Self-adaptive frequency domain equalization method and system based on decision feedback
By introducing the mean square value and signal-to-chain noise ratio estimation of crosstalk between data blocks in the frequency domain equalization system, a low-complex overlapping shear algorithm is constructed, which solves the compromise between noise interference and crosstalk between codes, and realizes the harsh channel Reliable communication and reduced bit error rate in.
Patent Information
- Application Number
- CN202410934406.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-12
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-07-12
AI Technical Summary
In harsh channels with large noise interference and severe multipath effect, the existing frequency domain equalization system cannot guarantee reliable and stable communication of lightweight feather quantization terminals, and the iterative algorithm is highly complex and is not suitable for real-time data processing.
Adaptive frequency domain equalization method based on judgment feedback is adopted, by introducing the mean square value of crosstalk between data blocks, a low-complex overlap shear algorithm is constructed, a minimum mean square error (MMSE) equalizer is generated, and a signal-to-stream noise ratio is used. The estimated value determines whether iteratively eliminates crosstalk between residual codes.
It realizes the reduction of bit error rate in bad channels, improves communication stability, and dynamic adjustment of iterative calculations, reducing algorithm complexity, and is suitable for real-time processing of light terminals.
Smart Images

Figure CN119011345A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to an adaptive frequency domain equalization method and system based on decision feedback, belonging to the technical field of communication. Background Art
[0002] The purpose of the frequency domain equalization system based on the minimum mean square error criterion is to minimize the system bit error rate. In essence, it is a compromise between channel noise interference and inter-code interference. In harsh channels with large noise interference and severe multipath effects, the current equalization system performance cannot guarantee reliable and stable communication of lightweight feather quantization terminals. The residual inter-code interference under the minimum mean square error criterion can be eliminated by an iterative algorithm, but the algorithm is highly complex and is not suitable for calculating each frame of data using the algorithm. At present, there is a lack of channel indicators that can simultaneously consider noise interference and crosstalk between data blocks. The terminal cannot judge the current communication situation and choose whether to use the algorithm. Summary of the invention
[0003] In order to solve the problem of high complexity of iterative elimination algorithm, the purpose of the present invention is to provide an adaptive frequency domain equalization method and system based on decision feedback, introduce the mean square value of crosstalk between data blocks into the received signal-to-noise ratio detection, derive the form of residual inter-code crosstalk in a low-complexity frequency domain equalization system based on the overlapping shearing algorithm, construct a signal-to-noise ratio index for simultaneously considering channel noise interference and crosstalk between data blocks, and based on this index, realize adaptive iterative elimination of residual inter-code crosstalk in the frequency domain equalization system.
[0004] The purpose of the present invention is achieved through the following technical solutions.
[0005] The present invention discloses an adaptive frequency domain equalization method based on decision feedback, comprising the following steps:
[0006] Step 1: Separate the pilot segment from the data and divide the data segment x(n) to be processed into r data blocks of equal length x1(n), x2(n), ...x r (n) Adjacent data blocks are combined and FFT transformed.
[0007] Step 2: Perform FFT transformation on the pilot band data to obtain the frequency domain channel estimation result and noise mean square value estimation to generate the MMSE equalizer. Use this equalizer to perform channel compensation on the data after r-1 times of combined FFT transformation to obtain the frequency domain response X of the MMSE equalization output result. MMSE (K), the compensated frequency domain signal is transformed into the time domain by block IFFT, and the time domain response x of the single carrier frequency domain MMSE equalization output result is obtained by cutting and merging. mmse (n).
[0008] The specific implementation method of Step 2 is:
[0009] Generate MMSE equalizer, the frequency domain response C(K) of MMSE equalizer is
[0010]
[0011] in: is the noise variance, is the average power of the transmitted signal, and H(K) is the channel frequency domain response.
[0012] This equalizer is used to perform channel compensation on the data after r-1 times of combined FFT transformation to obtain the frequency domain response X of the MMSE equalization output result. MMSE (K) Perform block IFFT transformation on the compensated frequency domain signal to the time domain.
[0013] Overlapping is performed to divide the data segment x(n) to be processed into r data blocks of the same length N.
