A method for estimating large frequency deviation of carrier in burst short frame communication
By combining DFT and signal autocorrelation function, the frequency deviation estimation is performed using multi-stage pilot bands, which solves the problem of small frequency deviation estimation range in the prior art, and achieves a larger frequency deviation search range and higher estimation accuracy. Improve the performance of the controller.
Patent Information
- Application Number
- CN202410036560.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-10
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-01-10
AI Technical Summary
In the prior art, the frequency deviation estimation range is small, making it difficult to adapt to the large Doppler frequency deviation, affecting the performance of the demodulator.
Combining the DFT-based frequency deviation estimation method and the signal autocorrelation function method, a suitable frame structure is selected, and frequency deviation estimation is performed through multi-level pilot bands, increasing the frequency deviation search range and improving the estimation accuracy.
The frequency offset search range is expanded and the estimation accuracy is improved, the adaptability to the majority Doppler frequency offset is enhanced, and the performance of the understanding controller is improved.
Smart Images

Figure CN117880043B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to frequency offset estimation, and in particular to a method for estimating large frequency offset of a burst short frame communication carrier. Background Art
[0002] Burst data communication is widely used in various communication fields. Due to factors such as Doppler frequency shift and clock accuracy, there is a certain deviation between the carrier of the received signal and the local carrier. The existence of this frequency difference seriously affects the performance of the demodulator, so carrier synchronization must be performed at the receiving end.
[0003] According to whether the training sequence is used, it can be divided into data-assisted synchronization method and non-data-assisted synchronization method. Burst data communication generally uses data-assisted synchronization method, and a set of known training sequences is placed at the head of each burst data frame for carrier synchronization, which has the advantages of high estimation accuracy, simple calculation, and large capture range.
[0004] In burst communication, traditional data-aided frequency offset estimation algorithms such as Kay algorithm, Fitz algorithm and L&R algorithm can obtain accurate frequency offset estimation, but the frequency offset estimation range is small. Therefore, it is of great significance to study how to improve the data-aided frequency offset estimation range.
[0005] At present, frequency offset estimation methods are mainly divided into two categories: estimation methods based on time domain and estimation methods based on frequency domain. Estimation methods based on time domain can be summarized into two forms: methods based on adjacent received signals and methods based on the autocorrelation function of received signals. The estimation range of the method based on adjacent received signals is large, but the estimation accuracy is low; the method based on the autocorrelation function of received signals has high accuracy, but the estimation range is small. The estimation method based on frequency domain refers to the estimation method based on DFT. This method has a large estimation range, but the estimation accuracy is related to the length of the data used. Summary of the invention
[0006] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a method for estimating large frequency offset of a burst short frame communication carrier, which combines the advantages of a large estimation range of an estimation method based on DFT and high accuracy of an estimation method based on signal autocorrelation function, combines these two methods and selects a suitable frame structure, while taking into account both the estimation range and the estimation accuracy.
[0007] The object of the present invention is to achieve the following technical solution: a method for estimating large frequency deviation of a burst short frame communication carrier, comprising the following steps:
[0008] S1. A burst short-time frame structure including a pre-pilot and its cyclic prefix, data and a post-pilot is given;
[0009] S2. Determine the oversampling rate of the signal according to the frequency deviation search range of the down-converted and demodulated data stream, and send the over-sampled data to the DFT transform sampling point. Sliding windows;
[0010] S3. When the amplitude peak exceeds the capture threshold Th, the peak position is recorded and the first rough frequency deviation estimate is calculated. , perform frequency offset correction on the frequency offset estimation value;
[0011] S4. Perform a cross-correlation operation on the frame data after the first coarse frequency offset correction with the locally known pilot sequence, find the position with the largest result in the cross-correlation operation as the frame header position, and obtain the position of the front and rear pilots and their cyclic sequences according to the known frame structure;
[0012] S5. Perform conjugate correlation operation on the pilot signal and its cyclic prefix to obtain the phase offset and calculate the phase angle, and finally calculate the second rough frequency deviation estimate. ;
[0013] S6. Perform conjugate correlation operation on the front pilot and the rear pilot to obtain the phase offset and calculate the angle, and finally calculate the rough estimate of the precise frequency deviation .
