KR100623064B1 - An adaptive equalization system and method consist of combined linear equalizer and non- linear equalizer - Google Patents

Abstract

The present invention is an adaptive equalization system and method that combines a linear equalizer and a nonlinear equalizer by combining a linear equalizer and a nonlinear equalizer to compensate for signal distortion caused by a channel through second-order correction of a signal. Linear equalizer, which firstly equalizes the input symbols and outputs a first equalized signal from which the first half of pre-ISI and post-ISI are removed, and a forward filter and a feedback filter from the first equalized signal in the linear equalizer. The present invention is characterized by comprising a DFE (Decision Feedback Equalizer), which is a nonlinear equalizer that outputs a second equalized signal from which residual ISI has been removed. The equalization of the signal is ensured through the correction of the signal through the second order through the DFE. It may be a it is possible to obtain an improved signal correction effect.

Linear Equalizer, Nonlinear Equalizer, DFE, Combined, Adaptive Equalizer

Description

Adaptive equalization system and method combined with linear equalizer and nonlinear equalizer {AN ADAPTIVE EQUALIZATION SYSTEM AND METHOD CONSIST OF COMBINED LINEAR EQUALIZER AND NON-LINEAR EQUALIZER}

1 is a block diagram of a general N-tap linear equalizer.

2 is a block diagram of a DFE which is a general nonlinear equalizer.

3 is a block diagram of an adaptive equalization system combining a linear equalizer and a nonlinear equalizer according to the present invention.

4 is a conceptual diagram of ISI separation removal of an equalization system in accordance with the present invention.

5 is a graph showing the channel adaptive convergence rate and the error level after convergence according to the use of a general linear equalizer.

6 is a graph showing channel adaptive convergence rate and error level after convergence according to general DFE usage.

7 is a graph showing channel adaptive convergence speed and error level after convergence of the adaptive equalization system according to the present invention.

<Explanation of symbols for the main parts of the drawings>

300: linear equalizer 302: FIR filter

303: filter coefficient update unit 400: DFE

402: Forward filter 403: Forward coefficient updating unit

405: feedback filter 411: feedback coefficient updating unit

TECHNICAL FIELD The present invention relates to an equalizer, and more particularly, to an adaptive equalizer combining a linear equalizer and a nonlinear equalizer that combines a linear equalizer and a nonlinear equalizer to compensate for signal distortion caused by a channel through correction of a signal over two orders of magnitude. .

In general, the biggest factor degrading the performance of a system in a high speed digital communication system is Inter-Symbol Interference (hereinafter referred to as ISI).

ISI is caused by the linear distortion of a channel, that is, multipath channel, non-ideal frequency response, group delay, etc. Equalizer is used to reduce the ISI generated by the channel. Is used.

Equalizer can be divided into Non-Linear, represented by Linear and Decision Feedback Equalizer (DFE), depending on the type of operation.Blind Equalization and Non-Non Equalization (Non) based on the use of training signals -Blind Equalization).

Both linear and nonlinear equalizers need to know the signal distortion behavior of a channel in order to correct the signal distortion by the channel.

Non-magnetic equalizers use training signals to perform this process, and adaptive algorithms are used to characterize the channel over time.

The adaptive algorithm is mainly used such as Least Mean Square (LMS), Recursive Least Square (RLS), Normalize LMS (NLMS), and the like.

The coefficient updating method of the LMS algorithm is shown in Equation 1.

Equation 1

는 갱신될 계수 벡터이고,

는 현재의 계수 벡터,

은 등화기에 입력되는 신호 벡터이며,

는 계수 갱신과정에서 사용되는 스텝-사이즈(Step-Size)이다.

Complex Conjugation을 나타내고,

는 채널에 의해 왜곡된 신호와 왜곡없는 신호 사이의 에러 값으로 수학식 2와 같이 갱신된다.

Is the coefficient vector to be updated,

Is the current coefficient vector,

Is the signal vector input to the equalizer,

Is a step-size used in the coefficient updating process.

Represents Complex Conjugation,

Is updated as Equation 2 with an error value between the signal distorted by the channel and the signal without distortion.

