A fast search method for pre-equalization coefficients based on scanning matrix

Abstract

The present invention discloses a method for fast searching of pre-equalization coefficients based on a scanning matrix, which is applied to an equalizer at a transmitting end of a communication system, including: initializing search parameters; obtaining a search range of each tap coefficient according to a search sequence number of a current row of a search matrix; sequentially traversing each position in a corresponding row, updating the tap coefficient value within the search range for any position and configuring the value through a host computer software, iteratively searching the tap coefficient of the position by reading the error vector amplitude result of a preset spectrum analyzer, and searching the next position in the row after completion; after completing the search of all tap coefficients in the row, proceed to the next row until the EVM index requirement is met or the search matrix traversal is completed, and the final tap coefficient value is obtained. The present invention has low implementation complexity, can complete the pre-equalization coefficient search in a relatively short time, and can effectively compensate for the reduced EVM performance of the transmitting end caused by group delay and distortion after configuration.

Description

Technical Field

[0001] The invention belongs to the technical field of digital satellite communications, and in particular relates to a fast search method for pre-equalization coefficients based on a scanning matrix. Background Art

[0002] With the development of digital satellite communication technology, the high rate and large bandwidth of digital signals have put forward higher requirements on the transmitter and receiver of the communication system. However, large bandwidth signals will produce group delay distortion after passing through the communication channel, and due to the nonlinear characteristics of the power amplifier, nonlinear distortion may also occur, which seriously deteriorates the error vector magnitude (EVM) performance of the transmitter. Adding a pre-equalization filter in the digital domain of the transmitter to offset the impact of the group delay distortion of the communication channel is a common method to improve EVM. Typical pre-equalization methods include static pre-equalization and dynamic pre-equalization. For dynamic pre-equalization, the transmitter needs to have the function of receiving sampled feedback data in real time, which greatly increases the processing complexity of the digital domain of the transmitter and causes the cost of the signal generator to be too high; for static pre-equalization, the pre-equalization coefficient is obtained by offline calculation, but this method requires additional high-speed data acquisition equipment, and the corresponding signal receiver needs to be designed to calculate the equalization coefficient. The high-speed data acquisition equipment can be a dedicated signal acquisition instrument or a spectrum analyzer with the function of collecting high-speed data, but in the absence of high-speed acquisition equipment, this method cannot be used.

[0003] As a static pre-equalization method, Zeng Dejun proposed a minimum mean square adaptive structure for digital pre-distortion in the article "Design of Digital Pre-distortion Based on New Parallel LMS Algorithm" published in the journal "Communication Technology" in 2022. It can improve the linearity of the power amplifier and compensate for the nonlinear distortion of the signal after passing through the power amplifier. This method optimizes the structure of the adaptive filter and simplifies the complexity of implementing the digital pre-distorter on the FPGA side. However, if this method is to be applied to the transmitter, it is first necessary to collect signal data passing through the communication channel and the power amplifier, and feed it back to the transmitter for calculation through the minimum mean square pre-distortion model, and then compensate for the EVM performance of the signal after obtaining the pre-distortion filter coefficients. Therefore, this method requires a device that can collect high-speed data and feed it back to the FPGA side. If there is no device for collecting data, this method cannot be used.

[0004] Therefore, how to reduce the implementation complexity of the existing static pre-equalization method and reduce the search time of the pre-equalization coefficient so as to quickly and easily compensate for the performance degradation caused by group delay and distortion on the transmitter is an urgent problem to be solved in this field. Summary of the invention

[0005] To solve the above problems existing in the prior art, the present invention provides a fast search method and device for pre - equalization coefficients based on a scanning matrix. The technical problems to be solved by the present invention are realized through the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention provides a fast search method for pre - equalization coefficients based on a scanning matrix, which is applied to an equalizer at the transmitting end of a communication system. The method includes:

[0007] Step 1, initialize search parameters, including initializing the tap - coefficient vector W of the equalizer ini , search step size μ and search matrix R, and initialize the error - vector magnitude E, row search sequence number i, and column search sequence number k to 1; where, the N elements in the tap - coefficient vector W ini are used to represent T complex equalizer coefficients, N = 2T; the dimension of the search matrix R is M×N, and its internal elements represent the search upper - bound values of the tap coefficients at the corresponding positions; M represents the maximum number of searches;

