A decision feedback based adaptive frequency domain equalization method and system

Abstract

The application discloses a decision feedback based adaptive frequency domain equalization method and system, and belongs to the technical field of communication. The system comprises a single carrier frequency domain MMSE unit, a signal-to-crosstalk noise ratio judgment unit, an adaptive iteration unit and an overlap-and-cut unit. The application introduces the mean square value of data block crosstalk into the received signal-to-noise ratio detection, deduces the form of the residual code crosstalk of the low complexity frequency domain equalization system based on the overlap-and-cut algorithm; the statistical quantity based signal-to-crosstalk noise ratio estimation algorithm obtains the dynamic received signal-to-crosstalk noise ratio estimation value, and constructs the signal-to-crosstalk noise ratio index for simultaneously considering the channel noise interference and the data block crosstalk; when the signal-to-crosstalk noise ratio estimation is higher, the residual code crosstalk does not need to be iteratively eliminated, and the communication index requirement in the environment can be met; and when the signal-to-crosstalk noise ratio estimation is lower, the residual code crosstalk is iteratively eliminated to improve the bit error rate performance of the single carrier frequency domain MMSE equalization system, that is, the decision feedback based adaptive frequency domain equalization is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to a decision feedback based adaptive frequency domain equalization method and system, belonging to the field of communication technology. BACKGROUND

[0002] The purpose of the frequency domain equalization system based on the minimum mean square error criterion is to make the system error rate minimum, which is essentially a compromise between channel noise interference and inter-symbol interference. In the severe channel with large noise interference and serious multipath effect, the performance of the current equalization system cannot guarantee the reliable and stable communication of the lightweight terminal. The residual inter-symbol interference under the minimum mean square error criterion can be eliminated by an iterative algorithm, but the algorithm has high complexity and is not suitable for calculating each frame of data. At present, there is a lack of channel index considering noise interference and inter-block interference, and the terminal cannot judge the current communication situation to select whether to use the algorithm. SUMMARY

[0003] In view of the problem of high complexity of the iterative elimination algorithm, the purpose of the present application is to provide a decision feedback based adaptive frequency domain equalization method and system, which introduces the inter-block interference mean square value into the received signal-to-noise ratio detection, deduces the form of the residual inter-symbol interference of the low complexity frequency domain equalization system based on the overlap-and-cut algorithm, constructs a signal-to-noise ratio index considering the channel noise interference and the inter-block interference, and realizes the adaptive iterative elimination of the residual inter-symbol interference of the frequency domain equalization system based on the index.

[0004] The purpose of the present application is realized by the following technical solutions.

[0005] The present application discloses a decision feedback based adaptive frequency domain equalization method, comprising the following steps:

[0006] Step 1: separate the pilot section from the data, divide the data section x(n) to be processed into r data blocks x1(n), x2(n),... x r (n) of equal length, and perform FFT transformation on the adjacent data blocks together.

[0007] Step 2: perform FFT transformation on the pilot section data to obtain the frequency domain channel estimation result and the noise mean square value estimation, and generate an MMSE equalizer. Use the equalizer to perform channel compensation on the data after r-1 times of FFT transformation, and obtain the frequency domain response X MMSE (K) of the MMSE equalization output result, perform block IFFT transformation on the compensated frequency domain signal to the time domain, perform cutting and merging to obtain the time domain response x mmse (n) of the single carrier frequency domain MMSE equalization output result.

[0008] The specific implementation method of Step 2 is:

[0009] The MMSE equalizer is generated, and the frequency domain response C(K) of the MMSE equalizer is

[0010]

[0011] Wherein: is the noise variance, is the average power of the transmitted signal, and H(K) is the channel frequency domain response.

[0012] The data after the r-1 times of combined FFT transformation is compensated by using the equalizer to obtain the frequency domain response X MMSE (K) of the MMSE equalization output result, and the compensated frequency domain signal is subjected to block IFFT transformation to the time domain.

[0013] The data segment x(n) to be processed is divided into r data blocks with a length of N / 2 by overlapping and covering

[0014] x(n) = [x1(n), x2(n),..., X T (n)]

[0015] Starting from the first data block x1(n), two data blocks are combined together for FFT transformation

[0016] X t-1，t (K) = FFT(x1(n), x2(n))

[0017] The combined FFT transformation is performed r-1 times, and the FFT transformation of the entire data segment is completed. For a data block x t (n), the FFT transformation result thereof appears in the t times of combined FFT transformation result and the t-1 times of combined FFT transformation result, and the overlapping and covering of the FFT transformation result is completed.

