KR20050045150A - Digital broadcasting receiver and coefficient initialization method of noise canceller - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital broadcast receiver that removes amplified noise during equalization by using a noise canceller at the output of the equalizer and a method of initializing coefficients of the noise canceller, and more particularly to using a noise canceller in a digital broadcast receiver having an equalizer. In this case, by setting an initial coefficient of the filter in the noise canceller using the channel impulse response (CIR) obtained from the channel estimator, the noise canceller may converge faster after the operation of the noise canceller. Therefore, the present invention can improve the initial convergence speed of the overall system by improving the initial convergence speed of the noise canceller.

Description

Digital broadcasting receiver and coefficient initialization method of noise canceller

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital broadcast receiver, and more particularly, to a method of initializing coefficients of a noise canceller of a digital broadcast receiver that removes amplified noise during equalization by using a noise canceller at an output of the equalizer.

When the ATSC 8VSB transmission system and PAM or QAM transmission system, which are proposed for most digital transmission systems and digital TV (DTV) transmission methods currently used, are transmitted by air or wire, the transmitted signals are transmitted to various reflectors. The reflected signals are summed and received. The reflected components distort the original signal, so the original signal cannot be obtained by the received signal alone.

As such, an equalizer is used to compensate for ghost or fading that distorts the transmission signal between the transmitter and the receiver. The equalizer can be divided into a time-domain equalizer and a frequency-domain equalizer, all of which remove components that distort the original signal.

1 is a schematic diagram of a digital broadcast receiver using the equalizer, i.e., a VSB receiving system, wherein a demodulator 101 is a carrier / timing of a VSB demodulation and digitization and a digitized signal for a received RF signal. The restoration is performed to convert the received RF signal into a base-band signal and output the same to the equalizer 102. The equalizer 102 compensates for the various fadings that occur while the signal arrives from the transmitter to the receiver.

At this time, if the distortion of the signal is compensated using the equalizer, the signal itself is compensated with the desired signal, but the noise is amplified according to the coefficient of the equalizer. At this time, if the amplified noise is compensated using a noise canceller, both the distortion of the signal and the amplification of the noise can be compensated.

That is, converting the impulse response of an ideal channel without fading into a frequency domain signal has a size of 1, while converting the impulse response of a fading channel into a frequency domain signal has a very low magnitude as shown in FIG. (spectrum null) occurs.

Figure 2 shows the frequency spectrum of a faded channel and received signal in a VSB receiving system. Figure 2a is a channel impulse response with a main signal and three fading signals with time delay. Figure 2b shows the frequency spectrum of the ideal VSB signal sent by the transmitter and the frequency spectrum when fading is present. And the white noise AWGN is added.

In this case, when the received signal having the fading is compensated using a frequency domain equalizer, as shown in FIG. 3, the channel is compensated and the size of the white noise component is increased at the lower portion of the spectrum of the channel impulse response. Noise enhancement occurs. Because of this phenomenon, even after frequency domain equalization, the signal to noise ratio (SNR) of the output signal is bad, which ultimately degrades the performance of the system.

Figure 3 shows the spectral change of the signal and noise when the fading component is compensated using a frequency domain equalizer. Figure 3a is a spectrum of the signal and the noise before the frequency domain equalization, it can be seen that the AWGN is added to the entire band with a certain size, the fading received signal can be seen that the magnitude of the spectrum is very small at a specific frequency. Figure 3b is a frequency spectrum of a signal that has undergone frequency domain equalization, where the signal spectrum coincides with the ideal signal, but the noise in the portion where the magnitude of the signal spectrum is small before compensation is increased (noise enhancement). The spectrum of the signal that has undergone the frequency domain equalization shows that fading is compensated, but performance is degraded due to the increase in noise of the entire signal due to noise enhancement in a specific frequency section.

Use a noise canceller to compensate for this. By using the noise canceller to remove the amplified noise as shown in FIG. 4, only pure white noise from which colored noise components are removed is output. The output of the noise canceller thus becomes a signal with both colored noise due to fading and noise amplification.

4 is a frequency spectrum when a noise canceller is used for an equalized signal. 4A is identical to the spectrum after equalization of FIG. 3 described above and shows that there is noise enhancement. In contrast, FIG. 4B shows that when the noise canceller is performed, the noise reduction component (colored noise) is removed to reduce the overall noise level.