[0008] Step 2, extract the row vector A corresponding to the current row search sequence number i in the search matrix R;

[0009] Step 3, for the k - th tap coefficient in the row vector A, determine its current tap - coefficient search lower - bound value d k , and combine the current search step size μ and the tap - coefficient vector W ini to obtain the corresponding tap - coefficient test vector W;

[0010] Step 4, configure the equalizer with the obtained tap - coefficient test vector W through the upper - computer software, read the error - vector magnitude result evm of the preset spectrum analyzer and compare it with the current error - vector magnitude E to obtain the updated tap - coefficient vector W new and the error - vector magnitude E new ; where, the preset spectrum analyzer has an EVM analysis function;

[0011] Step 5, determine whether k = N is satisfied currently; if not, execute Step 6, if so, execute Step 7;

[0012] Step 6, determine whether the current tap - coefficient search lower - bound value d k is equal to the corresponding search upper - bound value; if equal, increment the current k by one and return to Step 3, if not equal, increment the current d k by one and return to Step 3;

[0013] Step 7, determine whether i < M is satisfied currently; if satisfied, execute Step 8, if not satisfied, execute Step 9;

[0014] Step 8, increment the current i by one, and the current Wnew and E new As the updated tap coefficient vector W ini and error vector magnitude E, and reduce the search step size and update it, and return to step 2;

[0015] Step 9: Stop searching and output the current tap coefficient vector W new .

[0016] In one embodiment of the present invention, the tap coefficient vector W of the equalizer is initialized. ini The formulas used in the process include:

[0017]

[0018] Among them, w n Denotes the tap coefficient vector W ini The nth element in , N represents the total number of elements, and n and N are natural numbers greater than 0.

[0019] In one embodiment of the present invention, the T complex equalizer coefficients are c1, c2, ..., c t ,…,c T ;in,

[0020] c t =w 2t-1 +j·w 2t , t=1,2,…,T

[0021] w 2t-1 Denotes the tap coefficient vector W ini The elements corresponding to the odd positions in w 2t Denotes the tap coefficient vector W ini The elements corresponding to the even positions in ; j represents the imaginary unit; T is a natural number greater than 0.

[0022] In one embodiment of the present invention, in step 3, for the kth tap coefficient in the row vector A, at the first iteration, the current tap coefficient searches for the lower bound value d k It is the negative value of the search upper limit value of the tap coefficient at the corresponding position in the row vector A.

[0023] In one embodiment of the present invention, the combination of the current search step size μ and the tap coefficient vector W ini , get the corresponding tap coefficient test vector W, including:

[0024] The corresponding tap coefficient test vector W is determined using the tap coefficient test vector calculation formula, wherein the tap coefficient test vector calculation formula includes: W = W ini +d k ·μ.

[0025] In one embodiment of the present invention, the step of performing equalizer configuration on the obtained tap coefficient test vector W by the host computer software includes:

[0026] The obtained tap coefficient test vector W is sent to the FPGA using the host computer software;

[0027] The FPGA configures the equalizer using the received tap coefficient test vector W.

[0028] In one embodiment of the present invention, the error vector magnitude result evm of the preset spectrum analyzer is read and compared with the current error vector magnitude E to obtain the tap coefficient vector W after searching for update. new and the error vector magnitude E new ,include:

[0029] Sending the output signal obtained by the configured equalizer for its input signal to a preset spectrum analyzer; wherein the input signal is a digital signal output by the modulator to be equalized;

[0030] Read the test result of the preset spectrum analyzer to obtain the error vector magnitude result evm;

[0031] Compare the error vector magnitude result evm with the current error vector magnitude E, and obtain the searched updated tap coefficient vector W according to the result corresponding to the smaller value new and the error vector magnitude E new .

[0032] In one embodiment of the present invention, the error vector magnitude result evm is compared with the current error vector magnitude E, and the updated tap coefficient vector W is obtained according to the result corresponding to the smaller value. new and the error vector magnitude E new ,include:

[0033] The tap coefficient vector W after searching for update is determined by using the preset comparison formula new and the error vector magnitude E new ; Wherein, the preset comparison formula includes:

[0034]

[0035]

[0036] In one embodiment of the present invention, the updating of reducing the search step size includes:

[0037] Reduce the search step size to half of the current search step size.