[0018] For the first data block x1(n) and the last data block x T (n), the FFT transformation result thereof appears in only one combined FFT transformation. Therefore, the information of the two data blocks is still lost, and the partial zero padding method is used to compensate for the loss.

[0019] A full 0 sequence with a length of half a data block is added before the first data block x1(n) and after the last data block x T (n), and the data sequence x(n) z after zero padding is in the form of:

[0020] x(n) z = [zeros(n / 2), x1(n), x2(n),..., X T(n), zeros(n / 2)]

[0021] At this time, the sequence of the first combined FFT transform becomes:

[0022] X 1，2 (K) z = FFT(zeros(1, n / 2), x1(n), x2(1, n / 2))

[0023] The sequence of the second combined FFT transform becomes:

[0024] X 2,3 (K) z = FFT(x1(n / 2+1, n), X2(n), x3(1, n / 2))

[0025] At this time, the FFT transform result of the data block x1(n) appears in the two combined FFT transforms. The FFT transform result of the last data block x r (n) also appears in the last combined FFT transform and the second-to-last combined FFT transform, achieving full data segment overlap coverage.

[0026] After the overlap coverage link, r combined FFT transform results are obtained, and the length of each combined FFT transform is two data block lengths. The sequence of n / 2 at both ends of the equalization result is cut off, removing the inter-block crosstalk while ensuring that the FFT transform output length is still n, i.e., obtaining the time domain response x mmse (n) of the single-carrier frequency domain MMSE equalization output result.

[0027] Step 3: Perform signal-to-noise ratio estimation based on a low-complexity algorithm of a statistical quantity to obtain a signal-to-noise ratio estimate SNR and a signal-to-crosstalk noise ratio estimate SCNR.

[0028] The specific implementation method of Step 3 is:

[0029] The signal-to-crosstalk noise ratio estimation is performed based on a low-complexity algorithm of a statistical quantity to estimate the received signal strength and noise interference strength. The second moment and the fourth moment of the received signal are represented by the received signal energy S and the noise energy N, and the signal-to-noise ratio is further obtained by solving the equations.

[0030] Let The second moment L2 of the received signal R(K) is represented as:

[0031] L2 = E{r(n)r * (n)} = E{|b(n)| 2} + E{b(n)v * (n)}

[0032] + E{|v(n)| * + E{|v(n)| 2}

[0033] = S + N

[0034] where x(n) is the transmitted signal, h(n) is the channel time response, r(n) is the received signal, v(n) is the noise time response, and represents a circular convolution.

[0035] The fourth moment L4 of the received signal R(K) is expressed as:

[0036] L4 = E{(r(n)r * (n)) 2} = E{|b(n)| 4} + 2 * (E{|b(n)| 2 b(n)v * (n)}

[0037] +E{|b(n)| 2 b(n)v(n)}) + E{(b(n)v * (n)) 2}

[0038] +4*E{|b(n)| 2 |v * (n)| 2} + E{(b(n) * v(n)) 2}

[0039] +2*(E{|v(n)| 2 b(n)v * (n)} + E{|v(n)| 2 b(n)v(n)})

[0040] +E{|v(n)| 4} = S 2 +4SN + N 2

[0041] The second moment expression and the fourth moment expression are combined to obtain the signal-to-noise ratio estimate SNR:

[0042]

[0043] After adding crosstalk, the signal-to-crosstalk-and-noise ratio estimate SCNR is:

[0044]

[0045] where σ IBIdenotes the inter-block interference variance.

[0046] Step 4: compare the received signal-to-noise ratio estimate with the set threshold value, when the received signal-to-noise ratio is higher than the received signal-to-noise ratio threshold value, directly output the time domain response X mmse (n) of the single carrier frequency domain MMSE equalization output result, and output the decision. When the received signal-to-noise ratio at this time is lower than the received signal-to-noise ratio threshold value, the residual inter-symbol interference is iteratively eliminated, the x(n) is subjected to FFT transformation to obtain the frequency domain estimate value X(K) of the transmitted signal, and it is determined that the estimate value signal is the real transmitted signal X(K), the estimate value I(K) of the residual inter-symbol interference is calculated through the signal, and the single carrier frequency domain MMSE equalization output signal X MMSE (K) is compensated to obtain the iterative output signal X DFE (K) eliminating the residual inter-symbol interference, the frequency domain response is subjected to IFFT transformation and decision to obtain the output eliminating the residual inter-symbol interference, that is, the adaptive frequency domain equalization based on decision feedback is realized.

