT unit 51 for each branch, and for each frequency component, the signal of each branch is converted into the following equation through an adaptive algorithm. As in 7, the maximum ratio combining (MRC) is performed.

Any m-th received block is an equalizer weight vector obtained through an adaptive algorithm for each independent frequency component of the frequency domain adaptive equalizer 57. Take a weighted sum with.

Therefore, the weighted sum output of the n th frequency component is expressed by Equation 8 below.

H stands for conjugate transpose.

Then, the IFFT unit 52 converts the signal into a time domain signal.

Therefore, the final adaptive equalizer 50a output is Decision value through the decision unit 60 Get into.

In addition, the signal input to the adaptive equalizer 50a is weighted by each frequency component so that the MN of the PN code signal sequence and the equalizer output signal, which are the reference during the learning or training interval, is minimized by the adaptive algorithm. Adapt it.

And the determination value after the learning or training interval. Replace with the reference signal to continue to implement the adaptive algorithm.

As the adaptive algorithm, a power normalized least mean square (NLMS) algorithm for each frequency component is used.

The output error is expressed by the following equation (10), which is converted into the frequency domain through the second FFT unit 53 for use in the frequency domain.

e (m) = d (m)-y (m)

E (m) =

Equation 11 is an equalizer weight update equation based on NLMS (Normalized Least Mean Square).

In Equation 11 is an adaptation constant, Denotes the mth updated average input power of the equalizer nth frequency components. || · || represents a vector 2-norm operator, and λ represents a forgetting factor.

As such, the NLMS algorithm normalizes the value to update the equalizer weights with the input average power estimated for each frequency component.

In addition, the equalizer input average power is cyclically calculated using the lambda. The method of Equation 11 is more useful when the power change range of the input signal is severe.

Second embodiment

3 and 4 are configuration diagrams of a smart antenna system for a broadband wireless communication system employing an OFDM transmission scheme having a cyclic prefix.

The operation of the system shown in FIGS. 3 and 4 is almost the same as in the first embodiment.

Only OFDM signals transmit information through several subcarriers, and modulation schemes include binary phase shift keying (BPSK), quadrature PSK (QPSK), and M-ary PSK (MPS) with uniform amplitude in the frequency domain. Uses a modulation scheme of

The modulated signal is transformed into the time domain through an Inverse Discrecte Fourier Transform (IDFT), and is transmitted by adding a cyclic prefix.

The delay time estimation of the multipath signals is found using autocorrelation characteristics of an orthogonal frequency division multiplexing (OFDM) signal having a cyclic prefix.

Signal output from the symbol pattern generator 32a as shown in FIG. Is a value obtained by converting the reference signal D through the IDFT and taking the conjugate complex.

Therefore, a complex cyclic matched filter correlator device which obtains cross correlation with a sampled input signal while cyclically moving a signal output from the symbol pattern generator 32a through a cyclic shift register 32a. It is expressed as

If the autocorrelation function is defined as in Equation 12 below, the autocorrelation property in the time domain is the one having the same property as in Equation 13.

In Equation 13, a modulation scheme of binary phase shift keying (BPSK), quadrature PSK (QPSK), and M-ary PSK (M-ary PSK) having a uniform amplitude in the frequency domain is used. Will take the value of.

3 and 4, the output Y (m) of the signal adaptively applied by the frequency domain adaptive equalizer 57 is a desired output, and is determined by the decision unit 60. Get

Each signal thus output is obtained through the vector generator 55 to obtain a sequential data output value.

Therefore, an equation for obtaining an error as in Equation 10 is expressed as in Equation 14.

E (m) = D (m)-Y (m)

E (m) =

In addition, the equalizer weight update equation based on NLMS (Normalized Least Mean Square) may be expressed as Equation 11 expressed in the first embodiment.

As such, the NLMS algorithm normalizes the value to update the equalizer weights with the input average power estimated for each frequency component.

In addition, the equalizer input average power is cyclically calculated using the lambda, and a method such as Equation 11 is more useful when the power change range of the input signal is severe.

As described above, the adaptive equalizer in the frequency domain using the antenna array according to the present invention has the following effects.

First, in the presence of co-channel interference signal (CCI), by performing pre-beam shaping and delay time compensation for each of the multipath signals incident through the antenna array, a desired signal is obtained through an adaptive equalizer in the frequency domain. It can be extracted more efficiently.

Secondly, it is simple to estimate the incidence direction of the multipath signal and to implement the adaptive algorithm, which is effective for real-time implementation.

Those skilled in the art will appreciate that various changes and modifications can be made without departing from the spirit of the present invention.

Therefore, the technical scope of the present invention should not be limited to the contents described in the embodiments, but should be defined by the claims.