[0038] In one embodiment of the present invention, for the fast search method for pre-equalization coefficients based on the scanning matrix, the calculation formula for the total number of searches required for traversing the search matrix R is:

[0039]

[0040] Among them, r i,n represents the element in the i-th row and n-th column in the search matrix R; ∑(·) represents summation; Π(·) represents multiplication.

[0041] Beneficial effects of the present invention:

[0042] The embodiment of the present invention designs a fast search method for pre-equalization coefficients based on a scanning matrix. By designing a scanning matrix to perform online search for the pre-equalization coefficients used for the equalizer, the number of searches can be greatly reduced. This method only requires a spectrum analyzer with an EVM analysis function, and the setting of the pre-equalization coefficients can be completed with the help of host computer software. The implementation complexity of the embodiment of the present invention is low, and the search for the pre-equalization coefficients can be completed in a relatively short time. By configuring the pre-equalizer for the searched pre-equalization coefficients, the error vector amplitude of the broadband signal can be improved, and the performance degradation caused to the transmitting end by compensating for group delay and distortion can be compensated. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic flow chart of a method for quickly searching for pre-equalization coefficients based on a scanning matrix provided by an embodiment of the present invention;

[0044] Figure 2 Another flowchart diagram of a method for quickly searching for pre-equalization coefficients based on a scanning matrix provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] like Figure 1 As shown, a method for quickly searching for pre-equalization coefficients based on a scanning matrix provided in an embodiment of the present invention is applied to an equalizer at a transmitting end of a communication system and may include the following steps:

[0047] Step 1: Initialize the search parameters, including initializing the tap coefficient vector W of the equalizer ini, search step size μ and search matrix R, and initialize the error vector magnitude E, row search number i, and column search number k to 1;

[0048] Specifically, the search parameters of the embodiment of the present invention include the tap coefficient vector W ini , search step size μ, search matrix R, error vector magnitude E, row search number i and column search number k.

[0049] Among them, the tap coefficient vector W ini The N elements in [w1,w2,…,w n ,…,w N-1 ,w N ] is used to represent T complex equalizer coefficients, that is, to represent the real and imaginary values ​​of the pre-equalization complex coefficients; wherein the T complex equalizer coefficients are c1, c2, …, c t ,…,c T ; N=2T, N and T are natural numbers greater than 0.

[0050] In an optional implementation, the tap coefficient vector W can be used ini The values ​​in the odd positions represent the real part of the complex equalizer coefficients, represented by the tap coefficient vector W ini The values ​​in the even positions represent the imaginary part of the complex equalizer coefficient, or vice versa.

[0051] For example, for the first case, each complex equalizer coefficient is specifically expressed as:

[0052] c t =w 2t-1 +j·w 2t , t=1,2,…,T

[0053] Among them, w 2t-1 Denotes the tap coefficient vector W ini The elements corresponding to the odd positions in w 2t Denotes the tap coefficient vector W ini The elements corresponding to the even positions in ; j represents the imaginary unit.

[0054] Initialize the tap coefficient vector W of the equalizer ini In order to reduce the amount of calculation, in an optional implementation, W ini Any one of the N elements of is set to 1, and the rest are set to 0. For example, the element corresponding to any odd position can be set to 1, and the rest can be set to 0.

[0055] In an optional implementation, the tap coefficient vector W of the equalizer is initialized ini The formulas used in the process include:

[0056]

[0057] Among them, w n Denotes the tap coefficient vector W ini The nth element in , N represents the total number of elements, and n and N are natural numbers greater than 0.

[0058] This implementation method is to ini The center position of the element is set to 1, and the rest are set to 0. According to the characteristics of the equalizer in the communication system, the coefficient value at the center position of the converged equalizer tap coefficient is large and close to 1, so setting the center position to 1 during initialization can reduce the number of searches.

[0059] The search step size μ represents the step value for performing two adjacent searches on the tap coefficient values ​​in the search matrix R, and its value range is (0,1). The initial value of the search step size μ can be reasonably set within its value range as needed. For example, in an optional implementation, the initial value of the search step size μ can be 1 / 8, etc.