[0047] Step 4: the specific implementation method is as follows:

[0048] When the received signal-to-noise ratio is higher than the received signal-to-noise ratio threshold value, the frequency domain form I(K) of the residual inter-symbol interference is:

[0049]

[0050] At this time, the accurate transmitted signal X(K) cannot be known in advance at the receiving end, and the estimate value of I(K) is solved through iteration.

[0051] The first equalization result output by the iteration-free frequency domain algorithm, that is, the first estimate value X0(K) of the transmitted signal; the signal is fed back to the single carrier frequency domain equalization system as the transmitted signal. After the single carrier frequency domain equalization system obtains the feedback input, the estimate value is used to estimate the residual inter-symbol interference,

[0052]

[0053] The estimate value I1(K) of the residual inter-symbol interference is taken as the real residual inter-symbol interference value, and the first estimate value X0(K) of the transmitted signal is compensated, that is, the equalization output X DFE-1 (K) after one iteration is obtained:

[0054] X DFE-1 (K) = X0(K) - I1(K)

[0055] Since the first estimate value X0(K) of the transmitted signal still has errors compared with the real transmitted signal X(K), the estimate value I1(K) of the residual inter-symbol interference has errors compared with the first estimate value X0(K) of the transmitted signal.MMSE The actual residual intersymbol interference in (K) also has errors. The frequency domain response X DFE-1 (K) still has errors from the true transmitted data, and the reason for the errors is the inaccuracy of the estimated value I1(K) of the residual intersymbol interference, and the reason for the inaccuracy is the error between the feedback value fed back into the system and the true transmitted signal, i.e. the error between the signal X0(K) output by the non-iterative frequency domain equalization algorithm and the true transmitted signal: Therefore, reducing the error reduces the output error, and the frequency domain response X DFE-1 (K) after one iteration of the elimination of the residual intersymbol interference is selected as the output value, because the error between (K) and the true transmitted signal is smaller than the error between X0(K) and the true transmitted signal. DFE-1 (K) is taken as the feedback input and X DFE-1 (K) is taken as the feedback input and X

[0056]

[0057] The estimated value is used to compensate the non-iterative frequency domain equalization output signal X0(K), and a more accurate output value X DFE-2 (K) is obtained, i.e. the output value after two iterations of the elimination of the residual intersymbol interference:

[0058] X DFE-2 (K) = X DFE-1 (K) - I1(K)

[0059] According to the output value X DFE-2 (K) after the elimination, a decision feedback based adaptive frequency domain equalization is realized.

[0060] The application discloses a decision feedback based adaptive frequency domain equalization system, which is used for realizing a decision feedback based adaptive frequency domain equalization method.

[0061] The single-carrier frequency domain MMSE unit introduces the intersymbol interference mean square value between data blocks into received signal noise ratio detection, estimates a channel estimation result and a noise mean square value simultaneously by using pilot data, generates a frequency domain MMSE equalizer according to a minimum mean square error criterion by using the two estimated values, and compensates an effective data segment by using the equalizer.

[0062] The signal-to-crosstalk-plus-noise ratio judging unit is used for calculating the signal-to-crosstalk-plus-noise ratio and judging whether the signal-to-crosstalk-plus-noise ratio is within the set index allowable range, inputting the judging result and the noise ratio parameter to the adaptive iteration unit, and reducing the algorithm implementation difficulty.

[0063] The adaptive iteration unit introduces the crosstalk mean square value between data blocks into a signal-to-noise ratio estimation formula, obtains an estimation method of the parameter "signal-to-crosstalk noise ratio" considering the noise interference intensity and the channel fading condition, and judges whether to perform iteration equalization according to the estimation method judging result and the noise ratio parameter.

[0064] The overlap cutting unit divides the input signal into a plurality of data blocks, then overlaps the data blocks, and finally cuts and combines, so as to further eliminate the crosstalk between data blocks.

[0065] Beneficial effects:

[0066] 1. For the crosstalk between data blocks of the traditional single carrier frequency domain MMSE equalization system, the adaptive frequency domain equalization method and system based on decision feedback disclosed in the application introduces the crosstalk between data blocks, channel noise and inter-symbol interference based on the minimum error criterion frequency domain equalizer, and adopts overlap cutting to complete frequency domain equalization, so as to further eliminate the crosstalk between data blocks.