The noise canceller is typically implemented using an LMS filter. In the related art, an initial coefficient is arbitrarily set (for example, 0) when the filter is initialized. Therefore, there is a problem that the noise canceller takes time to converge. That is, a problem occurs in that the initial convergence speed of the noise canceller decreases.

The present invention is to solve the above problems, an object of the present invention is to set the initial coefficient of the noise canceller to remove the amplified noise at the time of equalization by using a channel impulse response (CIR), The present invention provides a method of initializing coefficients of a digital broadcast receiver and a noise canceller to improve the initial convergence speed of the canceller.

According to an aspect of the present invention, there is provided a digital broadcast receiver comprising: a channel estimator for estimating and outputting an impulse response of a transmission channel from a received signal passing through a transmission channel; An equalizer for compensating for channel distortion of the received signal passing through the transmission channel; And a noise canceller for initializing coefficients by using a channel impulse response output from the channel estimator and predicting noise from an output signal of the equalizer to remove noise included in the equalized signal. It is done.

The noise canceller may include a first subtractor for outputting reference noise by subtracting a determination value from an output signal of the equalizer, a slicer configured to receive a signal from which the noise is removed, and to output a determination result to the first subtractor; A second subtractor for detecting an error by subtracting the predicted noise from the reference noise, an initial coefficient calculating unit for calculating an initial coefficient from a channel impulse response output from the channel estimating unit, and a noise predicted from the output signal of the equalizer; After subtracting a third subtractor and an initial coefficient from the initial coefficient calculator, initializing tap coefficients, the noise is estimated using the reference noise output from the first subtractor and the error output from the second subtractor. And a filter output to the second and third subtractors.

The filter is connected to the N connected tap delays serially delaying the reference noise output from the first subtractor, the input terminal of the first tap delay unit and the output terminal of each tap delay unit. N + 1 coefficient updating units for initializing coefficients with an output of an initial coefficient calculating unit and then performing coefficient updating by using the reference noise, the output of each tap delayer that delayed the reference noise, and an error value, and the N An adder that adds all of the outputs of the +1 coefficient updater, a multiplier that multiplies the error output from the second subtractor with the step size, and selects the output of the initial coefficient calculator in initializing coefficients, and otherwise outputs the output of the multiplier And selectors for selecting and outputting the N + 1 coefficient update units, respectively.

The initial coefficient calculating unit includes an adder corresponding to each coefficient updating unit of the filter, and an adder for obtaining an initial coefficient of the nth coefficient updating unit, wherein n is a natural number, is a channel of the main signal output from the channel estimating unit. The channel impulse response of the previous n th fading signal and the channel impulse response of the subsequent n th fading signal are added based on the impulse response.

The adder for obtaining the initial coefficient of the nth coefficient updater, based on the channel impulse response of the main signal output from the channel estimator, if there is no channel impulse response of the previous nth fading signal, then adds' 0 to the channel impulse response of the subsequent nth fading signal. It is characterized by adding '.

The adder for calculating the initial coefficient of the n th coefficient updater, based on the channel impulse response of the main signal output from the channel estimator, if there is no channel impulse response of the n th fading signal, then adds' 0 to the channel impulse response of the previous n th fading signal. It is characterized by adding '.

The filter coefficient initialization method of the noise canceller according to the present invention includes estimating and outputting an impulse response of a transmission channel from a received signal passing through the transmission channel; And initializing the filter coefficients using the channel impulse response output in the step, and predicting noise from the equalized signal to remove noise included in the equalized signal.

In the above step, the nth coefficient of the filter may be obtained by adding the channel impulse response of the previous nth fading signal and the channel impulse response of the nth fading signal based on the channel impulse response of the main signal. .

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.

Hereinafter, with reference to the accompanying drawings illustrating the configuration and operation of the embodiment of the present invention, the configuration and operation of the present invention shown in the drawings and described by it will be described as at least one embodiment, By the technical spirit of the present invention described above and its core configuration and operation is not limited.