[0014] The beneficial effects of the present invention are:
[0015] (1) Large frequency offset search range: Using DFT+pilot for frequency offset estimation can increase the frequency offset search range and has greater Doppler frequency offset adaptability than traditional loops;
[0016] (2) High frequency offset estimation accuracy. Frequency offset estimation is performed through multi-level pilot bands, with high frequency offset estimation accuracy;
[0017] (3) Low implementation complexity and stable performance. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] Figure 1 It is a schematic diagram of the short frame data structure;
[0019] Figure 2 It is a schematic diagram of the multi-stage carrier frequency offset estimation process;
[0020] Figure 3 This is the demodulated constellation diagram without frequency offset estimation;
[0021] Figure 4 The demodulated constellation diagram after frequency offset estimation and correction;
[0022] Figure 5 It is a schematic diagram of the frequency estimation range curve;
[0023] Figure 6 This is a schematic diagram of the frequency estimation error curve. DETAILED DESCRIPTION
[0024] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings, but the protection scope of the present invention is not limited to the following.
[0025] The present invention takes into account that when using DFT for frequency offset estimation, the range of frequency offset estimation is large, but it is easily affected by the length of the data used, and its estimation accuracy is related to the length of the data used. The length of burst short-time frame data is limited, which limits the estimation accuracy of the DFT method.
[0026] When using the autocorrelation function of the demodulated signal for frequency offset estimation, using different time delays will result in different estimation performances: when using the long-delay autocorrelation function for frequency offset estimation, the anti-noise performance is stronger and the frequency offset estimation accuracy is higher, but the estimation range is related to the time delay length. The longer the time delay, the narrower the range and the more serious the phase folding phenomenon; when using the short-delay autocorrelation function for frequency offset estimation, there is a larger estimation range, but the estimation accuracy is lower and is more affected by noise. Therefore, combining the characteristics of the large DFT frequency offset estimation range and the high estimation accuracy of the autocorrelation function, a multi-level autocorrelation function is used for the frame structure of burst short-time frame data, that is, a multi-level frequency offset determination method combining the short-delay autocorrelation function with the long-delay autocorrelation function. First, a coarse DFT frequency offset estimation is performed, and after compensation, the frequency offset falls into the estimation range of the short-delay autocorrelation function, and then after compensation, the frequency offset falls into the estimation range of the long-delay autocorrelation function, which not only expands the frequency offset estimation range but also improves the frequency offset estimation accuracy. Specifically:
[0027] like Figure 1 As shown, a method for estimating large frequency deviation of a burst short frame communication carrier comprises the following steps:
[0028] S1. A burst short-time frame structure including a pre-pilot and its cyclic prefix, data and a post-pilot is given;
[0029] The burst short-time frame structure is as follows Figure 1 As shown, it includes three parts: the front pilot and its cyclic prefix, data and the rear pilot. The frame header and frame tail are the front pilot and its cyclic prefix and the rear pilot respectively. First, insert a section of pilot and its cyclic prefix at the front and rear positions of the data frame respectively, where the time interval between the front pilot and the rear pilot is recorded as T2, T2 is determined according to the data length loaded by the short frame, and the time interval between the front / rear pilot and its cyclic prefix is recorded as T1, T1 is determined according to the required carrier frequency offset estimation range. The relationship between the time interval and the sampling point interval is, , ,in , is the sampling time interval, is the sampling frequency.
[0030] S2. Determine the oversampling rate of the signal according to the frequency deviation search range of the down-converted and demodulated data stream, and send the over-sampled data to the DFT transform sampling point. Sliding windows, It needs to be greater than or equal to the length P of the front / rear pilot in the frame structure; generally , and search for the transformed amplitude peak;
[0031] S3. When the amplitude peak exceeds the capture threshold Th, the peak position is recorded and the first rough frequency deviation estimate is calculated. , perform frequency offset correction on the frequency offset estimation value;
[0032] S4. Perform a cross-correlation operation on the frame data after the first coarse frequency offset correction with the locally known pilot sequence, find the position with the largest result in the cross-correlation operation as the frame header position, and obtain the position of the front and rear pilots and their cyclic sequences according to the known frame structure;
[0033] S5. Perform conjugate correlation operation on the pilot signal and its cyclic prefix to obtain the phase offset and calculate the phase angle, and finally calculate the second rough frequency deviation estimate. ;
[0034] S6. Perform conjugate correlation operation on the front pilot and the rear pilot to obtain the phase offset and calculate the angle, and finally calculate the rough estimate of the precise frequency deviation .