Equation 2

는 Hermitian Matrix를 나타낸다.

Represents the Hermitian Matrix.

In general, LMS adaptation algorithms have the advantages of simple implementation and easy calculation, but have the disadvantage of relatively low convergence speed and poor stability after convergence.

On the other hand, the RLS adaptive algorithm has good convergence speed and stability after convergence, but has a large amount of calculation.

The coefficient update of the RLS algorithm is shown in Equation 3.

Equation 3

은 이득 벡터를,

은 추정된 에러 값을 나타내며 수학식 4로 갱신 된다.

Is a gain vector,

Represents an estimated error value and is updated to Equation 4.

Equation 4

Adaptive algorithms such as LMS and RLS adjust the filter tap coefficients of the equalizer to eliminate the ISI generated by the channel. ISI is an interference phenomenon of symbols occurring between front and back symbols in consecutive symbols and may be classified into pre-ISI (hereinafter referred to as Pre-ISI) and post-ISI (hereinafter referred to as Post-ISI).

1 shows a general linear equalizer, which is an N-tap FIR filter 102 for removing ISI from an input symbol input through the error corrector 101 according to a tap coefficient obtained by the coefficient updater 103. An input symbol input through the error estimator 106 and the error corrector 101 which estimate an error from the storage 104, the signal equalized through the FIR filter 102 and the output of the training sequence generator 105. And a tap delay unit 107 for delaying a predetermined time, and an error measurer 108 for removing a DC offset caused by a circuit offset or a pilot together with the error corrector 101.

) 부가부(103c) 및 상기 추정치 부가부(103c)의 출력과 탭 저장부(104)에 저장되어 있던 탭을 가산하는 가산기(103d)를 포함하여 구성되어 있다. 여기서, 상기 추정치 부가부(103c)는 등화기에서 수신 신호를 추정해 갈 때 점점 그 추정치를 훈련신호와 가깝게 해가게 되는데, 상기 추정치 부가부(103c)는 그 추정의 속도, 즉 어느 정도의 속도로 추정을 할지를 결정하는 요소인

를 곱해주는 기능을 수행하게 되며,

가 크게되면 그 추정 속도는 빨라지게 되지만 그만큼 추정후의 오차가 커지게 되고, 작을 경우 추정 속도는 느려지지만 추정후의 오차는 작아지게 된다.The FIR filter 102 is configured to remove pre-ISI at the filter front end 102a and post-ISI at the filter back end 102b, and the coefficient updater 103 is a tap delay unit 107. Multiplier 103a multiplying the input symbol delayed by a predetermined time through the error estimator 106 and an error estimate output from the error estimator 106;

) An adder 103c and an adder 103d for adding the output of the estimated value adder 103c and the tab stored in the tap storage unit 104. Here, the estimator 103c estimates the received signal in the equalizer, and gradually makes the estimate closer to the training signal. The estimator 103c is a speed of the estimation, that is, a certain speed. Is a factor that determines whether

Will multiply by

The larger the value, the faster the estimated speed, but the larger the estimated error, the smaller the lowered the estimated speed, but the smaller the estimated error.

The general linear equalizer generates a training sequence for compensating for channel distortion in the training sequence generator 105 and outputs the training sequence to the error estimator 106, which is an output of the FIR filter 102 from the error estimator 106. An error estimate is generated by subtracting the channel equalized signal. This error estimate is transmitted to the coefficient updater 103.

Accordingly, the coefficient updater 103 updates the tap coefficients based on the error estimate value. Accordingly, the FIR filter 102 filters the input symbols by the tap coefficients output from the coefficient updater 103 to generate channel equalized output signals.

In other words, such a linear equalizer is designed to remove ISI with one filter.

2 shows a DFE, which is a general nonlinear equalizer.

This is largely an N-tap forward filter 202 for filtering input symbols by tap coefficients output from the forward coefficient updating unit 209, and a training sequence through the multiplexer 207 with tap coefficients output from the feedback coefficient updating unit 211. M-tap feedback filter (203) for filtering the sum of the outputs of the forward filter 202 and the feedback filter 203 and the adder 204 for generating a channel equalization output signal, the multiplexer (207) ) Is configured to select one of the output of the training sequence generator 205 for generating the training sequence and the output of the slicer 206 for quantizing the channel equalized output, and the channel equalization with the output of the multiplexer 207. And an error estimator 208 that subtracts the signal to estimate the error.