5 is a schematic diagram of a VSB receiving system for initializing coefficients of a noise canceller according to the present invention, which estimates a path (impulse response) of a transmission channel from an output of a demodulator 501, an equalizer 502, and the demodulator 501. Noise calculated to estimate and remove amplified noise during equalization in the equalizer 502 after calculating and setting an initial coefficient from a channel estimator 503 and a channel impulse response (CIR) estimated by the channel estimator 503 A remover 504. The equalizer 504 uses a frequency domain equalizer as an embodiment. In this case, the equalizer 504 performs equalization in the frequency domain and then converts the equalization result back to the time domain and outputs the equalized result.

In FIG. 5 configured as described above, the demodulator 501 performs baseband-based reception on the received RF signal by performing VSB demodulation and digitization on the received RF signal, and performing carrier / timing recovery from the digitized signal. band) and then output to the equalizer 502 and the channel estimator 503.

The equalizer 502 compensates for various fading included in the baseband signal and outputs the same to the noise canceller 504.

The channel estimator 503 receives the output of the demodulator 501 as an input, estimates an impulse response (CIR) of a discrete equivalent channel that the original signal may have passed, and outputs it to the noise canceller 504.

In this case, assuming that the equalization is completely performed in the channel equalizer 502, the signal output from the equalizer 502 to the noise removing unit 504 is the sum of the original signal and the colored noise. It can be seen that. Therefore, the noise canceller 504 initializes the filter coefficients by using the CIR of the channel estimator 503 and estimates the colored noise from the output by the equalizer 502 to subtract the noise amplified during equalization. Remove

FIG. 6 is a detailed block diagram of the noise canceller 504. The first subtractor 603 outputting reference noise by subtracting a determination value from an output signal of the equalizer 502 and receiving a signal from which the noise is removed is determined. a VSB slicer 606 for determining and outputting a decision result to the first subtractor 603, a second subtractor 607 for detecting an error by subtracting a predicted noise from the reference noise, and the channel estimator 503 Initial coefficient calculation unit 609 that calculates an initial coefficient from a channel impulse response (CIR) value output from the N- s, and removes the predicted noise from the output signal of the equalizer 504 to remove the noise that is increased due to noise enhancement during equalization. And a filter 604 for predicting the noise and outputting the noise to the second and third subtractors 607. That is, the filter 605 receives and sets an initial coefficient from the initial coefficient calculating unit 609, and the reference noise output from the first subtractor 603 is used as an input signal of the tap delay unit. The error output from the subtractor 607 is used as an error input of the filter.

The filter 604 is a filter using a coefficient update method of a least mean square (LMS) method of a tapped-delay-line type. In the present invention, another type of filter capable of noise prediction may be used instead of the LMS type filter.

In addition, between the equalizer 502 and the first subtractor 603, between the VSB slice portion 606 and the first subtractor 603, and between the second subtractor 607 and the filter 604. Further delays 601, 602, 608 are provided for delaying the input signal by one clock for timing matching.

FIG. 7 illustrates the filter 604 of FIG. 6 in detail, wherein N series delay taps 7011 to 701N connected in series to delay the reference noise output from the first subtractor 603, respectively, and the first tap delay. It is connected to the input terminal of the device 7011 and the output terminal of each tap delayer (7011 ~ 701N), and when the filter is initialized, the coefficient is initialized by the output of the initial coefficient calculation unit 609, and then the reference noise and the N + 1 coefficient updating units 7021 to 702N + 1 and coefficient updating units 7021 to 702N + which perform coefficient updating using the output of each tap delayer that delayed the reference noise and the error value e (n). An adder 703 that adds all the outputs of 1), a multiplier 704 that multiplies the error e (n) by the step size μ, and an output of the initial coefficient calculating section 609 at the time of filter initialization, otherwise Select the output of the multiplier 704 and output it to a plurality of coefficient updater And a selection unit 705 for outputting.

The output of the adder 703 is output to the multiplier 704 via a delay 608 and a second subtractor 607. The selector 705 may use a multiplexer (ie, a mux) or a switching element.

Next, the initial coefficient setting and the noise removing process of the filter will be described in detail with reference to FIGS. 6 and 7.

That is, the signal output from the equalizer 502 includes both the original signal and the noise, which are output to the first subtractor 603 through the delay unit 601 and simultaneously to the third subtractor 605. Is output.