[0035] In order to improve the frequency offset estimation range and estimation accuracy, we use multi-stage carrier frequency offset estimation. The specific implementation steps are as follows:
[0036] 1. Set the sliding window size according to the set capture DFT point number N, and set the signal oversampling rate according to the frequency deviation search range , the oversampled data is sent to the sliding window, and the data in the window is subjected to DFT transformation. The maximum amplitude peak after transformation is compared with the preset threshold Th. When it exceeds the set threshold, it indicates that the burst signal has arrived, and the starting position S of the burst signal is recorded to complete the burst frame capture;
[0037] 2. From burst frame The starting position S is intercepted Sample data is transformed by DFT and the frequency estimation value corresponding to the amplitude peak is determined. The first rough estimation of carrier frequency offset is performed. The accuracy of the first rough estimation is The result of the first rough estimate is recorded as , and then the first rough estimation result is compared with the original data frame Compensate and record the compensated frame data as ; ;
[0038] 3. The compensated frame data It is cross-correlated with the local frame synchronization signal, and the arrival of a frame signal is detected according to the maximum amplitude, and then the starting point of a frame signal is located according to the maximum position. According to the starting position and the known frame structure information, the two pilots and their cyclic prefixes can be extracted. The extracted two pilots and the pre-pilot cyclic prefix are recorded as , , ;
[0039] 4. The compensated frame data The pre-pilot and its cyclic prefix are cross-correlated, and the cross-correlation result is used to make a second rough estimation of the carrier frequency offset. The result of the second rough estimation is recorded as , and then the second rough estimation result is compared with the original frame data Compensate and record the compensated frame data as ; ;
[0040] 5. The compensated frame data The front and rear pilots are cross-correlated, and the cross-correlation results are used to accurately estimate the carrier frequency offset to obtain the final carrier frequency offset estimate , and then the precise estimation result is compared with the original frame data Compensate and record the compensated frame data as , It can be used for subsequent calculations ;
[0041] 6. The pre-pilot and its cyclic prefix are cross-correlated. The relative positions of the pre-pilot and its cyclic prefix need to meet the range requirements of the second coarse frequency offset estimation, that is, , that is, the error of the first rough estimate needs to be within the range of the second rough estimate, .
[0042] DFT transformation and frequency offset estimation: Under burst short frame conditions, the frequency offset range of frequency offset estimation using pilots is narrow and is not suitable for large frequency offsets. When there is a large frequency offset under communication conditions, it is necessary to use a combination of DFT and pilot frequency offset estimation. Using DFT for rough spectrum estimation can increase the frequency offset search range by increasing the sampling multiple and compress the remaining frequency offset to a narrower range, that is, within the range where the pilot can be used for frequency offset estimation.
[0043] First, determine the signal oversampling rate according to the frequency deviation search range, assuming that the symbol rate is , the estimated frequency offset range required is , then the oversampling rate is , the oversampling rate is ,according to The received signal is sampled.
[0044] Number of points of DFT transformation Set to the sliding window size, The setting is related to the signal-to-noise ratio and the Doppler change rate. The lower the signal-to-noise ratio, the higher the requirement. The larger the Doppler change rate, the higher the required The smaller the value, the more you need to consider these two parameters to determine .
[0045] The peak value of each search is compared with the capture threshold Th. The threshold value Th can be determined based on the spectrum amplitude comparison between the data and pure noise under the working signal-to-noise ratio, and the false alarm probability and the error compensation probability need to be considered comprehensively. If the spectrum peak is less than Th, the sliding window is moved to input new data; when it exceeds Th, the starting position S of the burst signal is recorded, the burst frame capture is completed, and the subsequent multi-level pilot frequency deviation estimation is started.
