A method and system for equalization compensation of a communication channel

Abstract

The application belongs to the technical field of wireless communication, and relates to a communication channel equalization compensation method and system. A baseband signal sequence is obtained by time domain sampling of a receiving end, a signal segment is obtained by a sliding window, a channel impulse response is estimated by using a training sequence to obtain tap coefficients, a tap energy measurement value is determined, a tap exceeding a threshold value is marked as a candidate support set, a signal segment in the candidate support set is used as an observation vector, a time shift version of a sending signal is used as a dictionary atom, a matching atom is selected from an atom set corresponding to the candidate support set by using an orthogonal matching pursuit algorithm, and a channel sparse representation is reconstructed to obtain reconstructed local channel coefficients and an effective support set. Equalization symbols are output by equalizing and compensating a current window signal according to the effective support set. The application limits the reconstruction search to the candidate support set by two-stage screening, reduces the calculation amount, and improves the equalization stability under a time-varying channel by adjacent window support set continuity checking and double-window parallel fusion.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology, specifically a communication channel equalization compensation method and system. Background Technology

[0002] In wireless communication systems, inter-symbol interference (ISI) occurs when signals are transmitted through multipath channels, affecting communication quality and limiting the improvement of system transmission rate. Channel equalization technology eliminates ISI by constructing the inverse response of the channel at the receiver and is one of the core modules in the digital signal processing link of wireless communication receivers. Currently, the mainstream equalization schemes include linear equalizers, decision feedback equalizers, and sparse channel equalization methods based on compressed sensing. Among them, the equalization scheme based on compressed sensing utilizes the sparsity of the channel impulse response in the time domain and recovers the channel coefficients from a small number of observations through reconstruction algorithms such as orthogonal matched pursuit. Related research has been conducted on reducing pilot overhead.

[0003] However, the above-mentioned schemes still have room for optimization in practical applications. Channel equalization methods based on compressed sensing typically perform sparse reconstruction within the global tap range. The computational cost is related to the channel memory length, and in scenarios with significant channel delay spread, the computational resource consumption of the reconstruction process still needs further optimization. Furthermore, the location of the tap support set and the reconstruction of channel coefficients are usually coupled within the same algorithm framework, and the mechanism by which low-energy taps and noise taps participate in the calculation during the iteration process still has room for simplification. Moreover, existing equalizers require re-estimation and reconstruction of the channel when channel characteristics change, and their convergence speed and tracking capability remain areas of ongoing concern for those skilled in the art. Summary of the Invention

[0004] To address the problems mentioned in the background section, this invention provides a communication channel equalization compensation method and system to solve the problems of large computational load and insufficient tracking capability in channel equalization.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a communication channel equalization compensation method, comprising the following steps: The receiving end performs time-domain sampling on the received signal to obtain the baseband digital signal sequence; The baseband digital signal sequence is segmented according to the symbol reception order using a sliding time-domain window to obtain the signal segment within the current window; Based on the signal segment within the current window and the known training sequence, estimate the channel impulse response within the window and obtain the channel impulse response coefficients at each tap position; Based on the channel impulse response coefficient, determine the time-domain energy metric value of each tap position within the window; and mark the tap positions whose time-domain energy metric value exceeds a preset energy threshold as a candidate tap support set; The received signal segments within the candidate tap support set are used as observation vectors, and the time-shifted version of the transmitted signal is used as dictionary atoms. The orthogonal matching pursuit algorithm is used to select matching atoms from the atom set corresponding to the candidate tap support set and reconstruct the sparse representation of the channel impulse response to obtain the reconstructed local channel coefficients and effective tap support set. Based on the reconstructed local channel coefficients, equalization compensation is performed on the signal segments within the current window, and the equalized symbol sequence is output.

[0006] Optionally, the sliding time-domain window covers L consecutive symbol periods, and the sliding time-domain window slides forward once every M new symbols received, where M is no greater than 1 / 2 of L.

[0007] Optionally, the method for determining the time-domain energy metric is as follows: The amplitudes of the channel impulse response coefficients at each tap position within the current window are squared and accumulated, and the accumulated result is used as the time-domain energy metric for that tap position.