At the same time, the determination value determined by the VSB slice unit 606 is output to the first subtractor 603 through the delay unit 602. The first subtractor 603 subtracts a determination value input through the delayer 602 from an equalization signal input through the delayer 601 to obtain a reference noise x (n), and then calculates the first noise of the filter 604. Output to the tap retarder and the second subtractor 607. Here, the reference noise is noise to be compensated for in the noise canceller 604.

The VSB slicer 606 determines a value closest to the signal when a signal in which the amplified noise is removed from the third subtractor 605 of the noise canceller 604, that is, the noise is whitened, is input. This decision value is output to the first subtractor 603 through the delay unit 602. The VSB slicer 606 may use various types of algorithms for determining a VSB signal.

Meanwhile, the noise predicted by the filter 604 is output to the third subtractor 605 and to the second subtractor 607 through the delay unit 608. The second subtractor 607 calculates an error e (n) by subtracting the output of the delay unit 608 from the reference noise x (n) output from the first subtractor 603. The error value e (n) is input to the multiplier 704 of the filter 604, and the multiplier 704 multiplies the error value e (n) by the step size mu and outputs it to the selection unit 705. When the filter is initialized, the selector 705 selects the output of the initial coefficient calculator 609, and otherwise selects the output of the multiplier 704 and outputs the output to each coefficient updater.

The reference noise input to the filter 604 is delayed by one clock while passing through N number of tap delayers 7011 to 701N, and the output of the first subtractor 603 and the output of each tap delay are respectively corresponding to each other. Inputs are made to coefficient updating units 7021 to 702N + 1. The coefficients of the N + 1 coefficient update units 7021 to 702N + 1 are initialized to initial coefficient values calculated by the initial coefficient calculator 609 at the time of filter initialization, and then the reference noise and each tap delay unit. The coefficient is updated by the output of and the error value e (n).

Detailed operations of the initial coefficient calculator 609 will be described in detail later.

The coefficient update equation performed after the filter initialization is shown in Equation 1 below.

p (k, n + 1) = p (k, n) + u * e (n) * x (k, n)

Where n is 0, 1, 2, ..., and k is the k-th tap coefficient. u is an integer that determines the size of the value to be updated, referred to as the step size, and e (n) is an error value output from the second subtractor 607. And x (k, n) is the input value of the k th tap at n time.

As described above, the outputs of the coefficient update units 7020 to 702N, which have undergone coefficient update through the initial coefficient setting and coefficient update process, are added to the adder 703 and added to the delay unit 608 and the third subtractor 605. Is output. That is, the output of the adder 703 becomes the output of the filter 604, and eventually the predicted noise.

The third subtractor 605 removes the amplified noise included in the equalized signal by subtracting the noise predicted by the filter 604 from the equalized signal. That is, the amplified noise at the time of equalization can be removed by whitening the noise processing by subtracting the predicted noise from the equalized signal.

FIG. 8 illustrates an example of a method in which the initial coefficient calculator 609 obtains an initial coefficient of the noise canceller 504 by receiving a channel impulse response (CIR) from the channel estimator 503. The method for obtaining the initial coefficient of the noise canceller using the above-described method may be implemented in various ways in addition to the method of FIG. 8.

In general, the CIR estimated and output by the channel estimator 503 includes the CIR of the main signal and the CIR of the fading signal generated by a time delay before and after the CIR of the main signal. That is, the CIR in front of the CIR of the main signal is the CIR of the fading signal generated in time earlier than the main signal, and the CIR behind is the CIR of the fading signal generated in time delay from the main signal.

FIG. 8 shows a channel impulse response having a total of nine CIR coefficients, including the CIR coefficients of the central main signal. The CIR coefficient of the central main signal is c [0], based on c [0], the CIR coefficients c [01], c [-2], c [-3] with the previous three fading signals and the following five times. Suppose it consists of CIR coefficients c [1], c [2], c [3], c [4], c [5] with delayed fading signals.

In this case, the coefficients p [0] of the first tap of the filter 604 are c [-1] and c [, which are CIR coefficients of the first and subsequent fading signals based on the CIR coefficient c [0] of the central main signal. It is obtained by adding 1] (p [0] = c [-1] + c [1]).