[0014] x(n)=[x1(n),x2(n),...,x T (n)]
[0015] Starting from the first data block x1(n), merge the two data blocks together for FFT transformation
[0016] X t-1,t (K) = FFT(x1(n),x2(n))
[0017] A total of r-1 merging and FFT transformations are performed to complete the FFT transformation of the entire data segment. t (n), its FFT transformation result appears simultaneously in the t-th combined FFT transformation result and the t-1-th combined FFT transformation result, completing the overlapping coverage of the FFT transformation results.
[0018] For the first data block x1(n) and the last data block x T (n), its FFT transformation result only appears in one combined FFT transformation. Therefore, the information of these two data blocks is still lost, and the partial zero padding method is used to compensate for this part of the loss.
[0019] Before the first data block x1(n) and before the last data block x T (n) After that, add a sequence of all zeros with a length of half a data block, and the data sequence after zero filling is x(n) z The form is:
[0020] x(n) z =[zeros(n / 2),x1(n),x2(n),...,x T(n),zeros(n / 2)]
[0021] At this point, 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 data block x1(n) appears in two combined FFT transforms. r The FFT transformation result of (n) will also appear in the two combined FFT transformations, namely the last combined FFT transformation and the penultimate combined FFT transformation, to achieve overlapping coverage of the entire data segment.
[0026] After the overlapping link, a total of r merged FFT transformation results are obtained, and the length of each merged FFT transformation is two data blocks long. Cut off the n / 2 sequences at both ends of the equalization result to remove the crosstalk between data blocks while ensuring that the FFT transformation output length is still n, that is, the time domain response x of the single carrier frequency domain MMSE equalization output result is obtained. mmse (n).
[0027] Step 3: Use a low-complexity algorithm based on statistics to estimate the signal-to-noise ratio (SNR) and obtain the estimated value of the signal-to-noise ratio (SNR) and the estimated value of the signal-to-noise ratio (SCNR) after adding crosstalk.
[0028] Step 3 The specific implementation method is:
[0029] The signal-to-noise ratio estimation is based on a low-complexity algorithm based on statistics to estimate the received signal strength and noise interference strength. The received signal energy S and noise energy N are used to represent the second-order moment and fourth-order moment of the received signal, and the signal-to-noise ratio is further obtained by simultaneous equations.
[0030] make The second-order moment L2 of the received signal R(K) is expressed as:
[0031] L2=E{r(n)r*(n)}=E{|b(n)| 2}+E{b(n)v * (n)}+E{b(n) * v(n)}+E{|v(n)|2}
[0032] =S+N
[0033] Where: x(n) is the effective signal sent, h(n) is the channel time domain response, r(n) is the received signal, v(n) is the noise time domain response, ( represents circular convolution).
[0034] The fourth-order moment L4 of the received signal R(K) is expressed as:
[0035] 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
[0036] Combining the second-order moment expression with the fourth-order moment expression, we can get the signal-to-noise ratio estimate SNR:
[0037]
[0038] After adding crosstalk, the estimated signal-to-noise ratio SCNR is:
[0039]
[0040] Where: IBI represents the inter-block interference variance.
[0041] 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 calculate the time domain response x of the single-carrier frequency domain MMSE equalization output result. mmse(n) is judged and output. When the received signal-to-noise ratio is lower than the received signal-to-noise ratio threshold, the residual inter-symbol interference is iteratively eliminated, and x(n) is transformed by FFT to obtain the frequency domain estimated value X(K) of the transmitted signal, and it is determined that the estimated value signal is the real transmitted signal X(K). The estimated value I(K) of the residual inter-symbol interference is obtained through the signal, and the estimated value is used to equalize the single-carrier frequency domain MMSE equalization output signal X MMSE (K) compensation, and the iterative output signal X that eliminates the residual inter-symbol interference is obtained. DFE (K), the frequency domain response is transformed by IFFT and a decision is made to obtain an output for iteratively eliminating residual inter-symbol interference, that is, to achieve adaptive frequency domain equalization based on decision feedback.
[0042] Step 4 The specific implementation method is:
[0043] When the received signal-to-noise ratio is higher than the received signal-to-noise ratio threshold, the frequency domain form of the residual inter-symbol interference I(K) is:
[0044]
[0045] At this time, the accurate transmitted signal X(K) cannot be known in advance at the receiving end, and the estimated value of I(K) is solved through iteration.