[0067] 2. The adaptive frequency domain equalization method and system based on decision feedback disclosed in the application adopts an iteration method to eliminate the residual inter-symbol interference of the traditional frequency domain MMSE equalization system. Meanwhile, the crosstalk mean square value between data blocks is introduced into a signal-to-noise ratio estimation formula, so as to obtain an estimation method of the parameter "signal-to-crosstalk noise ratio" considering the noise interference intensity and the channel fading condition. The signal-to-crosstalk noise ratio estimation value is used to adaptively adjust whether to perform the residual inter-symbol interference iteration elimination calculation with higher complexity, so as to improve the adaptive frequency domain equalization efficiency.

[0068] 3. The adaptive frequency domain equalization method and system based on decision feedback disclosed in the application introduces the crosstalk mean square value between data blocks into the received signal-to-noise ratio detection, deduces the form of the residual inter-symbol interference of the low complexity frequency domain equalization system based on the overlap cutting algorithm; the signal-to-crosstalk noise ratio estimation algorithm based on the statistical quantity obtains a dynamic received signal-to-crosstalk noise ratio estimation value, and constructs the signal-to-crosstalk noise ratio index considering the channel noise interference and the crosstalk between data blocks; when the signal-to-crosstalk noise ratio estimation is high, the residual inter-symbol interference does not need to be iteratively eliminated, and the communication index requirement in the environment can be met; when the signal-to-crosstalk noise ratio estimation is low, the residual inter-symbol interference is iteratively eliminated, the bit error rate performance of the single carrier frequency domain MMSE equalization system is improved, and the adaptive frequency domain equalization based on decision feedback is realized. BRIEF DESCRIPTION OF DRAWINGS

[0069] In order to make the technical solutions in the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort based on these drawings.

[0070] Figure 1 Flow chart of the adaptive frequency domain equalization method based on decision feedback of the present application.

[0071] Figure 2 Frame structure diagram of the adaptive frequency domain equalization method based on decision feedback of the present application. DETAILED DESCRIPTION

[0072] The present application will be further described and explained with reference to the accompanying drawings and embodiments.

[0073] As shown in the figure, the adaptive frequency domain equalization method based on decision feedback disclosed in the present embodiment has the following specific implementation steps: Figure 1 Step 1: Referring to the frame structure, separate the pilot section from the data, divide the data section x(n) to be processed into 16 data blocks x1(n), x2(n),... xr(n) of equal length, and perform FFT transformation on the adjacent data blocks together.

[0074] Step 2: Perform FFT transformation on the pilot section data to obtain the frequency domain channel estimation result and noise mean square value estimation, and generate the MMSE equalizer. Use this equalizer to perform channel compensation on the 15 times of merged FFT transformed data to obtain the frequency domain response X(K) of the MMSE equalization output result, and perform block IFFT transformation on the compensated frequency domain signal to the time domain, and perform clipping and merging to obtain the time domain response x(n) of the single carrier frequency domain MMSE equalization output result. MMSE mmse

[0075] Step 2: The specific implementation method is as follows:

[0076] Generate the MMSE equalizer, and the frequency domain response C(K) of the MMSE equalizer is

[0077]

[0078]

[0079] Among them: is the noise variance, is the average power of the transmitted signal, and H(K) is the channel frequency domain response.

[0080] ​​​The equalizer is used to compensate the channel for the data after 15 times of combined FFT transform to obtain the frequency domain response X of the MMSE equalization output result MMSE (K), and the compensated frequency domain signal is subjected to block IFFT transform to the time domain.

[0081] Then, overlapping is performed, and the data segment x(n) to be processed is divided into 16 data blocks with a length of 256

[0082] x(n) = [x1(n), x2(n),..., X 16 (n)]

[0083] From the first data block x1(n), two data blocks are combined together for FFT transform

[0084] X t-1，t (K) = FFT(x1(n), x2(n))

[0085] There are 15 times of combined and FFT transform, and the FFT transform of the whole data segment is completed. For a data block x t (n), the FFT transform result thereof appears in the tth combined FFT transform result and the t-1th combined FFT transform result, and the overlapping of the FFT transform result is completed.

[0086] For the first data block x1(n) and the last data block x 16 (n), the FFT transform result thereof appears in only one combined FFT transform. Therefore, the information of the two data blocks is still lost, and the partial zero padding method is used to complete the compensation for the lost part.