[0008] Optionally, the iteration termination condition of the orthogonal matching pursuit algorithm is: The residual energy is lower than the preset residual threshold or the number of iterations reaches the preset maximum number of iterations; The preset maximum number of iterations is equal to half the total number of tap positions in the candidate tap support set.

[0009] Optionally, the equalization compensation constructs the tap weights of the linear equalizer based on the reconstructed local channel coefficients, and performs zero-forcing equalization or minimum mean square error equalization on the signal segments within the current window.

[0010] Optionally, a continuity check is performed on the effective tap support set between two adjacent sliding windows: If the number of overlapping taps between the current window's valid tap support set and the previous window's valid tap support set is greater than or equal to a preset percentage threshold of the total number of taps in the previous window's valid tap support set, then the current window's valid tap support set remains unchanged. Otherwise, a relocation process is triggered, which includes lowering a preset energy threshold and recalculating the time-domain energy metric.

[0011] Optionally, when the relocation process is triggered: The preset energy threshold is adjusted to half of the preset energy threshold used in the previous window, the time-domain energy metric is recalculated, and the candidate tap support set is updated. The adjustment stops when the preset energy threshold is reduced to the preset lower limit.

[0012] Optionally, a first sliding window and a second sliding window of different lengths are used to perform parallel sliding monitoring on the same received signal sequence. The length of the first sliding window is greater than the length of the second sliding window, and the two sliding windows slide synchronously with the same length M. When the size of the intersection of the first candidate tap support set located by the first sliding window and the second candidate tap support set located by the second sliding window exceeds a preset ratio threshold of the size of the second candidate tap support set, the intersection is used as the final candidate tap support set for subsequent channel reconstruction. Otherwise, the second candidate tap support set is used as the final candidate tap support set, and the window length of the first sliding window is reduced by M symbols.

[0013] Optionally, when the time-domain energy metric value at which no tap position is detected within the sliding time-domain window exceeds the preset energy threshold: If it is determined that there are no valid multipath components in the current window, the valid tap support set and corresponding local channel coefficients of the previous window are directly used for equalization compensation in the current window.