[0046] The first equalization result output by the non-iterative frequency domain algorithm, i.e., the first estimated value X0(K) of the transmitted 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 estimated value is used to estimate the residual inter-symbol interference.
[0047]
[0048] The estimated value of residual inter-symbol interference I1(K) is used as the real residual inter-symbol interference value to compensate the first estimated value X0(K) of the transmitted signal, that is, the equalized output X after one iteration can be obtained. DFE-1 (K):
[0049] X DFE-1 (K) = X0(K) - I1(K)
[0050] Since there is still an error between the first estimated value X0(K) of the transmitted signal and the actual transmitted signal X(K), the estimated value I1(K) of the residual inter-symbol interference is different from the first estimated value X0(K) of the transmitted signal. MMSE The actual residual ISI in (K) also has errors. The frequency domain response X after one iteration of eliminating the residual ISI is DFE-1(K) still has an error with the actual transmitted data. The cause of the bit error is the inaccuracy of the estimated value I1(K) of the residual inter-symbol interference, and the reason for the inaccuracy is the error between the feedback value fed back to the system and the actual transmitted signal, that is, the error between the signal X0(K) output by the non-iterative frequency domain equalization algorithm and the actual transmitted signal: Therefore, reducing this error reduces the output error, and the frequency domain response X after an iterative elimination of the residual inter-symbol interference DFE-1 (K) Compared with the signal X0(K) output by the non-iterative frequency domain equalization algorithm, the error with the real transmitted signal is smaller, and the frequency domain response X after one iteration of eliminating the residual inter-symbol interference is selected. DFE-1 (K) as feedback input and using X DFE-1 (K) to obtain a more accurate estimate of the residual inter-symbol interference I2(K):
[0051]
[0052] Using this estimated value to compensate the non-iterative frequency domain equalization output signal X0(K), a more accurate output value X can be obtained. DFE-2 (K), that is, the output value after the residual inter-symbol interference is eliminated by secondary iteration:
[0053] X DFE-2 (K) = X DFE-1 (K)-I1(K)
[0054] According to the output value X after elimination DFE-2 (K) Implement adaptive frequency domain equalization based on decision feedback.
[0055] The invention discloses an adaptive frequency domain equalization system based on decision feedback, which is used to implement an adaptive frequency domain equalization method based on decision feedback. The adaptive frequency domain equalization system based on decision feedback comprises a single carrier frequency domain MMSE unit, a signal-to-noise ratio judgment unit, an adaptive iteration unit and an overlap shearing unit.
[0056] The single-carrier frequency-domain MMSE unit introduces the mean square value of crosstalk between data blocks into the received signal-to-noise ratio detection, uses the pilot segment data to simultaneously estimate the channel estimation result and the mean square value of noise, uses these two estimated values to generate a frequency-domain MMSE equalizer according to the minimum mean square error criterion, and then uses the equalizer to compensate for the valid data segment.
[0057] The signal-to-crosstalk-noise ratio judgment unit is used to calculate the signal-to-crosstalk-plus-noise ratio and judge whether the signal-to-crosstalk-plus-noise ratio is within the set indicator allowable range, and input the judgment result and the noise ratio parameter into the adaptive iteration unit to reduce the difficulty of algorithm implementation.
[0058] The adaptive iteration unit introduces the mean square value of crosstalk between data blocks into the signal-to-noise ratio estimation formula to obtain an estimation method for the parameter "signal-to-noise ratio" that simultaneously considers the noise interference intensity and channel fading conditions, and selects whether to perform iterative equalization based on the estimation method judgment result and the noise ratio parameter.
[0059] The overlapping and cutting unit divides the input signal into a plurality of data blocks, then overlaps the data blocks, and finally cuts and merges them, so as to further eliminate crosstalk between the data blocks.
[0060] Beneficial effects:
[0061] 1. Aiming at the problem of crosstalk between data blocks in the traditional single-carrier frequency-domain MMSE equalization system, the present invention discloses an adaptive frequency-domain equalization method and system based on decision feedback, designs a frequency-domain equalizer based on the minimum error criterion that introduces crosstalk between data blocks, channel noise, and inter-code crosstalk, and uses overlapping shearing to complete frequency-domain equalization, thereby further eliminating crosstalk between data blocks.