Claims (9)

A synchronizer for performing synchronization using signals received through a plurality of antenna arrays; The signal output from the synchronization unit estimates a delay time for a multipath using characteristics of a pilot signal during a training or learning period, and uses a pre-beam shaping weight vector using the estimated delay time information of the multipath signal. An estimator for estimating, A pre-beamformer for forming an independent beam for each multipath signal by applying the pre-beamforming weight vector output from the estimator to the signal output from the synchronizer to combine signal components for the same path; A delay time compensator for compensating for the delayed time of the multipath signal output through the pre-beam forming unit by the estimated delay time to match the received time of each multipath signal; The pre-combined signals for each path whose time synchronization is output by the delay time compensator are converted into a frequency domain and include an adaptive equalizer for equalizing and combining the same frequency components of the signals for each path. Smart antenna system. The method of claim 1, The adaptive equalizer in the single carrier transmission method having a cyclic prefix includes a first FFT unit for converting pre-combined signals for each path that are time-synchronized by the delay time compensator into a frequency domain signal; An equalizer unit for equalizing and combining the same frequency components of the signal converted into the frequency domain for each path in the first FFT unit; An IFFT unit for converting a plurality of output signals equalized and coupled by the equalizer unit into time-domain signals; A divider for obtaining a difference signal by using the time domain signal and the training signal output from the IFFT unit, or a signal obtained by determining the output time domain signal; A second FFT unit converting the difference signal output from the difference unit into a frequency domain signal; And a weight updater configured to form an adaptive beam in the frequency domain by using the respective frequency domain signals output from the second FFT unit and the first FFT unit. The method of claim 1, The adaptive equalizer in the OFDM transmission scheme is an FFT unit for converting pre-combined signals for each path that are time-synchronized by the delay time compensator into a frequency domain signal; An equalizer unit for equalizing and combining the same frequency components of the signal converted into the frequency domain for each path in the FFT unit; A vector generator for tying a plurality of output signals equalized by the equalizer in one vector; A difference unit for obtaining a difference signal using a signal output from the vector generator and a signal determined from the training signal or the output time-domain signal; And a weight updater configured to form an adaptive beam in the frequency domain by using the difference signal output from the difference and the respective frequency domain signal output from the FFT unit. The method of claim 1, The estimator includes a plurality of cyclic prefix cancellers for removing a prefix from a plurality of sample signals output from the synchronizer; A plurality of complex cyclic matched filter correlators for correlating each signal output from the plurality of cyclic prefix removers with a reference signal; A signal combiner for combining the signal output from the complex cyclic matched filter correlator into one signal; A delay time estimator for estimating a delay time for each multipath signal using a value of a signal output from the signal combiner; And a beamforming weight vector estimator for estimating a pre-beamforming weight vector from a signal output from the cyclic prefix using the delay time information estimated by the delay estimator. The method of claim 1, The pre-beam forming unit includes a plurality of cyclic prefix cancellers for removing cyclic prefixes from a plurality of sample signals output from the synchronizer; A plurality of beamformers for forming independent beams for each multipath signal by applying the pre-beamforming weight vector output from the estimator to the signals output from the cyclic prefix remover to combine signal components for the same path; Smart antenna system comprising a plurality of serial / parallel converter for converting each signal output from the plurality of beam former in parallel. The method of claim 4, wherein The complex cyclic matched filter correlator comprises: a symbol pattern generator for outputting a reference signal; A cyclic shift register unit which cyclically shifts and outputs a plurality of signals output from the symbol pattern generator; And a cross correlation unit configured to obtain a cross correlation between a reference signal output while being cyclically moved through the cyclic shift register unit and a block data signal output through the cyclic prefix remover stored in the block data buffer. system. The method of claim 4, wherein And the signal combiner uses either a signal power combiner or a signal amplitude combiner for power or amplitude combining the outputs of the plurality of complex cyclic filter correlators for use as a delay time estimator input signal. The method of claim 4, wherein The beamforming weight vector estimator includes a symbol pattern generator for outputting a reference signal; A cyclic shift register unit cyclically shifting the signal output from the symbol pattern generator by a delay time for each multipath signal estimated by the delay time estimator; A plurality of beamforming weight generators for obtaining a pre-beamforming weight for each multipath signal by obtaining a weighted sum of the output signals of the cyclic shift register unit with the input signals from which the cyclic prefix is removed through the cyclic prefix remover for each antenna element. Wow, A beamforming weight vector generator for tying output signals of the plurality of beamforming weight generators into one vector; And a bank structure of a beamforming weight generator including a plurality of beamforming weight generators and beamforming weight vector generators for obtaining a pre-beamforming weight of a multipath signal. The method of claim 1, And an determining unit for compensating a plurality of time-domain signals outputted from the adaptive equalizing unit.