[0014] A communication channel equalization compensation system, comprising: The time-domain sampling module is used to sample the received signal in the time domain to obtain the baseband digital signal sequence; The sliding window truncation module is used to truncate the baseband digital signal sequence segment by segment according to the symbol reception order using a sliding time-domain window, and obtain the signal segment within the current window; The channel estimation module is used to estimate the channel impulse response within the current window based on the signal segment and the known training sequence, and to obtain the channel impulse response coefficients at each tap position. The tap support set positioning module is used to determine the time-domain energy metric value of each tap position within the window based on the channel impulse response coefficient, and to mark the tap positions whose time-domain energy metric value exceeds a preset energy threshold as candidate tap support sets. The local channel reconstruction module is used to take the received signal segments within the range of the candidate tap support set as observation vectors, the time-shifted version of the transmitted signal as dictionary atoms, and use the orthogonal matching pursuit algorithm to select matching atoms from the atom set corresponding to the candidate tap support set and reconstruct the sparse representation of the channel impulse response to obtain the reconstructed local channel coefficients and effective tap support set. The equalization compensation output module is used to perform equalization compensation on the signal segment within the current window based on the reconstructed local channel coefficients, and output the equalized symbol sequence.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention employs a sliding time-domain window to segment the signal, independently estimating the channel impulse response within each window. Based on this, it uses a threshold comparison of time-domain energy metrics to initially screen tap positions, marking taps exceeding the threshold as candidate tap support sets. This eliminates a large number of low-energy and purely noisy taps before entering the orthogonal matched pursuit (ORP) reconstruction phase, significantly reducing the search space for subsequent sparse reconstruction and decreasing the iteration count and multiply-accumulate computation of the ORP algorithm. Furthermore, ORP atom selection and sparse reconstruction of the channel impulse response are performed only within the atom sets corresponding to the candidate tap support sets, focusing the reconstruction process on tap positions with concentrated energy. This reduces the possibility of noisy atoms being mistakenly selected in the global search, improving the accuracy of channel coefficient estimation while lowering the false alarm probability. Simultaneously, by performing segmentation of the effective tap support sets between adjacent sliding windows... The continuity verification of the number of overlapping taps is performed. When the channel characteristics change gradually, the support set results of the previous window are directly used to avoid redundant calculations. When the changes are drastic, a repositioning process is triggered and the energy threshold is adjusted to half of the previous window to reduce the positioning threshold and capture newly emerging multipath components. A reasonable balance is achieved between computational efficiency and channel tracking sensitivity. Preferably, the present invention also uses a first sliding window and a second sliding window of different lengths for parallel sliding monitoring. It utilizes the complementary advantages of the stability of the long window and the sensitivity of the short window. The composition of the final candidate support set is dynamically determined by the intersection decision result of the candidate tap support sets of the two windows. When the intersection is insufficient, the positioning result of the short window is used as the final candidate support set and the length of the long window is shortened simultaneously. This achieves adaptive adjustment of the window length and further improves the robustness and equalization performance of the system under time-varying channels. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 This is a schematic diagram of the overall system flow of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] like Figures 1 to 2 As shown, the present invention provides a communication channel equalization compensation method, comprising the following steps: The receiving end performs time-domain sampling on the received signal to obtain the baseband digital signal sequence; The baseband digital signal sequence is segmented according to the symbol reception order using a sliding time-domain window to obtain the signal segment within the current window; Based on the signal segment within the current window and the known training sequence, estimate the channel impulse response within the window and obtain the channel impulse response coefficients at each tap position; Based on the channel impulse response coefficient, determine the time-domain energy metric value of each tap position within the window; and mark the tap positions whose time-domain energy metric value exceeds a preset energy threshold as a candidate tap support set; The received signal segments within the candidate tap support set are used as observation vectors, and the time-shifted version of the transmitted signal is used as dictionary atoms. The orthogonal matching pursuit algorithm is used to select matching atoms from the atom set corresponding to the candidate tap support set and reconstruct the sparse representation of the channel impulse response to obtain the reconstructed local channel coefficients and effective tap support set. Based on the reconstructed local channel coefficients, equalization compensation is performed on the signal segments within the current window, and the equalized symbol sequence is output.

[0019] Specifically, in one embodiment of the present invention, after the communication channel equalization compensation method starts to be executed, the analog-to-digital converter at the receiving end samples the analog signal received by the antenna at a set sampling frequency and converts the analog signal into a baseband digital signal sequence for output.

[0020] The baseband digital signal sequence enters the sliding window processing pipeline. At the start of signal processing, the receiver positions the sliding time-domain window at the beginning of the baseband digital signal sequence, capturing a signal segment within the continuous symbol period covered by the current window. This window slides forward on the signal sequence according to the symbol reception order. After processing each window, the window slides forward by a set step size, entering the processing cycle of the next window, until the entire signal sequence has been processed. There is an overlapping area between adjacent windows; the processing results of this overlapping area are used for parameter transfer between adjacent windows.

[0021] Within each window period, the receiver performs channel estimation using the training sequence within that window. The training sequence is inserted at the header of the data frame at the transmitter, and the receiver stores a copy of the same training sequence locally. The receiver performs cross-correlation between the local training sequence and the corresponding data in the received signal segment. Based on the position of the peak in the cross-correlation result, the arrival delay of each multipath component is determined. Based on the amplitude and phase of the peak, the amplitude attenuation and phase rotation of each multipath component are determined, resulting in a set of channel impulse response coefficients arranged by tap index. Each coefficient value in this sequence corresponds to a delay tap position, reflecting the channel gain of the multipath component at that delay.

[0022] After obtaining the channel impulse response coefficients, the receiver performs energy measurement on each tap position. For each tap position within the current window, the receiver calculates the cumulative energy value based on its channel impulse response coefficients over multiple consecutive symbol periods. Specifically, the receiver takes the channel impulse response coefficients for each tap position over each symbol period, calculates the sum of the squares of the real and imaginary parts of each coefficient to obtain the instantaneous energy for each symbol period, and accumulates the instantaneous energy over multiple symbol periods. The accumulated result is used as the time-domain energy measurement value for that tap position. Through the above calculations, each tap position within the current window obtains a corresponding time-domain energy measurement value, and the measurement values ​​of all tap positions constitute the energy distribution sequence of the window.