[0062] 2. The present invention discloses an adaptive frequency domain equalization method and system based on decision feedback, which uses an iterative method to eliminate the residual inter-code interference of the traditional frequency domain MMSE equalization system. At the same time, the mean square value of the crosstalk between data blocks is introduced into the signal-to-noise ratio estimation formula to obtain an estimation method for the parameter "signal-to-noise ratio" that takes into account both the noise interference intensity and the channel fading. The signal-to-noise ratio estimation value is used to adaptively adjust whether a more complex iterative elimination calculation of the residual inter-code interference is required, thereby improving the efficiency of adaptive frequency domain equalization.
[0063] 3. The present invention discloses an adaptive frequency domain equalization method and system based on decision feedback, which introduces the mean square value of crosstalk between data blocks into the detection of received signal-to-noise ratio, and derives the form of residual inter-code crosstalk in a low-complexity frequency domain equalization system based on an overlapping shearing algorithm; a signal-to-noise ratio estimation algorithm based on statistics obtains a dynamic received signal-to-noise ratio estimation value, and constructs a signal-to-noise ratio index that simultaneously considers channel noise interference and crosstalk between data blocks; when the signal-to-noise ratio estimate is high, there is no need to iteratively eliminate the residual inter-code crosstalk, and the communication index requirements under the environment can be met; and when the signal-to-noise ratio estimate is low, the bit error rate performance of the single-carrier frequency domain MMSE equalization system is improved by iteratively eliminating the residual inter-code crosstalk, that is, adaptive frequency domain equalization based on decision feedback is realized. BRIEF DESCRIPTION OF THE DRAWINGS
[0064] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0065] Figure 1 This is a flow chart of the adaptive frequency domain equalization method based on decision feedback of the present invention.
[0066] Figure 2 The frame structure diagram is applicable to the adaptive frequency domain equalization method based on decision feedback of the present invention. DETAILED DESCRIPTION
[0067] The present invention is further described and described in detail below with reference to the accompanying drawings and embodiments.
[0068] like Figure 1 As shown, the present embodiment discloses an adaptive frequency domain equalization method based on decision feedback, and the specific implementation steps are as follows:
[0069] Step 1: Refer to the frame structure shown in the figure, separate the pilot segment from the data, and divide the data segment x(n) to be processed into 16 data blocks of equal length x1(n), x2(n), ...x r (n) Adjacent data blocks are combined and FFT transformed.
[0070] Step 2: Perform FFT transformation on the pilot band data to obtain the frequency domain channel estimation result and noise mean square value estimation to generate the MMSE equalizer. Use this equalizer to perform channel compensation on the data after 15 combined FFT transformations to obtain the frequency domain response X of the MMSE equalization output result. MMSE (K), the compensated frequency domain signal is transformed into the time domain by block IFFT, and the time domain response x of the single carrier frequency domain MMSE equalization output result is obtained by cutting and merging. mmse (n).
[0071] Step 2 The specific implementation method is:
[0072] Generate MMSE equalizer, the frequency domain response C(K) of MMSE equalizer is
[0073]
[0074] in: is the noise variance, is the average power of the transmitted signal, and H(K) is the channel frequency domain response.
[0075] The equalizer is used to perform channel compensation on the data after 15 combined FFT transformations to obtain the frequency domain response X of the MMSE equalization output result. MMSE (K) Perform block IFFT transformation on the compensated frequency domain signal to the time domain.
[0076] Then, overlap and overwrite to divide the data segment x(n) to be processed into 16 data blocks of length 256.
[0077] x(n)=[x1(n),x2(n),...,x 16 (n)]
[0078] Starting from the first data block x1(n), merge the two data blocks together for FFT transformation
[0079] X t-1,t (K) = FFT(x1(n),x2(n))
[0080] A total of 15 merges and FFT transformations are performed to complete the FFT transformation of the entire data segment. t (n), its FFT transformation result appears simultaneously in the t-th combined FFT transformation result and the t-1-th combined FFT transformation result, completing the overlapping coverage of the FFT transformation results.
[0081] For the first data block x1(n) and the last data block x 16 (n), its FFT transformation result only appears in one combined FFT transformation. Therefore, the information of these two data blocks is still lost, and the partial zero padding method is used to compensate for this part of the loss.