The coefficient p [1] of the second tap of the filter 604 is the cIR coefficient of each second fading signal before and after the CIR coefficient c [0] of the central main signal, as with the first filter tap. 2] and c [2] are added (p [1] = c [-2] + c [2]). The subsequent filter taps are obtained in the same manner, and if there is no CIR coefficient of the corresponding fading signal before or after the CIR coefficient of the central main signal, 0 is added as in the coefficient initialization general formula of Equation 2 below.

CIR coefficients: c [n], n = -L, -L + 1, ..., 0,1, ..., M where L and M are natural numbers

Coefficient of noise canceller: p [l], l = p [0], p [1], ..., p [N] where N is an integer

P [k] = c [k] + c [-k], (1 ≤ k ≤ N)

At this time, if -n <-L, 0 is added instead of c [-n], and if n> M, 0 is added instead of c [n].

In the noise canceller coefficient initialization general formula of Equation 2, c [n] is a channel impulse response coefficient, n is an integer ranging from -L to M, and L and M are positive integers, that is, natural numbers. The coefficient of the noise canceller is p [l] and the range of l is a positive integer from 1 to N.

As described above, p [n] is obtained as the sum of the CIR coefficient c [−n] of the previous fading signal and the CIR coefficient c [n] of the subsequent fading signal. Where n is a positive integer between 1 and N, and if -n <-L, then there is no CIR coefficient, then 0 is added as the CIR coefficient value of the previous fading signal, and if n> M, the CIR coefficient of the fading signal is also Since there is no interval, 0 is added as a CIR coefficient value of a fading signal.

As described above, in the present invention, the coefficient of the noise canceller is initialized using the channel impulse response to improve the initial convergence speed of the noise canceller. Therefore, although one method is described as an example in the present invention, it is possible to obtain an initial coefficient of the noise canceller using the channel impulse response in various ways.

The channel impulse response used in the present invention may be obtained using not only the channel estimator but also another method of obtaining the channel impulse response.

As described above, according to the apparatus and method for initializing a coefficient of a noise canceller according to the present invention, when a noise canceller is used in a VSB receiver having an equalizer, the filter in the noise canceller is obtained by using a channel impulse response (CIR) obtained from a channel estimator. By setting an initial coefficient, the noise canceller can converge at a faster time after the noise canceller operation starts. That is, the present invention can improve the initial convergence speed of the entire system by improving the initial convergence speed of the noise canceller.

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.

1 is a general schematic diagram of a VSB receiving system using an equalizer

2A is a general diagram showing an example of a channel impulse response with a main signal and three time-delayed fading signals.

2b is a general diagram showing an example of a frequency spectrum of an ideal VSB signal sent from a transmitter and a frequency spectrum in the presence of fading

3A is a general diagram showing an example of a spectrum of a signal and noise before frequency domain equalization

3b is a general diagram illustrating an example of a frequency spectrum of a signal that has undergone frequency domain equalization

4A is a general diagram showing an example of a frequency spectrum when a noise canceller is not used after frequency domain equalization

4B is a general diagram showing an example of a frequency spectrum after performing both frequency domain equalization and noise cancellation

5 is a schematic diagram of a VSB receiving system for initializing coefficients of a noise canceller in accordance with the present invention.

6 is a detailed block diagram of the noise canceller of FIG.

7 is a detailed block diagram of the filter of FIG.

8 is a view illustrating a coefficient initialization process of the initial coefficient calculating unit of FIG. 7;

Explanation of symbols for main parts of the drawings

501: demodulator 502: equalizer

503: channel estimator 504: noise canceller

601,602,608: delayer 603,605,607: subtractor

604: LMS filter 606: VSB slice portion

609: initial coefficient calculation unit

Claims (13)