[0060] The search matrix R is set according to the empirical value. Specifically, it can be known from the characteristics of the equalizer that the center position of the tap coefficient of the equalizer and the adjacent 1 to 2 tap coefficients are key coefficients, which require multiple searches, and the element values ​​at the corresponding positions in the search matrix are large. For example, for an equalizer with 9 tap coefficients, the center position and the nearby 1 to 2 tap coefficients are key coefficients, so only 3 to 5 tap coefficient values ​​need to be searched, and the search matrix element values ​​can be increased row by row to achieve the purpose of reducing the number of searches. The dimension of the search matrix R is M×N, and the search matrix R can be expressed as:

[0061]

[0062] Among them, r m,n represents the search for the element in the mth row and nth column of the matrix R, where m = 1, 2, ..., M, n = 1, 2, ..., N, and It indicates that each element in the search matrix R is a natural number; the elements in the search matrix R represent the search upper limit value of the tap coefficient at the corresponding position; M represents the maximum number of searches, which can be reasonably set as needed.

[0063] The error vector magnitude E indicates how close the real and imaginary parts generated when the transmitter demodulates the signal are to the ideal signal components. The error vector magnitude ranges from [0,1], and its initial value can be set to 1, that is, it is initialized with the maximum value 1 for search.

[0064] The row search number i indicates that the i-th row of the matrix is ​​currently being searched, and the column search number k indicates that the position of the k-th tap coefficient in a row of the matrix R is being searched. To facilitate traversal, in the embodiment of the present invention, the initial values ​​of the row search number i and the column search number k are both set to 1, that is, the search starts from the position corresponding to the first row and the first column of the search matrix R.

[0065] Step 2, extracting the row vector A corresponding to the current row search number i in the search matrix R;

[0066] Specifically, the search upper limit value of the tap coefficient of the i-th row of the search matrix R is extracted according to the current row search sequence number i, and the result is:

[0067] A=R i =[a i,1 ,a i,2 ,…,a i,n ,…,a i,N ]

[0068] Among them, a i,n =r i,n , n=1,2,…,N.

[0069] It can be understood that when step 2 is executed for the first time, the extracted row vector A is the first row in the search matrix R.

[0070] Step 3: For the kth tap coefficient in the row vector A, determine the current tap coefficient search lower bound d k , and combined with the current search step size μ and tap coefficient vector W ini , get the corresponding tap coefficient test vector W;

[0071] Among them, in step 3, for the kth tap coefficient in the row vector A, at the first iteration, its current tap coefficient searches for the lower bound value d k is the negative value of the search upper limit value of the tap coefficient at the corresponding position in the row vector A. That is, when the iterative search starts for the k-th tap coefficient in the row vector A, d is determined according to the column search number k of the tap coefficient. k =-a i,k , as the initial value for searching the lower bound.

[0072] In an optional implementation manner, when determining the current tap coefficient search lower bound value d of the kth tap coefficient in the row vector A, k After that, the combination of the current search step size μ and the tap coefficient vector W ini , get the corresponding tap coefficient test vector W, including:

[0073] The corresponding tap coefficient test vector W is determined using the tap coefficient test vector calculation formula, wherein the tap coefficient test vector calculation formula includes: W = W ini +d k ·μ.

[0074] Of course, the tap coefficient test vector calculation formula is not limited to the above.

[0075] In the embodiment of the present invention, the obtained tap coefficient test vector W represents the new tap coefficient vector searched, and its function is to determine the new tap coefficient vector and error vector amplitude after equalizer configuration and related tests.

[0076] It can be understood that when step 3 is performed for the first time, the kth tap coefficient is the tap coefficient corresponding to the first row and first column in the search matrix R.

[0077] Step 4: Use the host computer software to configure the tap coefficient test vector W for the equalizer, read the error vector amplitude result evm of the preset spectrum analyzer and compare it with the current error vector amplitude E to obtain the tap coefficient vector W after searching and updating. new and the error vector magnitude E new ;

[0078] The preset spectrum analyzer has an EVM analysis function, which can be implemented by any existing spectrum analyzer with this function.