[0082] Before the first data block x1(n) and before the last data block x 16 (n) After that, add a sequence of all zeros with a length of half a data block, and the data sequence after zero filling is x(n) z The form is:
[0083] x(n) z =[zeros(n / 2),x1(n),x2(n),...,x T (n),zeros(n / 2)]
[0084] At this point, the sequence of the first combined FFT transform becomes:
[0085] X 1,2 (K) z =FFT(zeros(1,n / 2),x1(n),x2(1,n / 2))
[0086] The sequence of the second combined FFT transform becomes:
[0087] X 2,3 (K) z =FFT(x1(n / 2+1,n),x2(n),x3(1,n / 2))
[0088] At this time, the FFT transform result of data block x1(n) appears in two combined FFT transforms.r The FFT transformation result of (n) will also appear in the two combined FFT transformations, namely the last combined FFT transformation and the penultimate combined FFT transformation, to achieve overlapping coverage of the entire data segment.
[0089] After the overlapping link, a total of 16 merged FFT transformation results are obtained, and the length of each merged FFT transformation is two data blocks long. The n / 2 sequences at both ends of the equalization result are cut off to remove the crosstalk between data blocks while ensuring that the FFT transformation output length is still 256.
[0090] Step 3: Use a low-complexity algorithm based on statistics to estimate the signal-to-noise ratio (SNR) and obtain the estimated value of the signal-to-noise ratio (SNR) and the estimated value of the signal-to-noise ratio (SCNR) after adding crosstalk.
[0091] Step 3 The specific implementation method is:
[0092] The signal-to-noise ratio estimation is based on a low-complexity algorithm based on statistics to estimate the received signal strength and noise interference strength. The received signal energy S and noise energy N are used to represent the second-order moment and fourth-order moment of the received signal, and the signal-to-noise ratio is further obtained by simultaneous equations.
[0093] make The second-order moment L2 of the received signal R(K) is expressed as:
[0094] L2=E{r(n)r * (n)}=E{|b(n)| 2}+E{b(n)v * (n)}+E{b(n) * v(n)}+E{|v(n)| 2}
[0095] =S+N
[0096] Where: x(n) is the effective signal sent, h(n) is the channel time domain response, r(n) is the received signal, v(n) is the noise time domain response, ( represents circular convolution).
[0097] The fourth-order moment L4 of the received signal R(K) is expressed as:
[0098] 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
[0099] Combining the second-order moment expression with the fourth-order moment expression, we can get the signal-to-noise ratio estimate SNR:
[0100]
[0101] After adding crosstalk, the estimated signal-to-noise ratio SCNR is:
[0102]
[0103] Where: IBI represents the inter-block interference variance.
[0104] 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 calculate the time domain response x of the single-carrier frequency domain MMSE equalization output result. mmse (n) is judged and output. When the received signal-to-noise ratio is lower than the received signal-to-noise ratio threshold, the residual inter-symbol interference is iteratively eliminated, and x(n) is transformed by FFT to obtain the frequency domain estimated value X(K) of the transmitted signal, and it is determined that the estimated value signal is the real transmitted signal X(K). The estimated value I(K) of the residual inter-symbol interference is obtained through the signal, and the estimated value is used to equalize the single-carrier frequency domain MMSE equalization output signal X MMSE (K) compensation, and the iterative output signal X that eliminates the residual inter-symbol interference is obtained. DFE (K), the frequency domain response is transformed by IFFT and a decision is made to obtain an output for iteratively eliminating residual inter-symbol interference, that is, to achieve adaptive frequency domain equalization based on decision feedback.
[0105] Step 4 The specific implementation method is:
[0106] When the received signal-to-noise ratio is higher than the received signal-to-noise ratio threshold, the frequency domain form of the residual inter-symbol interference I(K) is:
[0107]
[0108] The accurate transmitted signal X(K) cannot be known in advance at the receiving end, and the estimated value of I(K) is solved by iteration.
[0109] The first equalization result output by the non-iterative frequency domain algorithm, i.e., the first estimated value X0(K) of the transmitted 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 estimated value is used to estimate the residual inter-symbol interference.