[0087] A full zero sequence with a length of half of the data block is added before the first data block x1(n) and after the last data block x 16 (n), and the data sequence x(n) z after zero padding is in the form of:

[0088] x(n) z = [zeros(n / 2), x1(n), x2(n),..., X T (n), zeros(n / 2)]

[0089] At this time, the sequence of the first combined FFT transform becomes:

[0090] X 1，2 (K) z = FFT(zeros(1, n / 2), x1(n), x2(1, n / 2))

[0091] The sequence of the second combined FFT transform becomes:

[0092] X 2,3 (K) z = FFT(x1(n / 2+1, n), X2(n), x3(1, n / 2))

[0093] At this time, the FFT transform result of the data block x1(n) appears in two merging FFT transforms. The FFT transform result of the last data block x r (n) also appears in the last merging FFT transform and the second-to-last merging FFT transform, realizing the full data segment overlap coverage.

[0094] After the overlap coverage link, there are 16 merging FFT transform results, and the length of each merging FFT transform is two data block lengths. The sequence of n / 2 at both ends of the equalization result is cut off, removing the data block crosstalk while ensuring that the FFT transform output length is still 256.

[0095] Step 3: Perform signal-to-crosstalk noise ratio estimation based on a low-complexity algorithm based on statistics to obtain a signal-to-noise ratio estimate SNR and a signal-to-crosstalk noise ratio estimate SCNR.

[0096] The specific implementation method of Step 3 is:

[0097] The signal-to-crosstalk noise ratio estimation is based on a low-complexity algorithm based on statistics to estimate the received signal strength and noise interference strength. The second moment and fourth moment of the received signal energy S and noise energy N are used to further solve the signal-to-noise ratio by solving the equations.

[0098] Let The second moment L2 of the received signal R(K) is expressed as:

[0099] L2 = E{r(n)r * (n)} = E{|b(n)| 2} + E{b(n)v * (n)}

[0100] +E{b(n) * v(n)} + E{|v(n)| 2}

[0101] = S + N

[0102] Where: x(n) is the transmitted effective signal, h(n) is the channel time domain response, r(n) is the received signal, v(n) is the noise time domain response, and represents circular convolution.

[0103] The fourth moment L4 of the received signal R(K) is expressed as:

[0104] L4 = E{(r(n)r * (n)) 2} = E{|b(n)| 4} + 2*(E{|b(n)| 2 b(n)v * (n)}

[0105] +E{|b(n)| 2 b(n)v(n)}) + E{(b(n)v * (n)) 2}

[0106] +4*E{|b(n)| 2 |v * (n)| 2} + E{(b(n) * v(n)) 2}

[0107] +2*(E{|v(n)| 2 b(n)v * (n)} + E{|v(n)| 2 b(n)v(n)})

[0108] +E{|v(n)| 4} = S 2 +4SN + N 2

[0109] The second moment expression is combined with the fourth moment expression to obtain the signal-to-noise ratio estimate SNR:

[0110]

[0111] After adding crosstalk, the signal-to-crosstalk noise ratio estimate SCNR is:

[0112]

[0113] Where: σ IBI represents the inter-block interference variance.

[0114] Step 4: Compare the received signal-to-noise ratio estimate with the set threshold value. When the received signal-to-crosstalk noise ratio is higher than the received signal-to-crosstalk noise ratio threshold value, directly output the time domain response X mmse(n) decision and output. When the received signal-to-crosstalk ratio is lower than the received signal-to-crosstalk ratio threshold value at this time, the residual inter-symbol interference is iteratively cancelled, the x(n) is subjected to FFT transformation to obtain the frequency domain estimation value X(K) of the transmitted signal, and it is determined that the estimation value signal is the true transmitted signal X(K), the estimation value I(K) of the residual inter-symbol interference is calculated through the signal, and the single-carrier frequency domain MMSE equalization output signal X MMSE (K) is compensated to obtain the iterative output signal X DFE (K) which is subjected to IFFT transformation and decision to obtain the output of the iterative cancellation of the residual inter-symbol interference, that is, the adaptive frequency domain equalization based on decision feedback is realized.

[0115] The specific implementation method of Step 4 is:

[0116] When the received signal-to-crosstalk ratio is higher than the received signal-to-crosstalk ratio threshold value, the frequency domain form I(K) of the residual inter-symbol interference is:

[0117]

[0118] The accurate transmitted signal X(K) cannot be known in advance at the receiving end, and the estimation value of I(K) is solved through iteration.