In addition, the forward coefficient updating unit 209 multiplies the input symbol delayed by the forward tap delay unit 212 with the error estimate output from the error estimator 208, and the multiplier 209a. The accumulator 209b accumulates a signal through the () and an adder 209d adds the signal stored in the accumulator 209b and the tap stored in the forward tap storage unit 210.

The feedback coefficient updater 211 includes a multiplier 211a for multiplying an error signal output from the error estimator 208 and an equalized output signal, an accumulator 211b for accumulating signals through the multiplier, and an accumulator 211b. The adder 211d adds a signal through the feedback tab and the feedback tap stored in the feedback tap storage unit 212.

When the DFE, which is a general nonlinear equalizer, receives a data frame having a large amplitude and phase distortion due to a channel change, the DFE performs a nonlinear decision feedback equalization operation to restore a signal compensated for the amplitude and phase distortion.

That is, the training sequence generator 205 generates a training sequence for compensating for channel distortion, and outputs it through the multiplexer 207, and the error estimator 208 outputs a channel equalized signal to the output signal of the multiplexer 207. Subtracted to generate an error estimate, the error estimate is input to the forward coefficient updater 209 and the feedback coefficient updater 211, respectively.

Accordingly, the forward coefficient updater 209 updates the tap coefficients based on the error estimate, and at the same time, the feedback coefficient updater 211 also updates the tap coefficients based on the error estimate.

Accordingly, the forward filter 202 filters the input symbols by the tap coefficients output from the forward coefficient updater 209, and the feedback filter 203 uses the multiplexer (TAP) with the tap coefficients output from the feedback filter updater 211. Filtering train 207).

Therefore, the summer 204 sums the signals through the forward filter 202 and the feedback filter 204 to generate a channel equalization output signal.

The DFE which is such a nonlinear equalizer removes pre-ISI using a forward filter 202 and post-ISI using a feedback filter 203.

A filter that implements a forward filter of the linear equalizer and the nonlinear equalizer generally uses a finite impulse response (FIR) filter so that the filter converges stably.

However, the nonlinear equalizer has the characteristics of the Infinite Impulse Response (IIR) filter due to the feedback filter, which causes a problem of inferior stability of the equalizer.

In general, the filter unit for removing post-ISI should be longer than the maximum delay time of the channel. Therefore, as the channel condition is poor and the delay time of the signal increases, the length of the required filter also increases. This increase in filter length results in an increase in the time and complexity required to calculate the coefficients.

That is, in the conventional nonlinear equalizer, the feedback component is present to reduce the stability of the equalizer, and the number of taps of the feedback filter used in the equalizer increases as the delay time of the signal by the channel becomes longer. This increase in the filter tap increases the amount of determinant calculations, thereby slowing down the channel adaptation speed of the equalizer and increasing the number of symbols that are affected by false error estimation, resulting in a reduction in signal correction capability by the equalizer.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and an object of the present invention is to reduce the number of taps of the feedback filter of the DFE, which is a nonlinear equalizer, by using a linear equalizer as a preprocessor. The linear equalizer is designed to compensate for the instability of the feedback filter by inputting to the equalizer DFE, and to improve the speed and signal quality of the system by distributing and repeatedly removing the ISI from the linear equalizer and the nonlinear equalizer DFE. An adaptive equalization system and method incorporating a nonlinear equalizer are provided.

The adaptive equalization system combining the linear equalizer and the nonlinear equalizer according to the present invention for achieving the object of the present invention, in the equalization system for reducing the ISI generated in the digital communication system, by using the FIR filter A linear equalizer which outputs a primary equalization signal by firstly equalizing and removing portions of pre-ISI and post-ISI; And a decision feedback equalizer (DFE), which is a nonlinear equalizer for outputting a second equalized signal from which residual ISI is removed from the first equalized signal in the linear equalizer by using a forward filter and a feedback filter.