[0110]
[0111] The estimated value of residual inter-symbol interference I1(K) is used as the real residual inter-symbol interference value to compensate the first estimated value X0(K) of the transmitted signal, that is, the equalized output X after one iteration can be obtained. DFE-1 (K):
[0112] X DFE-1 (K) = X0(K) - I1(K)
[0113] Since there is still an error between the first estimated value X0(K) of the transmitted signal and the actual transmitted signal X(K), the estimated value I1(K) of the residual inter-symbol interference is different from the first estimated value X0(K) of the transmitted signal. MMSE The actual residual ISI in (K) also has errors. The frequency domain response X after one iteration of eliminating the residual ISI is DFE-1 (K) still has an error with the actual transmitted data. The cause of the bit error is the inaccuracy of the estimated value I1(K) of the residual inter-symbol interference, and the reason for the inaccuracy is the error between the feedback value fed back to the system and the actual transmitted signal, that is, the error between the signal X0(K) output by the non-iterative frequency domain equalization algorithm and the actual transmitted signal: Therefore, reducing this error reduces the output error, and the frequency domain response X after an iterative elimination of the residual inter-symbol interference DFE-1 (K) Compared with the signal X0(K) output by the non-iterative frequency domain equalization algorithm, the error with the real transmitted signal is smaller, and the frequency domain response X after one iteration of eliminating the residual inter-symbol interference is selected. DFE-1 (K) as feedback input and using X DFE-1 (K) to obtain a more accurate estimate of the residual inter-symbol interference I2(K):
[0114]
[0115] Using this estimated value to compensate the non-iterative frequency domain equalization output signal X0(K), a more accurate output value X can be obtained. DFE-2 (K), that is, the output value after the residual inter-symbol interference is eliminated by secondary iteration:
[0116] XDFE-2 (K) = X DFE-1 (K)-I1(K)
[0117] According to the output value X after elimination DFE-2 (K) Implement adaptive frequency domain equalization based on decision feedback.
[0118] The present embodiment discloses an adaptive frequency domain equalization system based on decision feedback, which is used to implement an adaptive frequency domain equalization method based on decision feedback. The adaptive frequency domain equalization system based on decision feedback includes a single carrier frequency domain MMSE unit, a signal-to-noise ratio judgment unit, an adaptive iteration unit, and an overlap shearing unit.
[0119] The single-carrier frequency-domain MMSE unit introduces the mean square value of crosstalk between data blocks into the received signal-to-noise ratio detection, uses the pilot segment data to simultaneously estimate the channel estimation result and the mean square value of noise, uses these two estimated values to generate a frequency-domain MMSE equalizer according to the minimum mean square error criterion, and then uses the equalizer to compensate for the valid data segment.
[0120] The signal-to-crosstalk-noise ratio judgment unit is used to calculate the signal-to-crosstalk-plus-noise ratio and judge whether the signal-to-crosstalk-plus-noise ratio is within the set indicator allowable range, and input the judgment result and the noise ratio parameter into the adaptive iteration unit to reduce the difficulty of algorithm implementation.
[0121] The adaptive iteration unit introduces the mean square value of crosstalk between data blocks into the signal-to-noise ratio estimation formula to obtain an estimation method for the parameter "signal-to-noise ratio" that simultaneously considers the noise interference intensity and channel fading conditions, and selects whether to perform iterative equalization based on the estimation method judgment result and the noise ratio parameter.
[0122] The overlapping and cutting unit divides the input signal into a plurality of data blocks, then overlaps the data blocks, and finally cuts and merges them, so as to further eliminate crosstalk between the data blocks.