[0119] The first equalization result output by the iterative frequency domain algorithm, that is, the first estimation value X0(K) of the transmitted signal; the signal is fed back to the single-carrier frequency domain equalization system as the transmitted signal. After the single-carrier frequency domain equalization system obtains the feedback input, the estimation value is used to estimate the residual inter-symbol interference,

[0120]

[0121] The estimation value I1(K) of the residual inter-symbol interference is taken as the true residual inter-symbol interference value, and the first estimation value X0(K) of the transmitted signal is compensated, that is, the equalization output X DFE-1 (K) after one iteration can be obtained:

[0122] X DFE-1 (K) = X0(K) - I1(K)

[0123] Since the first estimation value X0(K) of the transmitted signal still has errors compared with the true transmitted signal X(K), the estimation value I1(K) of the residual inter-symbol interference also has errors compared with the actual residual inter-symbol interference in the first estimation value X MMSE (K) of the transmitted signal. After one iteration of the residual inter-symbol interference cancellation, the frequency domain response X DFE-1(K) still has error with the real transmitted data, the error is caused by the inaccuracy of the estimated value I1(K) of the residual inter-symbol interference, and the inaccuracy is caused by the error between the feedback value fed back to the system and the real transmitted signal, i.e. the error between the signal X0(K) output by the non-iterative frequency domain equalization algorithm and the real transmitted signal; therefore, reducing the error can reduce the output error, and the frequency domain response X DFE-1 (K) has smaller error with the real transmitted signal than the signal X0(K) output by the non-iterative frequency domain equalization algorithm, and the frequency domain response X DFE-1 (K) is selected as the feedback input and is compensated by X DFE-1 (K) to obtain a more accurate estimated value I2(K) of the residual inter-symbol interference:

[0124]

[0125] Compensating the non-iterative frequency domain equalization output signal X0(K) by the estimated value can obtain a more accurate output value X DFE-2 (K), i.e. the output value after twice iterative elimination of the residual inter-symbol interference:

[0126] X DFE-2 (K) = X DFE-1 (K) - I1(K)

[0127] According to the output value X DFE-2 (K) after elimination, the adaptive frequency domain equalization based on decision feedback is realized.

[0128] The embodiment discloses a kind of adaptive frequency domain equalization based on decision feedback system, for realizing a kind of adaptive frequency domain equalization based on decision feedback method.The kind of adaptive frequency domain equalization based on decision feedback system, including single carrier frequency domain MMSE unit, signal-to-interference-plus-noise ratio judging unit, adaptive iteration unit and overlap shear unit.

[0129] The single carrier frequency domain MMSE unit introduces the inter-block interference mean square value into received signal-to-noise ratio detection, estimates the channel estimation result and noise mean square value simultaneously using pilot segment data, generates frequency domain MMSE equalizer according to minimum mean square error criterion using the two estimated values, and then compensates effective data segment using the equalizer.

[0130] The signal-to-interference-plus-noise ratio judging unit is used to calculate signal-to-interference-plus-noise ratio and judge whether signal-to-interference-plus-noise ratio is within the set index allowable range, input the judgment result and noise ratio parameter to the adaptive iteration unit, and reduce the algorithm implementation difficulty.

[0131] The adaptive iteration unit introduces the mean square value of the crosstalk between data blocks into a signal-to-noise ratio estimation formula, obtains an estimation method of a parameter "signal-to-crosstalk-to-noise ratio" considering the noise interference intensity and the channel fading condition, and judges whether to perform iteration equalization according to the estimation method and the noise ratio parameter.

[0132] The overlap cutting unit divides the input signal into a plurality of data blocks, then overlaps the data blocks, and finally cuts and combines, so as to further eliminate the crosstalk between the data blocks.

[0133] The above detailed description further describes the purpose, technical scheme and beneficial effects of the application. It should be understood that the above description is only a specific embodiment of the application and is not used to limit the protection scope of the application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the application should be included in the protection scope of the application.