The FIR filter of the linear equalizer has a relatively long asymmetrical structure from the center of the filter to remove a part of pre-ISI and post-ISI, and is set in the filter coefficient generator based on the LMS algorithm. The signal is filtered according to a filter coefficient, and the feedback filter of the DFE is configured to filter the corresponding signal according to a filter coefficient set by a filter coefficient generator based on an RLS algorithm.

The adaptive equalization method combining the linear equalizer and the nonlinear equalizer according to the present invention is an equalization method for reducing the ISI generated in the digital communication system, and the linear equalizer employing the FIR filter first equalizes the input symbols to pre- A first equalization process of outputting a first equalization signal from which the first half of the ISI and the post-ISI are removed; And a second equalization process of outputting a second equalization signal from which residual ISI is removed from the first equalization signal in the linear equalizer by a DFE, which is a nonlinear equalizer employing a forward filter and a feedback filter.

The filter coefficient of the FIR filter is adjusted according to the LMS algorithm, and the feedback filter of the DFE can reduce the number of taps of the feedback filter by using the linear equalizer as a preprocessor for post-ISI removal and according to the RLS algorithm. Characterized in that the filter coefficient is adjusted.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the following examples are merely to illustrate the present invention is not limited to the contents of the present invention.

3 is a block diagram of an adaptive equalization system combining a linear equalizer and a nonlinear equalizer according to the present invention, which reduces ISI by a channel while reducing the number of feedback taps of a DFE, which is a nonlinear equalizer, after stable convergence and convergence of a filter. It is configured to lower the error level.

As shown, the present invention uses the FIR filter 302 to first equalize the input symbols and output a linear equalizer 300 that outputs a first equalized signal from which the first half of the pre-ISI and the post-ISI are removed. It is composed of a DFE 400 which is a nonlinear equalizer that outputs a second equalized signal from which residual ISI is removed from the first equalized signal generated by the linear equalizer 300 using the filter 402 and the feedback filter 405. .

Since the linear equalizer 300 has only the same reference numerals as the general linear equalizer of FIG. 1, and the configuration thereof is the same, detailed descriptions of the detailed configuration and the same operation as in the prior art are weak.

The DFE 400 has an output of the linear equalizer 300 as an input of the DFE 400, except that the error corrector 201 and the error measuring unit 214 of FIG. 2 are not configured. Since only the DFE and the reference numerals are different, the configuration is the same, the detailed description of the configuration and the description of the same operation as in the prior art are weak.

This invention complements the function of the feedback filter 405 of the DFE 400 to remove post-ISI with the linear equalizer 300.

Linear equalizer 300 removes some of the ISI and pre-ISI and post-ISI prior to DFE 400 due to the delay of the signal generated by the channel.

As shown in FIG. 4, the first half of the post-ISI is removed using the FIR filter 302 having excellent convergence stability. In this case, the FIR filter 302 has a relatively long asymmetric shape in the latter half from the center of the filter to effectively remove the post-ISI.

In the linear equalizer 300, a signal from which a part of the post-ISI is first removed is input to the DFE 400 to remove residual ISI. Since the linear equalizer 300 is used as the preprocessor, the feedback filter 405 may use fewer than the taps of the feedback filter 203 of FIG. 2.

This reduction in the number of taps shows good performance at convergence speed and post-convergence error level, but it allows the use of algorithms such as RLS, which have a large amount of computation, delay signal processing time, and increase hardware complexity.

Therefore, since the linear equalizer 300 serving as a preprocessor has a longer post-ISI elimination compared to the DFE 400, the filter coefficient updater 303 based on the LMS algorithm showing an advantage of less computation is used. The DFE 400, which sets the coefficients of the FIR filter 302 and reduces the number of taps of the feedback filter 405, has a large amount of calculation, but is based on an RLS algorithm having excellent performance in channel adaptive convergence speed and error level after convergence. The coefficient update unit 411 may be used to set the coefficients of the feedback filter 405.

5 and 6 are graphs showing the channel adaptive convergence speed and the error level after convergence when the linear equalizer and the DFE are used alone.