[0123] The specific description above further illustrates the purpose, technical solutions and beneficial effects of the invention in detail. It should be understood that the above is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. An adaptive frequency domain equalization method based on decision feedback, characterized in that: The following steps are included: Step 1: Separate the pilot segment from the data and divide the data segment x(n) to be processed into r data blocks of equal length x1(n), x2(n), ...x r (n), perform a combined FFT transform on adjacent data blocks; Step 2: Perform FFT transformation on the pilot band data to obtain the frequency domain channel estimation result and the noise mean square value estimation to generate the MMSE equalizer; use this equalizer to perform channel compensation on the data after r-1 times of combined FFT transformation to obtain the frequency domain response X of the MMSE equalization output result MMSE (K), the compensated frequency domain signal is transformed into the time domain by block IFFT, and the time domain response x of the single carrier frequency domain MMSE equalization output result is obtained by cutting and merging. mmse (n); Step 3: Use a low-complexity algorithm based on statistics to estimate the signal-to-noise ratio (SNR) and obtain the estimated value of the signal-to-noise ratio (SNR) and the estimated value of the signal-to-noise ratio (SCNR) after adding crosstalk. 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 calculate the time domain response x of the single-carrier frequency domain MMSE equalization output result. mmse (n) Making a judgment and outputting it; When the received signal-to-noise ratio is lower than the received signal-to-noise ratio threshold, the residual inter-symbol interference is iteratively eliminated, and x(n) is transformed by FFT to obtain the frequency domain estimated value X(K) of the transmitted signal, and it is determined that the estimated value signal is the real transmitted signal X(K). The estimated value I(K) of the residual inter-symbol interference is obtained through the signal, and the estimated value is used to equalize the single-carrier frequency domain MMSE equalization output signal X MMSE (K) compensation, and the iterative output signal X that eliminates the residual inter-symbol interference is obtained. DFE (K), the frequency domain response is transformed by IFFT and a decision is made to obtain an output for iteratively eliminating residual inter-symbol interference, that is, to achieve adaptive frequency domain equalization based on decision feedback.
2. The adaptive frequency domain equalization method based on decision feedback according to claim 1, characterized in that: The specific implementation method of Step2 is: Generate MMSE equalizer, the frequency domain response C(K) of MMSE equalizer is in: is the noise variance, is the average power of the transmitted signal, H(K) is the channel frequency domain response; This equalizer is used to perform channel compensation on the data after r-1 times of combined FFT transformation to obtain the frequency domain response X of the MMSE equalization output result. MMSE (K), performing block IFFT transformation on the compensated frequency domain signal to the time domain; Overlapping is performed to divide the data segment x(n) to be processed into r data blocks of the same length N. x(n)=[x1(n),x2(n),…,x T (n)] Starting from the first data block x1(n), merge the two data blocks together for FFT transformation X t-1,t (K)=FFT(x1(n),x2(n)) A total of r-1 merging and FFT transformations are performed to complete the FFT transformation of the entire data segment; for a data block x t (n), its FFT transformation result appears simultaneously in the t-th combined FFT transformation result and the t-1-th combined FFT transformation result, completing the overlapping coverage of the FFT transformation results; For the first data block x1(n) and the last data block x T (n), its FFT transformation result only appears in one combined FFT transformation; therefore, the information of these two data blocks is still lost, and the partial zero padding method is used to compensate for this part of the loss; Before the first data block x1(n) and before the last data block x T (n) After that, add a sequence of all zeros with a length of half a data block, and the data sequence after zero filling is x(n) z The form is: x(n) z =[zeros(n / 2),x1(n),x2(n),...,x T (n),zeros(n / 2)] At this point, the sequence of the first combined FFT transform becomes: X 1,2 (K) z =FFT(zeros(1,n / 2),x1(n),x2(1,n / 2)) The sequence of the second combined FFT transform 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 in two combined FFT transforms; the last data block x r The FFT transformation result of (n) will also appear in the two combined FFT transformations, namely the last combined FFT transformation and the penultimate combined FFT transformation, to achieve overlapping coverage of the entire data segment; After the overlapping link, a total of r merged FFT transformation results are obtained, and the length of each merged FFT transformation is two data blocks long; the n / 2 sequences at both ends of the equalization result are cut off to remove the crosstalk between data blocks while ensuring that the FFT transformation output length is still n, that is, the time domain response x of the single carrier frequency domain MMSE equalization output result is obtained. mmse (n).
3. The adaptive frequency domain equalization method based on decision feedback according to claim 2, characterized in that: Step 3 The specific implementation method is: The signal-to-noise ratio estimation estimates the received signal strength and noise interference strength based on a low-complexity algorithm based on statistics; the received signal energy S and the noise energy N are used to represent the second-order moment and the fourth-order moment of the received signal, and the signal-to-noise ratio is further obtained by simultaneous equations; make The second-order 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 effective signal sent, h(n) is the channel time domain response, r(n) is the received signal, v(n) is the noise time domain response, ( represents circular convolution); The fourth-order moment L4 of the received signal R(K) is expressed 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 Combining the second-order moment expression with the fourth-order moment expression, we can get the signal-to-noise ratio estimate SNR: After adding crosstalk, the estimated signal-to-noise ratio SCNR is: Where: IBI represents the inter-block interference variance.