[0046] Assume that the frequency deviation of the converted signal is The receiving sequence Point DFT transformation can be obtained:
[0047]
[0048] Where F[.] is the discrete Fourier transform, is the sampling frequency of the discrete Fourier transform sampling points, is the symbol rate, sin(.) is the sine function, and the maximum value in the discrete Fourier transform result R(f) is taken, and its corresponding position is , then the result of the first rough estimate is , the estimated accuracy is .
[0049] Multi-stage pilot frequency offset estimation:
[0050] Assume that the preamble sequence is , , is the length of the preamble sequence, and its corresponding cyclic prefix is , , is the cyclic prefix length, usually ; The data sequence is d(n), n=0,1,2,...D-1, D is the length of the data sequence; the pilot sequence is , , is the length of the post-pilot sequence. The symbol rate is , the symbol period is , the carrier frequency used for up-conversion at the transmitter is The pre-pilot signal after up-conversion can be expressed as , , the pilot signal can be expressed as , .
[0051] Assume that due to the Doppler factor in the channel, the carrier frequency deviation of the received signal is given by becomes , then the front and rear pilot signals at the receiving end can be expressed as , ; , .
[0052] Assume that the down-conversion local oscillator frequency used by the receiver is After down-conversion at the receiving end, the baseband representation of the front-end pilot signal is:
[0053]
[0054] The baseband representation of the latter pilot signal is:
[0055]
[0056] in is the carrier frequency offset to be estimated. The pre-pilot is cross-correlated with its cyclic sequence. Since the pilot signal used is a Zadoff-Chu sequence, the modulus value of each complex symbol is 1, that is, , then the relevant results can be expressed as:
[0057]
[0058]
[0059] Then the second coarse frequency offset estimation Fine frequency offset estimation The calculation formulas are:
[0060]
[0061] In the above formula, the physical meaning of the numerator is the phase change caused by the carrier frequency offset from the cyclic prefix of the pre-pilot to the pre-pilot center moment. .
[0062]
[0063] In the above formula, the physical meaning of the numerator is the phase change caused by the carrier frequency deviation from the front pilot to the rear pilot center. , the denominator is a constant coefficient that depends on the time interval between the previous and next sequences used for calculation.
[0064] because The scope of the function is , so the corresponding estimated range of carrier frequency offset is:
[0065]
[0066]
[0067] According to the designed frame structure, , , then the above formula can be transformed into:
[0068]
[0069]
[0070] When the carrier frequency deviation exceeds this range, The function will have phase wrapping. For example, the actual phase accumulation value is , but using The result of the function is , which leads to a large carrier frequency offset estimation error. At the same time, the larger the time interval between the previous and next pilot signals, the smaller the carrier frequency offset estimation range and the higher the estimation accuracy; the smaller the time interval between the previous and next pilot signals, the larger the carrier frequency offset estimation range and the lower the estimation accuracy.
[0071] According to the previous derivation formula, after completing the first rough estimation, i.e., integer frequency offset estimation, the second rough estimation needs to complete the fractional frequency offset estimation, and the conditions for completing the fractional frequency offset estimation are:
[0072]
[0073] Right now ,but .
[0074] According to the formula, the condition for the fractional frequency deviation to be completed is that the second rough estimation range is within the accuracy of the first rough estimation range, and the first rough estimation DFT sample data value is required. , excluding the length of the preamble and its cyclic prefix, the minimum length of the designed short frame data is .
[0075] In the embodiment of the present application, the received signal is a QAM modulated signal under an AWGN channel, and the number of pre-pilot symbols for auxiliary frequency deviation is is 16, is 32, the number of valid data symbols D is 100, and the number of pilot symbols is The number of symbols in the test short frame is 32, and the symbol rate is 75MHz, and the frequency deviation is 3.75MHz. After the signal is shaped and filtered, noise and frequency deviation are added to the signal, and the object of frequency deviation estimation is obtained after matched filtering at the receiving end. Both shaping and matched filtering use root raised cosine filters with a roll-off factor of 0.35.
[0076] Figure 3 When the SNR is 10dB and the frequency offset is 1MHz, the bit error rate of the demodulated constellation diagram without frequency offset estimation is 21%.