[0023] The receiver compares the time-domain energy metric value at each tap position with a preset energy threshold. (Preset energy threshold) The method for determining it is as follows: The receiver calculates the temporal energy metric for all tap positions within the current window and obtains the maximum value. The receiver simultaneously estimates the noise power within the current window. The estimation method involves selecting several tap positions with the lowest energy metric values ​​within the current window, calculating their average energy, and using this average energy as an estimate of the noise power.

[0024] The receiver calculates the first reference value. ,in In this embodiment, the noise power amplification factor is... The value is set to 4, which ensures the threshold is higher than the noise floor, preventing noise taps from being misidentified as valid signal taps. The receiver calculates the second reference value. ,in In this embodiment, the preset proportional threshold coefficient is used. The value is 0.1.

[0025] Receiver comparison and The larger of the two values ​​is taken as the preset energy threshold. In this threshold determination method, the noise power estimate ensures that the threshold is not lower than the noise floor, and the maximum energy value ensures that multipath components with strong energy can be effectively detected. Taking the larger of the two values ​​balances the needs of noise suppression and effective signal acquisition. For tap positions whose metric values ​​exceed the threshold, the receiver marks them as candidate tap support sets; for tap positions whose metric values ​​do not exceed the threshold, the receiver determines that they do not have sufficient signal energy and excludes them from the subsequent reconstruction process.

[0026] The receiver uses received signal segments within the candidate tap support set as observation vectors and time-shifted versions of the transmitted signal as dictionary atoms, employing an orthogonal matching pursuit algorithm for channel reconstruction. During the algorithm's iteration, the receiver restricts atom selection to the atom set corresponding to the candidate tap support set, avoiding searches outside this set. In each iteration, the algorithm selects the atom with the largest inner product with the current residual from the atom set corresponding to the candidate support set, adds this atom to the selected atom set, updates the channel coefficient estimate using the least squares method, and then updates the residual vector. The algorithm terminates when the residual energy falls below a preset residual threshold or the number of iterations reaches a preset maximum. Upon termination, the algorithm outputs the reconstructed local channel coefficients, with the tap position corresponding to these local channel coefficients forming the effective tap support set.

[0027] The receiver constructs the linear equalizer tap weights for the current window based on the reconstructed local channel coefficients. The number of taps in this equalizer is consistent with the number of symbol periods covered by the sliding time-domain window. The receiver inputs the signal segment within the current window into the equalizer, and the equalizer tap weights are convolved with the signal segment to cancel inter-symbol interference introduced by the multipath channel, outputting the equalized symbol sequence.

[0028] After completing the equalization output for the current window, the receiver passes the effective tap support set and local channel coefficients of the current window to the next window. The sliding time-domain window moves forward by a set step size, entering the next processing cycle, and repeating the above steps of channel estimation, energy measurement, threshold filtering, orthogonal matched pursuit reconstruction, and equalization compensation until the entire baseband digital signal sequence has been processed. The sliding of the window drives the continuous operation of the entire processing flow, and the processing results are connected between adjacent windows through the transfer of support set information and channel coefficient information.

[0029] The sliding time-domain window covers L consecutive symbol periods. The sliding time-domain window slides forward once every M new symbols received, where M is no greater than 1 / 2 of L.

[0030] Specifically, during the operation of the sliding time-domain window, the window length L is configured according to the system's operating scenario. In one application scenario, the communication system operates indoors or in a low-speed mobile environment, where the channel coherence time is relatively long and the multipath delay spread is relatively stable. In this case, the window length L is configured as 64 symbol periods. The longer window can accumulate more training sequence samples, resulting in higher statistical estimation accuracy of the channel impulse response coefficient. In another application scenario, the communication system operates in a high-speed mobile environment, such as high-speed rail communication, where the channel coherence time is short and channel characteristics change rapidly. In this case, the window length L is configured as 16 symbol periods. The shorter window can respond to channel changes more quickly, avoiding estimation distortion caused by significant changes in channel characteristics within the window. In a third application scenario, the communication system operates in a moderate-speed urban environment, where the window length L is configured as 32 symbol periods.