4. The adaptive frequency domain equalization method based on decision feedback according to claim 3, characterized in that: Step 4 The specific implementation method is: When the received signal-to-noise ratio is higher than the received signal-to-noise ratio threshold, the frequency domain form of the residual inter-symbol interference I(K) is: The accurate transmitted signal X(K) cannot be known in advance at the receiving end, and the estimated value of I(K) is solved by iteration; The first equalization result output by the non-iterative frequency domain algorithm is the first estimated value X0(K) of the transmitted signal; the signal is used as the transmitted signal to feed back to the single carrier frequency domain equalization system; after the single carrier frequency domain equalization system obtains the feedback input, the estimated value is used to estimate the residual inter-symbol interference. The estimated value of residual inter-symbol interference I1(K) is used as the real residual inter-symbol interference value to compensate the first estimated value X0(K) of the transmitted signal, that is, the equalized output X after one iteration can be obtained. DFE-1 (K): X DFE-1 (K)=X0(K)-I1(K) Since there is still an error between the first estimated value X0(K) of the transmitted signal and the actual transmitted signal X(K), the estimated value I1(K) of the residual inter-symbol interference is different from the first estimated value X0(K) of the transmitted signal. MMSE The actual residual ISI in (K) also has errors; the frequency domain response X after one iteration of eliminating the residual ISI DFE-1 (K) still has an error with the actual transmitted data; the frequency domain response X after one iteration of eliminating the residual inter-symbol interference DFE-1 (K) Compared with the signal X0(K) output by the non-iterative frequency domain equalization algorithm, the error with the real transmitted signal is smaller, and the frequency domain response X after one iteration of eliminating the residual inter-symbol interference is selected. DFE-1 (K) as feedback input and using X DFE-1 (K) to obtain a more accurate estimate of the residual inter-symbol interference I2(K): Using this estimated value to compensate the non-iterative frequency domain equalization output signal X0(K), a more accurate output value X can be obtained. DFE-2 (K), that is, the output value after the residual inter-symbol interference is eliminated by secondary iteration: X DFE-2 (K)=X DFE-1 (K)-I1(K) According to the output value X after elimination DFE-2 (K) Implement adaptive frequency domain equalization based on decision feedback.
5. A decision feedback based adaptive frequency domain equalization system, used to implement a decision feedback based adaptive frequency domain equalization method as claimed in claim 1, 2, 3 or 4, characterized in that: It includes a single carrier frequency domain MMSE unit, a signal-to-noise ratio judgment unit, an adaptive iteration unit and an overlap cutting unit; The single-carrier frequency-domain MMSE unit introduces the mean square value of crosstalk between data blocks into the received signal-to-noise ratio detection, uses the pilot segment data to simultaneously estimate the channel estimation result and the mean square value of noise, uses the two estimated values to generate a frequency-domain MMSE equalizer according to the minimum mean square error criterion, and then uses the equalizer to compensate the valid data segment; The signal-to-crosstalk-to-noise ratio judgment unit is used to calculate the signal-to-crosstalk-to-noise ratio, and judge whether the signal-to-crosstalk-to-noise ratio is within the set index allowable range, and input the judgment result and the noise ratio parameter into the adaptive iteration unit; The adaptive iteration unit introduces the mean square value of crosstalk between data blocks into the signal-to-noise ratio estimation formula to obtain an estimation method for the parameter "signal-to-noise ratio" that takes into account both the noise interference intensity and the channel fading conditions, and selects whether to perform iterative equalization based on the estimation method judgment result and the noise ratio parameter; The overlapping and cutting unit divides the input signal into a plurality of data blocks, then overlaps the data blocks, and finally cuts and merges them, so as to further eliminate crosstalk between the data blocks.
Citation Information
Patent Citations
Single carrier frequency domain equalization method for shortwave communication channel
CN103986676A
Decision feedback equalization processing method and equalizer
CN106302282A
Transform domain quadratic estimation method uniting weighted threshold de-noising and balanced decision
CN106789764A