[0077] Figure 4 When the SNR is 10dB and the frequency deviation is 1MHz, Figure 3 The same channel environment, but after frequency offset estimation and correction, the demodulated constellation diagram has a bit error rate of 0.25%.
[0078] Figure 5 In the figure, the horizontal axis is the carrier frequency deviation, from 0Hz to 15MHz, and the vertical axis is the spectrum estimation value when SNR=10dB. It can be seen that the carrier frequency deviation can be correctly estimated from this range.
[0079] Figure 6 In the figure, the horizontal axis is the SNR changing from 0dB to 20dB, the frequency deviation is 15MHz, and the vertical axis is the error value of the spectrum estimation. It can be seen that the carrier frequency deviation can be correctly estimated from this range.
[0080] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for estimating large frequency deviation of a burst short frame communication carrier, characterized in that: The following steps are involved: S1. A burst short-time frame data structure is given, which includes a pre-pilot and its cyclic prefix, data and a post-pilot; S2. Down-convert the burst short-time frame data and then oversample it, and send the oversampled data to the DFT transformation sampling point N. p Sliding windows; S3. Perform amplitude peak search on the data output by the sliding window. When the amplitude peak exceeds the capture threshold Th, record the peak position and calculate the first rough frequency deviation estimate. Using the frequency offset estimate value to perform frequency offset compensation; In step S3, the data output by the sliding window is searched for an amplitude peak value, and the amplitude peak value is compared with a preset capture threshold Th. If the amplitude peak is less than Th, the sliding window is moved to input the next sliding window data; When the peak amplitude exceeds Th, record the starting position S of the burst signal and intercept N from the starting position S. p The data of each sampling point is used to complete the burst frame capture and start the multi-level pilot frequency offset estimation; Assume that the frequency deviation of the converted signal is Δf, and perform N p Point DFT transform; set up f sym is the sampling frequency of the discrete Fourier transform sampling point, f sym =1 / T sym , T sym is the sampling period, and the maximum value in the discrete Fourier transform result is taken, and its corresponding position is i max , then the result of the first rough estimate is The estimated accuracy is f sym / N p ; The first rough estimation result is used to compensate the original data frame c0(n), and the compensated frame data is recorded as c1(n): S4. Perform a cross-correlation operation on the frame data after the first coarse frequency offset correction with the locally known pilot sequence, find out the position with the largest result in the cross-correlation operation as the frame header position, and obtain the position of the front and rear pilots and their cyclic sequence according to the known frame structure; S5. Perform conjugate correlation operation on the pilot signal and its cyclic prefix to obtain the phase offset and calculate the phase angle, and finally calculate the second rough frequency deviation estimate. And perform frequency deviation compensation; The step S5 comprises: The pilot signal and cyclic prefix of the frame data c1(n) after the first rough frequency offset estimation compensation are cross-correlated, and the result of the cross-correlation is used to make a second rough estimation of the carrier frequency offset. The result of the second rough estimation is recorded as The second rough estimation result is then used to compensate the frame data c1(n) after the first rough frequency offset estimation compensation, and the compensated frame data is recorded as c2(n): Assume that the preamble sequence is p 11 (n), n = 0, 1, 2, ... N2-1, N2 is the length of the preamble sequence, and its corresponding cyclic prefix is p 12 (n), n = 0, 1, 2, ... N1-1, N1 is the cyclic prefix length, take N1 = N2 / 2; the data sequence is d(n), n = 0, 1, 2, ... D-1, D is the data sequence length; the post-pilot sequence is p 21 (n), n = 0, 1, 2, ... N2-1, N2 is the length of the post-pilot sequence; the sampling frequency is f sym , the sampling period is T sym =1 / f sym , the carrier frequency used for up-conversion at the transmitter is f c ; The pre-pilot signal after up-conversion is expressed as The subsequent pilot signal can be expressed as Assume that due to the Doppler factor in the channel, the carrier frequency deviation of the received signal is f c becomes f c,rcv , then the front and rear pilot signals at the receiving end are expressed as Assume that the down-conversion local oscillator frequency used by the receiver is f c,ddc , then after the receiving end down-conversion, the baseband of the front-end pilot signal is expressed as: Where