A channel estimator for estimating and outputting an impulse response of the transmission channel from the received signal passing through the transmission channel; An equalizer for compensating for channel distortion of the received signal passing through the transmission channel; And And a noise canceller for initializing coefficients using a channel impulse response output from the channel estimator and removing noise included in the equalized signal by estimating noise from an output signal of the equalizer. Digital broadcast receiver. The method of claim 1, wherein the noise canceller A first subtractor for outputting reference noise by subtracting a determination value from an output signal of the equalizer; A slice unit configured to receive a signal from which noise is removed and to output a determination result to the first subtractor; A second subtractor for detecting an error by subtracting a predicted noise from the reference noise; An initial coefficient calculator for calculating an initial coefficient from a channel impulse response output from the channel estimator; A third subtractor for subtracting predicted noise from an output signal of the equalizer, After receiving the initial coefficient from the initial coefficient calculator and initializing the tap coefficients, the noise is predicted using the reference noise output from the first subtractor and the error output from the second subtractor. And a filter to output. The method of claim 2, wherein the filter And a filter using a coefficient update method of an LMS type in the form of a tap-delay-line. The method of claim 2, wherein the filter N series delay taps connected in series to sequentially delay the reference noise output from the first subtractor; Each tap connected to the input terminal of the first tap delay unit and the output terminal of each tap delay unit, and initializes the coefficient to the output of the initial coefficient calculating unit when the coefficient is initialized, and then delays the reference noise and the reference noise. N + 1 coefficient updating units which perform coefficient updating by using the output of the delay and the error value, An adder for adding all of the outputs of the N + 1 coefficient update units; A multiplier for multiplying an error output from the second subtractor by a step size, And a selector which selects an output of the initial coefficient calculator and selects an output of the multiplier and outputs the N + 1 coefficient updater to each other. The method of claim 2, wherein the initial coefficient calculation unit A number of adders corresponding to each coefficient updating unit of the filter, An adder for calculating an initial coefficient of the n th coefficient update unit, wherein n is a natural number, is based on the channel impulse response of the main signal output from the channel estimator, and then the channel of the n th fading signal. A digital broadcast receiver, comprising adding an impulse response. 6. The adder of claim 5, wherein the adder for obtaining an initial coefficient of the nth coefficient updater And if there is no channel impulse response of the previous nth fading signal based on the channel impulse response of the main signal output from the channel estimator, '0' is added to the channel impulse response of the subsequent nth fading signal. 6. The adder of claim 5, wherein the adder for obtaining an initial coefficient of the nth coefficient updater And if there is no channel impulse response of the n-th fading signal based on the channel impulse response of the main signal output from the channel estimator, '0' is added to the channel impulse response of the previous n-th fading signal. The method of claim 4, wherein each coefficient update unit And multiplying the output of each tap delayer by the output of the multiplier after the initialization of the coefficients by the initial coefficient calculating unit, and updating the result of adding the previous coefficient to the multiplication result. A filter coefficient initialization method of a noise canceller, which is connected to an output of an equalizer for compensating for channel distortion of a received signal passing through a transmission channel and removes amplified noise during equalization. (a) estimating and outputting an impulse response of the transmission channel from the received signal passing through the transmission channel; And (b) initializing the filter coefficients using the channel impulse response output in step (a) and predicting noise from an output signal of the equalizer to remove noise included in the equalized signal. A coefficient initialization method of the noise canceller, characterized in that. The method of claim 9, wherein step (b) (b-1) subtracting a determination value from the output signal of the equalizer and outputting reference noise; (b-2) receiving and determining a signal from which noise is removed and outputting the determined value to the step (b-1); (b-3) detecting an error by subtracting a predicted noise from the reference noise; (b-4) calculating an initial coefficient from the channel impulse response; (b-5) removing the amplified noise during equalization by subtracting the predicted noise from the output signal of the equalizer; (b-6) After initializing the coefficients of the filter by receiving the initial coefficients in the step (b-4), predicting the amplified noise during equalization using the reference noise and the error output in the step and then the (b- 3), (b-5) step of outputting the coefficients of the noise canceller, characterized in that it comprises the step of outputting. The method of claim 10, In step (b-6), the nth coefficient of the filter is equal to the channel impulse response of the previous nth fading signal and the channel impulse response of the nth fading signal based on the channel impulse response of the main signal. Coefficient initialization method of the noise canceller, characterized in that The method of claim 11, If there is no channel impulse response of the previous nth fading signal based on the channel impulse response of the main signal, the nth coefficient of the filter is added to '0' to the channel impulse response of the subsequent nth fading signal based on the channel impulse response of the main signal. Coefficient initialization method of the noise canceller, characterized in that The method of claim 11, If there is no channel impulse response of the subsequent nth fading signal based on the channel impulse response of the main signal, the nth coefficient of the filter adds '0' to the channel impulse response of the previous n th fading signal based on the channel impulse response of the main signal. Coefficient initialization method of the noise canceller, characterized in that