[0031] When the window length L is 64, the current window covers 64 symbol periods. Within these 64 symbol periods, the receiver completes all operations including channel estimation, energy measurement, candidate support set localization, orthogonal matching pursuit reconstruction, and equalization compensation. When the window length L is 16, the current window covers 16 symbol periods, and the computational load within each window is approximately one-quarter of that when L is 64. When the window length L is 32, a compromise is achieved between the two considerations mentioned above.

[0032] The selection of the window length L is also related to the preset maximum number of iterations of the Orthogonal Matching Pursuit algorithm. When L is 64, half of the total number of tap positions in the candidate tap support set is 32, meaning the maximum number of iterations for the Orthogonal Matching Pursuit algorithm is 32. When L is 16, the maximum number of iterations is 8. When L is 32, the maximum number of iterations is 16.

[0033] The sliding time-domain window slides forward once for every M new symbols received. In this embodiment, M is configured as half of L. When L is 64, M is 32, and the number of overlapping symbols between adjacent windows is 32. When L is 16, M is 8, and the number of overlapping symbols between adjacent windows is 8. When L is 32, M is 16, and the number of overlapping symbols between adjacent windows is 16.

[0034] There is an overlapping region between adjacent windows, and the symbols within the overlapping region participate in the channel estimation and equalization processing of the preceding and following windows. The effective tap support set output by the preceding window after processing the overlapping region is passed to the current window through the overlapping region, serving as a reference benchmark for the continuity verification of the current window.

[0035] In actual operation, the receiver maintains a symbol counter to record the number of new symbols arriving. This counter increments each time a symbol is received. When the counter value reaches M, the receiver triggers a window sliding operation, moving the starting symbol position of the current window forward by M symbols, and simultaneously resetting the counter to 0 to begin counting the next round of new symbols.

[0036] The values ​​of window length L and sliding step size M satisfy the constraint that M is no greater than half of L. In this embodiment, M is half of L, meaning the overlap ratio of adjacent windows is 50%, the previous window and the current window share half of the symbol period, and the continuity of the effective tap support set between adjacent windows is relatively high. In other embodiments, M is one-quarter of L, the overlap ratio of adjacent windows is 75%, the window sliding is smoother, the amount of support set information transmitted between adjacent windows is larger, and the probability of passing the continuity check is higher, but the number of windows that need to be processed per unit time increases accordingly. In scenarios that require faster response to channel changes, M being half of L is the preferred configuration. In this embodiment, M being half of L achieves a balance between computational burden and tracking sensitivity.

[0037] When the window length L is 64 and M is 32, the valid tap support set of the previous window is passed to the current window within the overlapping area. The current window first performs a continuity check. If the check passes, the support set is directly used without triggering a relocation process. If the channel changes rapidly between adjacent windows, causing the check to fail, a relocation process is triggered.

[0038] When the window length L is 16, the number of samples within a single window is relatively small due to the short symbol period covered by the window, resulting in a relatively large variance in the estimated channel impulse response coefficient. Therefore, the preset proportional threshold coefficient α in the preset energy threshold is configured as 0.2 to improve tolerance to noise fluctuations. When the window length L is 64, the number of samples is larger, and the estimation variance is relatively smaller, so the preset proportional threshold coefficient α is configured as 0.05. When the window length L is 32, the preset proportional threshold coefficient α is configured as 0.1, achieving a trade-off among the three configurations.

[0039] After each window slide, the receiver uses the new window's starting position as a reference to extract L consecutive symbols from the baseband digital signal sequence starting at that position, forming a signal segment for the current window. This signal segment serves as input data for subsequent channel estimation steps. Simultaneously, the effective tap support set and local channel coefficients output when the previous window was processed serve as reference data for continuity verification and full noise fallback processing of the current window. Inter-frame parameter transfer is achieved through the temporal correspondence of overlapping regions.