[0079] The step of performing equalizer configuration on the obtained tap coefficient test vector W through the host computer software includes:

[0080] The obtained tap coefficient test vector W is sent to the FPGA using the host computer software;

[0081] The FPGA configures the equalizer using the received tap coefficient test vector W.

[0082] For the specific configuration process, please refer to the relevant technical understanding, which will not be explained in detail here.

[0083] The error vector magnitude result evm of the preset spectrum analyzer is read and compared with the current error vector magnitude E to obtain the tap coefficient vector W after searching for update. new and the error vector magnitude E new ,include:

[0084] (1) sending the output signal obtained by the configured equalizer for its input signal to a preset spectrum analyzer;

[0085] The input signal is a digital signal output by a modulator to be equalized.

[0086] (2) reading the test result of the preset spectrum analyzer to obtain the error vector magnitude result evm;

[0087] Among them, for the specific process of presetting the spectrum analyzer to obtain the error vector magnitude result evm, please refer to the relevant technical understanding.

[0088] (3) Compare the error vector magnitude result evm with the current error vector magnitude E, and obtain the searched updated tap coefficient vector W according to the result corresponding to the smaller value new and the error vector magnitude E new .

[0089] This step includes:

[0090] The tap coefficient vector W after searching for update is determined by using the preset comparison formula new and the error vector magnitude E new ; Wherein, the preset comparison formula includes:

[0091]

[0092]

[0093] It can be seen that for E new , is to select the smaller value between evm and E, and for W new , is a tap coefficient correlation vector corresponding to a smaller value selected from evm and E. The embodiment of the present invention achieves the purpose of compensating the distorted digital signal at the transmitting end and reducing its EVM value by screening the search direction where evm and E decrease.

[0094] It can be understood that for the kth tap coefficient in the row vector A, the corresponding E can be obtained through steps 3 to 4. new and W new .

[0095] Step 5, determine whether k=N is currently satisfied; if not, execute step 6; if so, execute step 7;

[0096] If k=N is satisfied, it means that all tap coefficients in the row vector A have been searched, and it is necessary to execute step 7 to try the next row search.

[0097] If k=N is not satisfied, it means that there are tap coefficients in the row vector A that have not been searched completely, and then step 6 is executed to continue searching in the row vector A.

[0098] Step 6: Determine the current tap coefficient search lower limit value d k Is it equal to the corresponding search upper bound? If so, increase the current k by 1 and return to step 3. If not, update the current d kAfter performing the increment update, return to step 3;

[0099] As can be seen from the foregoing, the lower bound value d of the tap coefficient search k has an initial value that is the negative of the corresponding upper bound value of the search, that is, the search range for each tap coefficient is [-r m,n , r m,n .

[0100] If the current lower bound value d of the tap coefficient search k is equal to the corresponding upper bound value of the search, it means that the search for this tap coefficient within the search range has been completed, and it is necessary to perform the search for the next tap coefficient in this row, that is, after k + 1, return to step 3 to continue the search. It can be understood that, compared with before, after returning to step 3, k changes, and its current lower bound value d of the tap coefficient search k also changes, becoming the negative of the upper bound value of the tap coefficient at the corresponding position in the row vector A, while the row search sequence number i, the search step size μ, and the tap coefficient vector W ini do not change.

[0101] If the current lower bound value d of the tap coefficient search k is not equal to the corresponding upper bound value of the search, it means that the search for this tap coefficient within the search range has not been completed, and it is necessary to continue the search for this tap coefficient. At this time, after d k + 1, return to step 3 to continue the search. It can be understood that, compared with before, after returning to step 3, k does not change, the row search sequence number i, the search step size μ, and the tap coefficient vector W ini do not change either, and d k changes.

[0102] Step 7, determine whether the current i < M is satisfied; if satisfied, execute step 8, if not satisfied, execute step 9;

[0103] If i < M is satisfied, it means that the rows in the search matrix R have not been completely searched, and it is necessary to search the next row. At this time, use the existing results to perform the update in step 8, and then return to step 2 to search a new row.

[0104] If i < M is not satisfied, that is, there is i = M, it means that the rows in the search matrix R have been completely searched, then execute step 9, stop the search, and output the current search result.