Claims

1. A decision feedback based adaptive frequency domain equalization method characterized by: It comprises the following steps, Step1: separate the pilot section from the data section, divide the data section x(n) to be processed into r data blocks x1(n), x2(n),... x r (n) of equal length, combine adjacent data blocks and perform FFT transform together; Step 2: the pilot segment data is subjected to FFT transformation to obtain a frequency domain channel estimation result and a noise mean square value estimation to generate a MMSE equalizer; the equalizer is used to compensate the data subjected to r-1 times of merging FFT transformation respectively to obtain a frequency domain response X of the MMSE equalizer output result MMSE (K); the compensated frequency domain signal is subjected to block IFFT transformation to the time domain, and is subjected to cutting and merging to obtain a time domain response x of the single carrier frequency domain MMSE equalizer output result mmse (n); Step 3: low-complexity algorithm based on statistics is used to estimate the signal-to-crosstalk noise ratio to obtain the signal-to-noise ratio SNR and the crosstalk-added signal-to-crosstalk noise ratio SCNR; Step 4: comparing the received signal-to-noise ratio estimation value with a set threshold value, when the received signal-to-noise ratio is higher than the received signal-to-noise ratio threshold value, directly outputting the time domain response x mmse (n) making a decision and outputting; When the received signal-to-interference ratio at this time is lower than the received signal-to-interference ratio threshold value, the residual intersymbol interference is iteratively eliminated, the x(n) is subjected to FFT transformation to obtain the frequency domain estimation value X(K) of the transmitted signal, and it is determined that the estimation value signal is the real transmitted signal X(K), the estimation value I(K) of the residual intersymbol interference is calculated through the signal, the estimation value is used to compensate the single-carrier frequency domain MMSE equalizer output signal X MMSE (K), to obtain the iterative output signal X DFE (K) eliminating the residual intersymbol interference DFE (K) is subjected to IFFT transformation and decision to obtain the output of the iterative elimination of the residual intersymbol interference, that is, the adaptive frequency domain equalization based on decision feedback is realized.

2. The method for decision feedback based adaptive frequency domain equalization as recited in claim 1, wherein: The specific implementation method of Step 2 is, The MMSE equalizer is generated, and the frequency domain response C(K) of the MMSE equalizer is wherein: is the noise variance, is the average power of the transmitted signal, H(K) is the channel frequency domain response; The equalizer is used to compensate the channel of the data after r-1 times of combined FFT transformation to obtain the frequency domain response X of the output result of the MMSE equalizer MMSE (K), and block IFFT transformation is performed on the compensated frequency domain signal to the time domain. The data segment x(n) to be processed is divided into r data blocks with a length of N by overlapping coverage x(n) = [x1(n), x2(n),..., xN(n)]T r (n)] Starting from the first data block x1(n), the adjacent two data blocks are merged together for FFT transformation X t-1,t (K) = FFT(x1(n), x2(n)) The r-1 times of merging and FFT transforming are performed to complete the FFT transforming of the whole data segment; for a data block x t (n), the FFT transforming result of which appears in the t times of merging FFT transforming result and the t-1 times of merging FFT transforming result, to complete the overlapping of the FFT transforming result; For the first data block x1(n) and the last data block x r (n), the FFT transform results only appear once in the combined FFT transform; therefore, the information of these two data blocks is still lost, and the partial zero padding method is used to compensate for the loss. A half-block long all-zero sequence is added before the first data block x1(n) and after the last data block xN(n). r The zero-padded data sequence x(n) is of the form: z x(n) = x1(n) x2(n) x3(n) x4(n) x5(n) x6(n) x7(n) x8(n) x(n) z = [zeros(n / 2), x1(n), x2(n),..., x r (n), zeros(n / 2)] At this time, the sequence of the first merged FFT transformation becomes: X 1,2 (K) z = FFT(zeros(1, n / 2), x1(n), x2(1, n / 2)) The sequence of the second merged FFT transformation becomes: X 2,3 (K) z = FFT(x1(n / 2 + 1, n), x2(n), x3(1, n / 2)) At this time, the FFT transform result of the data block x1(n) appears twice in the merging FFT transform; the FFT transform result of the last data block x r (n) also appears twice in the merging FFT transform, i.e., the last merging FFT transform and the second last merging FFT transform, to realize the overlapping coverage of the whole data segment. After the overlapping cover step, r times of combined FFT transform results are obtained, and the length of each combined FFT transform is two data block lengths; the sequences of n / 2 at both ends of the equalization result are cut off, so that the crosstalk between data blocks is removed and the FFT transform output length is still n, that is, the time domain response x of the single carrier frequency domain MMSE equalizer output result is obtained mmse (n).

3. The method for decision feedback based adaptive frequency domain equalization as recited in claim 2, wherein: The specific implementation method of Step 3 is, The low-complexity algorithm based on statistics is used to estimate the received signal strength and noise interference strength; the second moment and fourth moment of the received signal are represented by the received signal energy S and noise energy N, and the signal-to-noise ratio is further obtained by solving the equations; Let The second moment L2 of the received signal R(K) is expressed as: L2 = E{r(n)r * (n)} = E{|b(n)| 2} + E{b(n)v * (n)} + E{b(n) * v(n)} + E{|v(n)| 2} =S+N where x(n) is the transmitted signal, h(n) is the channel time response, r(n) is the received signal, and v(n) is the noise time response, denotes a circular convolution; The fourth moment L4 of the received signal R(K) is represented as: L4 = E{(r(n)r * (n)) 2} = E{|b(n)| 4} + 2*(E{|b(n)| 2 b(n)v * (n)} +E{|b(n)| 2 b(n)v(n)})+E{(b(n)v * (n)) 2 } +4*E{|b(n)| 2 |v * (n)| 2 }+E{(b(n)*v(n)) 2 } + 2 * (E{|v(n)| 2 b(n)v * (n)}+E{|v(n)| 2 b(n)v(n)}) + E{|v(n)| 4} = S 2 + 4SN + N 2 The second moment expression and the fourth moment expression are combined to obtain the signal-to-noise ratio SNR: After adding crosstalk, the crosstalk-added signal-to-crosstalk noise ratio SCNR is: where: σ IBI denotes the inter-block interference variance.