Combined Adaptive equalization system combined with linear equalizer 300 and non-linear equalizer DFE 400 according to the present invention is a pre-ISI, post-ISI in the linear equalizer 300 operating as a preprocessor. Since the signal corrected by first removing the signal is applied to the DFE 400 as an input, the number of filter taps of the DFE 400 may be smaller than the number of filter taps of the existing DFE. This reduction in the number of filter taps not only reduces the range of influence of the symbol fed back when an error occurs, but also reduces the amount of computation, which leads to algorithms such as RLS that perform well at the adaptive speed of the equalizer to the channel and the error level after the adaptation. Can be used. In addition, by performing channel equalization over the second through the DFE 400, an improved signal quality can be obtained by improving the reception signal correction capability of the system requiring the equalizer.

7 shows the channel adaptive convergence speed and the post-convergence error level according to the present invention. From this, it can be seen that the system of the present invention has improved convergence speed and post-convergence error level compared to FIGS. 5 and 6. .

As described above, although described with reference to a preferred embodiment of the present invention, those skilled in the art various modifications of the present invention without departing from the spirit and scope of the invention described in the claims below Or it may be modified.

As described above, the present invention ensures a stable equalization operation through the correction of the signal through the second order through the DFE using the linear equalizer employing the FIR filter and the forward filter and the feedback filter, the convergence of the feedback filter of the DFE Since the stability can be improved, an improved signal correction effect can be obtained. In addition, as the quality of the received signal is improved, the transmission power can be reduced, and the system can be stably maintained even in a poor channel environment.

Claims (8)

An equalization system for reducing inter-symbol interference (ISI) occurring in a digital communication system, A linear equalizer which firstly equalizes input symbols by using an FIR filter and outputs a primary equalization signal from which a part of pre-ISI and post-ISI are removed; And A decision feedback equalizer (DFE) which is a nonlinear equalizer for outputting a second equalized signal from which residual ISI is removed from the first equalized signal in the linear equalizer using a forward filter and a feedback filter; Adaptive equalization system that combines a linear equalizer and a nonlinear equalizer. The FIR filter of claim 1, wherein the FIR filter of the linear equalizer is formed. Adaptive equalization system combining a linear equalizer and a nonlinear equalizer, characterized in that the latter part has a relatively long asymmetric structure from the center of the filter to remove a part of the post-ISI. The FIR filter of claim 1, wherein the FIR filter of the linear equalizer is formed. An adaptive equalization system combining a linear equalizer and a nonlinear equalizer, characterized in that the signal is filtered according to the filter coefficient set by the filter coefficient updater based on the LMS algorithm. The method of claim 1, wherein the feedback filter of the DFE An adaptive equalization system combining a linear equalizer and a nonlinear equalizer, characterized in that the signal is filtered according to the filter coefficient set by the filter coefficient updater based on the RLS algorithm. An equalization method for reducing inter-symbol interference (ISI) occurring in a digital communication system, A first equalization process for linearly equalizing an input symbol with a linear equalizer employing an FIR filter to output a first equalized signal from which a part of pre-ISI and post-ISI are removed; And A second equalization process of outputting a second equalized signal from which residual ISI is removed from the first equalized signal in the linear equalizer by a decision feedback equalizer (DFE), which is a nonlinear equalizer employing a forward filter and a feedback filter; Adaptive equalization method combining a linear equalizer and a nonlinear equalizer, characterized in that consisting of. The method of claim 5, wherein the FIR filter Adaptive equalization method combining a linear equalizer and a nonlinear equalizer, characterized in that the filter coefficient is adjusted according to the LMS algorithm. The method of claim 5, wherein the feedback filter of the DFE is Adaptive equalization method combining a linear equalizer and a nonlinear equalizer, characterized in that the number of taps of the feedback filter can be reduced by using the linear equalizer as a preprocessor for post-ISI removal. The method of claim 7, wherein the feedback filter is Adaptive equalization method combining a linear equalizer and a nonlinear equalizer, characterized in that the filter coefficients are adjusted according to the RLS algorithm.

Images