[0105] Step 8, increment the current i by one, and use the current W new and E new as the updated tap coefficient vector W ini and the error vector magnitude E, and update by reducing the search step size, then return to step 2;

[0106] In an optional implementation manner, the updating of reducing the search step size includes:

[0107] Update the search step size to half of the current search step size, that is, set the search step size to That is, the search step size is updated using the binary method.

[0108] Based on the above implementation, the row search number updated in step 8 is i+1, and the updated tap coefficient vector W ini =W new , the updated error vector magnitude E = E new , the updated search step length is Return to step 2 using these updated parameters and perform a search for row i+1. See the previous section for details.

[0109] Of course, in the embodiment of the present invention, the method of reducing the search step size is not limited to setting The 1 / 2 can also be 1 / 3, 1 / 4, 1 / 5, or other values ​​between (0,1).

[0110] Step 9: Stop searching and output the current tap coefficient vector W new .

[0111] Specifically, the current tap coefficient vector W new The final searched pre-equalization coefficient can be used to configure the equalizer to compensate for the deterioration of the transmitter performance. Specifically, the equalizer will perform equalization filtering on the digital signal of the transmitter according to the configured searched pre-equalization coefficient to achieve the purpose of offsetting the distortion and distortion in the signal, while increasing the error vector amplitude of the transmitter signal to compensate for the deterioration of the transmission performance caused by distortion.

[0112] As mentioned above, since the search range of each tap coefficient is [-r m,n ,r m,n ],d k The change step is 1. For the fast search method for pre-equalization coefficients based on the scanning matrix, the calculation formula for the total number of searches required for traversing the search matrix R is:

[0113]

[0114] Among them, r i,n represents the element in the i-th row and n-th column in the search matrix R; ∑(·) represents summation; Π(·) represents multiplication.

[0115] Therefore, when the search matrix R is determined, the total number of searches can be clearly calculated using the above calculation formula.

[0116] A specific example is given below to illustrate the method of the present invention, and please refer to Figure 2 understand.

[0117] The specific steps include:

[0118] (1) Initialize search parameters:

[0119] Initialize the tap coefficient value of the equalizer to W ini =[0,0,1,0,0,0,0,0], representing c1, c2, c3, c4, a total of 4 complex equalizer coefficients, where each complex equalizer coefficient can be expressed as c t =w 2t-1 +i·w 2t , t = 1, 2, 3, 4, c2 is the center tap coefficient value, c1 is the forward tap coefficient value, c3 and c4 are the backward tap coefficient values, and the initial search step size is

[0120] The search matrix is:

[0121]

[0122] The error vector amplitude E is initialized to E=1; i represents the row search number, indicating that the i-th row of the matrix R is being searched, and is initialized to i=1; k represents the column search number, indicating that the k-th tap coefficient in a row of the matrix R is being searched, and is initialized to k=1.

[0123] (2) Iteratively searching for equalizer coefficients, i.e., tap coefficient values, specifically includes the following steps:

[0124] (2a) According to the current row search number i of the search matrix R, the search upper limit value of the tap coefficient of the i-th row of the search matrix R is taken out, that is, A = R i =[a i,1 ,a i,2 ,…,a i,8 ], where a i,n =r i,n , n=1,2,…,8;

[0125] (2b) Iteratively searching each tap coefficient in A by searching the upper bound value, specifically comprising the following steps:

[0126] (2b1) According to the column search number k of each tap coefficient, its initial search lower limit value is d k =-a i,k ;

[0127] (2b2) ​​Through the current search lower bound value d k The tap coefficient test vector W that characterizes the tap coefficient of the equalizer is obtained as:

[0128] W = W ini + d k · μ

[0129] (2b3) Use the host computer software to send the tap coefficient test vector W obtained in (2b2) to the FPGA;

[0130] (2b4) The FPGA configures the equalizer with the equalizer tap coefficient W in (2b3), and connects the output signal of the equalizer to a preset spectrum analyzer for analysis;

[0131] (2b5) Observe its error vector magnitude evm through the preset spectrum analyzer, and judge the tap coefficient test vector W and the error vector magnitude E according to evm to obtain the updated tap coefficient vector W after search new and the error vector magnitude E new :

[0132]

[0133]

[0134] (2b6) Judge that if k = N, then execute step (2c), otherwise judge d k and make a judgment: if d k = a i,k , execute step (2b7), otherwise execute step (2b8);

[0135] (2b7) Update k = k + 1, and repeat steps (2b1) to (2b6);

[0136] (2b8) Update d k = d k + 1, and repeat steps (2b2) to (2b6);

[0137] (2c) Judge that if i < M, execute step (2c1), otherwise execute step (2c2):

[0138] (2c1) Update the search row number i, the tap coefficient vector W ini , the error vector magnitude E and the search step μ: update i = i + 1, W ini = W new , E = E new , Repeat steps (2a) to (2c);

[0139] (2c2) Stop the search and obtain the tap coefficient value W of the equalizer new .