Δf = f c,rcv -f c,ddc is the carrier frequency offset to be estimated; The pre-pilot is cross-correlated with its cyclic sequence. Since the pilot signal used is a Zadoff-Chu sequence, the modulus value of each complex symbol is 1, that is, |p zc (n)|=1, then the second coarse frequency offset correlation result is expressed as C head ; The second coarse frequency offset estimation S6. Perform conjugate correlation operation on the front pilot and the rear pilot to obtain the phase offset and calculate the angle, and finally calculate the precise frequency offset estimate The step S6 comprises: The front and rear pilot signals of the frame data c2(n) after the second rough frequency offset estimation compensation are cross-correlated, and the cross-correlation result is used to estimate the carrier frequency offset accurately to obtain the final carrier frequency offset estimation. The estimation result is a fine frequency offset estimation result, which is used to compensate the frame data c2(n) after the second coarse frequency offset estimation compensation. The compensated frame data is recorded as c3(n): Among them, the baseband of the latter pilot signal is expressed as: The relevant results are expressed as: Fine frequency offset estimation Calculation formula: In the above formula, the physical meaning of the numerator is the phase change caused by the carrier frequency deviation from the front pilot to the rear pilot center. The denominator is a constant coefficient that depends on the time interval between the previous and next series used for calculation.
2. The method for estimating large frequency deviation of a burst short frame communication carrier according to claim 1, characterized in that: The burst short-time frame data structure includes: a front pilot, a rear pilot, and a data frame located between the front pilot and the rear pilot; There are a total of P sampling lengths from the front pilot to the rear pilot, and the sampling period of each sampling length is T sym ; The pre-pilot includes a cyclic prefix part and a pilot part. The cyclic prefix has a total sampling length of N1, and the pilot part has a total sampling length of N2, N1 = N2 / 2, and the sampling length from the cyclic prefix to 1 / 2 of the pilot part is N2; The time interval between the pre-pilot and the post-pilot is recorded as T2, and the time interval from the cyclic prefix to the 1 / 2 pilot part is recorded as T1: PT sym =T2,N2T sym =T1, where T sym =1 / f sym , T sym is the sampling period, f sym is the sampling frequency.
3. The method for estimating large frequency deviation of a burst short frame communication carrier according to claim 2, characterized in that: The sliding window N p Greater than or equal to P.
4. The method for estimating large frequency offset of a burst short frame communication carrier according to claim 2, characterized in that: The step S2 comprises: Set the sliding window size N p , set the signal oversampling rate f sam ; Since the transmitter needs to perform modulation and up-conversion during the burst short-time frame data communication process, the receiver down-converts and modulates the burst short-time frame data and then performs oversampling according to the set oversampling rate; The oversampled data is then sent to a sliding window, and a DFT transform is performed on the data in the window.
5. The method for estimating large frequency deviation of a burst short frame communication carrier according to claim 1, characterized in that: In step S4, the receiving end uses sliding correlation to achieve frame synchronization, cross-correlates the compensated frame data c1(n) with the local frame synchronization signal, and then locates the starting point of a frame signal according to the maximum value position; The period is N z The pilot sequence is defined as: where k = 0, 1, 2, 3, N z -1, j is the imaginary unit, j 2 =-1; r is a positive integer, take r=N z -1, then the autocorrelation characteristics of z(k) are: The modulated symbols of the locally generated correlation sequence are correlated with the received signal. The received signal slides one symbol each time. When the received signal coincides with the local symbol, a peak value N is generated in the correlation value. z , the frame start position is determined by the peak value, and the correlation result when the two sequences do not overlap satisfies the characteristic of zero side lobe; According to the starting position and the known frame structure information, the two pilots and their cyclic prefixes can be extracted. The extracted two pilots and the pre-pilot cyclic prefix are respectively denoted as p 11 (n), p 12 (n), p 21 (n).
Citation Information
Patent Citations
Frequency offset estimation method for high-dynamic large-frequency-offset burst signals
CN104852875A
Receiver burst signal synchronization method and device for single carrier frequency domain equalization system
CN111585933A