4. The method for decision feedback based adaptive frequency domain equalization as recited in claim 3, wherein: The specific implementation method of Step 4 is, When the received signal-to-crosstalk noise ratio is higher than the received signal-to-crosstalk noise ratio threshold value, the frequency domain form I(K) of the residual inter-code crosstalk is: The accurate transmitted signal X(K) cannot be known in advance at the receiving end, and the estimation value of I(K) is solved by iteration; The first equalization result output by the non-iterative frequency domain algorithm is the first estimation value X0(K) of the transmitted signal; this signal is fed back as the transmitted signal to the single-carrier frequency domain equalization system; after the single-carrier frequency domain equalization system obtains the feedback input, the estimation value is used to estimate the residual inter-code crosstalk, The estimated value of the residual intersymbol interference I1(K) is used as the real residual intersymbol interference value to compensate the first estimated value X0(K) of the transmission signal, i.e. the output of the equalization after one iteration X1(K) is obtained DFE-1 (K) X DFE-1 (K) = X0(K) - I1(K) Since the first estimate of the transmitted signal X0(K) still has errors from the real transmitted signal X(K), the estimate of the residual intersymbol interference I1(K) also has errors from the actual residual intersymbol interference in the first estimate of the transmitted signal X0(K); the frequency domain response X DFE-1 (K) after one iteration of the residual intersymbol interference still has errors from the real transmitted data; the frequency domain response X DFE-1 (K) after one iteration of the residual intersymbol interference has less errors from the real transmitted signal than the signal X0(K) output by the non-iterative frequency domain equalization algorithm, the frequency domain response X DFE-1 (K) after one iteration of the residual intersymbol interference is selected as the feedback input and the more accurate estimate of the residual intersymbol interference I2(K) is obtained using X DFE-1 (K) after one iteration of the residual intersymbol interference is selected as the feedback input and the more accurate estimate of the residual intersymbol interference I2(K) is obtained using X Compensating the non-iterative frequency domain equalization output signal X0(K) with the estimation value, a more accurate output value X DFE-2 (K) can be obtained, i.e. the output value after the secondary iteration elimination of the residual inter-symbol interference: X DFE-2 (K) = X DFE-1 (K) - I1(K) According to the eliminated output value X DFE-2 (K) Implement adaptive frequency domain equalization based on decision feedback.

5. A decision feedback based adaptive frequency domain equalization system for implementing a decision feedback based adaptive frequency domain equalization method as claimed in claim 1, 2, 3 or 4, characterized by: The single-carrier frequency domain MMSE unit, the signal-to-crosstalk noise ratio judgment unit, the adaptive iteration unit, and the overlap cutting unit are included. The single-carrier frequency domain MMSE unit introduces the inter-block crosstalk mean square value into the received signal-to-noise ratio detection, estimates the channel estimation result and the noise mean square value at the same time using the pilot segment data, generates the frequency domain MMSE equalizer according to the minimum mean square error criterion using the two estimation values, and compensates the effective data segment using the equalizer; The signal-to-crosstalk noise ratio judgment unit is used to calculate the signal-to-crosstalk noise ratio and judge whether the signal-to-crosstalk noise ratio is within the set index allowable range, and input the judgment result and the noise ratio parameter to the adaptive iteration unit; The adaptive iteration unit introduces the inter-block crosstalk mean square value into the signal-to-noise ratio estimation formula to obtain an estimation method of the parameter "signal-to-crosstalk noise ratio" considering the noise interference strength and the channel fading condition at the same time, and selects whether to perform iterative equalization according to the estimation method judgment result and the noise ratio parameter; The overlap cutting unit divides the input signal into multiple data blocks, overlaps the data blocks, and finally cuts and merges to further eliminate the inter-block crosstalk.

Images