[0140] Combined with the search matrix in the embodiment, the embodiment of the present invention searches the tap coefficients of the non-zero element positions of the first row of the search matrix with a step size of 1 / 8 for the first time, and a total of 3×3×3=27 searches are required; the second time, based on the first search result, the tap coefficients of the non-zero element positions of the second row of the search matrix are searched with a step size of 1 / 16, and a total of 3×5×5=75 searches are required; the third time, based on the second search result, the tap coefficients of the non-zero element positions of the third row of the search matrix are searched with a step size of 1 / 32, and a total of 3×3×5=75 searches are required. ×5=225 times; the fourth time, based on the search results of the third time, the tap coefficients of the non-zero element positions in the 4th row of the search matrix are searched with a step size of 1 / 64, and a total of 3^7=2187 searches are required; the fifth time, based on the search results of the fourth time, the tap coefficients of the non-zero element positions in the 5th row of the search matrix are searched with a step size of 1 / 128, and 3^7=2187 searches are also required; the last time, the tap coefficients of the non-zero element positions in the 6th row of the search matrix are searched with a step size of 1 / 256, and 3^8=6561 searches are required. After many actual tests, it has been proved that using the above search matrix, after the first three searches, the EVM can generally be reduced to less than 5%, after the first 5 searches, it can be reduced to 3.5%, and after the last search, the EVM can be reduced to 3.2%. In other words, only 27+75+225=327 searches are required to achieve the required EVM performance. Assuming that each search takes 100 ms, the basic search can be completed within 33 seconds, which greatly reduces the number of searches and shortens the time required to search for pre-equalization coefficients.

[0141] The embodiment of the present invention designs a fast search method for pre-equalization coefficients based on a scanning matrix. By designing a scanning matrix to perform online search for the pre-equalization coefficients used for the equalizer, the number of searches can be greatly reduced. This method only requires a spectrum analyzer with an EVM analysis function, and the setting of the pre-equalization coefficients can be completed with the help of the host computer software. The implementation complexity of the embodiment of the present invention is low, and the search for the pre-equalization coefficients can be completed in a relatively short time. By configuring the pre-equalizer for the searched pre-equalization coefficients, the error vector amplitude of the broadband signal can be improved, and the performance degradation caused to the transmitting end by compensating for group delay and distortion can be compensated.

[0142] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. A fast search method for pre-equalization coefficients based on a scanning matrix, characterized in that: Applied to the equalizer at the transmitting end of a communication system, including: Step 1: Initialize the search parameters, including initializing the tap coefficient vector W of the equalizer ini , search step size μ and search matrix R, and initialize the error vector magnitude E, row search number i, column search number k to 1; among them, the tap coefficient vector W ini The N elements in are used to represent T complex equalizer coefficients, where N=2T; the dimension of the search matrix R is M×N, and the elements in it represent the search upper bound values ​​of the tap coefficients at the corresponding positions; M represents the maximum number of searches; Step 2: Extract the row vector A corresponding to the current row search sequence number i in the search matrix R; Step 3: For the kth tap coefficient in the row vector A, determine the current tap coefficient search lower bound d k , and combined with the current search step size μ and tap coefficient vector W ini , get the corresponding tap coefficient test vector W; Step 4: Use the host computer software to configure the tap coefficient test vector W for the equalizer, read the error vector amplitude result evm of the preset spectrum analyzer and compare it with the current error vector amplitude E to obtain the tap coefficient vector W after searching and updating. new and the error vector magnitude E new ; Wherein, the preset spectrum analyzer has an EVM analysis function; Step 5: Determine whether k = N is satisfied currently; if not, execute Step 6, if so, execute Step 7; Step 6: Determine the current tap coefficient search lower limit value d k Is it equal to the corresponding search upper bound? If so, increase the current k by 1 and return to step 3. If not, update the current d k After adding one to the update, return to step 3; Step 7: Determine whether i < M is satisfied currently; if satisfied, execute Step 8, if not satisfied, execute Step 9; Step 8: Update the current i by one and set the current W new and E new As the updated tap coefficient vector W ini and error vector magnitude E, and reduce the search step size and update it, and return to step 2; Step 9: Stop searching and output the current tap coefficient vector W new .

2. The method for quickly searching for pre-equalization coefficients based on a scanning matrix according to claim 1, characterized in that: Initialize the tap coefficient vector W of the equalizer ini The formula used in the process Including: Among them, w n Denotes the tap coefficient vector W ini The nth element in , N represents the total number of elements, and n and N are natural numbers greater than 0.

3. The method for fast searching pre-equalization coefficients based on scanning matrix according to claim 1, characterized in that: The T complex equalizer coefficients are c1, c2, ..., c t ,…,c T ; Wherein, c t =w 2t-1 +j·w 2t ,t=1,2,…,T w 2t-1 Denotes the tap coefficient vector W ini The elements corresponding to the odd positions in w 2t Denotes the tap coefficient vector W ini The elements corresponding to the even positions in ; j represents the imaginary unit; T is a natural number greater than 0.

4. The method for fast searching pre-equalization coefficients based on scanning matrix according to claim 1, characterized in that: In step 3, for the kth tap coefficient in the row vector A, at the first iteration, the current tap coefficient searches for the lower bound value d k It is the negative value of the search upper limit value of the tap coefficient at the corresponding position in the row vector A.

5. The method for quickly searching for pre-equalization coefficients based on a scanning matrix according to claim 4, characterized in that: The combination of the current search step size μ and the tap coefficient vector W ini , get the corresponding tap coefficient test vector W, including: The corresponding tap coefficient test vector W is determined using the tap coefficient test vector calculation formula, wherein the tap coefficient test vector calculation formula includes: W = W ini +d k ·μ.

6. The method for fast searching pre-equalization coefficients based on scanning matrix according to claim 5, characterized in that: The equalizer configuration by using the obtained tap coefficient test vector W through the host computer software includes: Use the host computer software to send the obtained tap coefficient test vector W to the FPGA; The FPGA configures the equalizer by using the received tap coefficient test vector W.

7. The method for quickly searching for pre-equalization coefficients based on a scanning matrix according to claim 6, characterized in that: The error vector magnitude result evm of the preset spectrum analyzer is read and compared with the current error vector magnitude E to obtain the tap coefficient vector W after searching for update. new and the error vector magnitude E new ,include: Send the output signal obtained by the configured equalizer for its input signal to a preset spectrum analyzer; wherein, the input signal is the digital signal output by the modulator to be equalized; Read the test result of the preset spectrum analyzer to obtain the error vector magnitude result evm; Compare the error vector magnitude result evm with the current error vector magnitude E, and obtain the searched updated tap coefficient vector W according to the result corresponding to the smaller value new and the error vector magnitude E new .

8. The method for fast searching pre-equalization coefficients based on scanning matrix according to claim 7, characterized in that: The error vector magnitude result evm is compared with the current error vector magnitude E, and the updated tap coefficient vector W is obtained according to the result corresponding to the smaller value. new and the error vector magnitude E new ,include: The tap coefficient vector W after searching for update is determined by using the preset comparison formula new and the error vector magnitude E new ; Wherein, the preset comparison formula includes:

9. The method for fast searching pre-equalization coefficients based on scanning matrix according to claim 1, characterized in that: The reduction and update of the search step size include: Reduce and update the search step size to half of the current search step size.

10. The method for fast searching pre-equalization coefficients based on scanning matrix according to any one of claims 1 to 9, characterized in that: For the fast search method of the pre-equalization coefficient based on the scan matrix, the calculation formula for the total number of search times required for traversing and searching the search matrix R is: Among them, r i,n represents the element in the i-th row and n-th column in the search matrix R; ∑(·) represents summation; Π(·) represents multiplication.

Images