[0040] The method for determining the time-domain energy metric is as follows: The amplitudes of the channel impulse response coefficients at each tap position within the current window are squared and accumulated, and the accumulated result is used as the time-domain energy metric for that tap position.

[0041] Specifically, the determination of the time-domain energy metric is performed after the channel estimation step is completed. The receiver has obtained the channel impulse response coefficients at each tap position within the current window. These coefficients are in complex form and include both real and imaginary parts.

[0042] For the current window's first The receiver obtains the tap position in a continuous sequence. The channel impulse response coefficient sequence over symbol periods is represented as: ,in The value of is consistent with the total number of symbols in the training sequence within the current window. In this embodiment, the window length is... When the value is 32, the total number of symbols in the training sequence is 16, therefore Take 16.

[0043] For the Channel impulse response coefficient over one symbol period The receiver calculates the square of the modulus, which is the sum of the squares of the real and imaginary parts, to obtain the tap position in the symbol period. The instantaneous energy. The receiver... to Repeat the above calculation for all symbol periods to obtain The instantaneous energy value corresponding to each symbol period.

[0044] The receiving end will the above The instantaneous energy values ​​are accumulated, and the accumulated result is the tap position. Time-domain energy metric The cumulative relationship is expressed as Through the above calculations, each tap position within the current window obtains a corresponding time-domain energy metric value, and the metric values ​​of all tap positions constitute the energy distribution sequence of the window.

[0045] For example, suppose the coefficient with tap index 1 in the current window has the following values ​​over three consecutive symbol periods: , , Then the receiving end calculates the first symbol period. Calculate the second symbol period Calculate the third symbol period Summing 0.34, 0.40, and 0.32 yields 1.06, which is the time-domain energy metric value at tap position 1. The receiving end performs the above calculations one by one for each tap position within the window to obtain the corresponding time-domain energy metric value.

[0046] The calculation cycle of the energy metric is consistent with the processing cycle of the sliding window. That is, after the channel estimation is completed, the time-domain energy metric of all tap positions within the window is calculated independently. After the current window slides to the next position, the receiver re-executes the above square accumulation calculation for each tap position in the new window to obtain the energy distribution sequence of the new window. The energy metric of the previous window is not carried over across windows.

[0047] The magnitude of the time-domain energy metric reflects the energy intensity of the multipath component at the corresponding tap position. Tap positions with larger energy metric values ​​correspond to stronger multipath paths, while tap positions with smaller energy metric values ​​correspond to weaker paths or pure noise positions. A preset energy threshold is applied to this energy distribution sequence to distinguish tap positions with energy metric values ​​exceeding the threshold from those with values ​​below the threshold.

[0048] When window length When changes occur, the total number of symbols in the training sequence As the changes occur, the number of times the energy measurement value is accumulated is adjusted accordingly. It was 16:00. The value is 8, and the energy metric is the sum of 8 symbol periods. When it is 64, The value is 32, meaning the energy metric is the result of accumulation over 32 symbol periods. The more accumulations, the more significant the noise suppression effect of the energy metric, but the computational cost increases accordingly.

[0049] The iteration termination condition of the orthogonal matching pursuit algorithm is: The residual energy is lower than the preset residual threshold or the number of iterations reaches the preset maximum number of iterations; The preset maximum number of iterations is equal to half the total number of tap positions in the candidate tap support set.

[0050] Specifically, the iterative process of the orthogonal matching pursuit algorithm begins after the candidate tap support set is formed. The receiving end uses the candidate tap support set as a constraint on the range of atom selection, and performs iterative atom selection and residual update within this range.

[0051] The iteration termination condition of the orthogonal matching pursuit algorithm is either that the residual energy is lower than a preset residual threshold or that the number of iterations reaches a preset maximum number of iterations. These two conditions are OR logically related; the iteration terminates when either condition is met. In one case, if the current residual energy... It has fallen below the preset residual threshold. This indicates that the fitting accuracy of the currently selected atom set to the observation vector meets the requirements, even with fewer iterations. The preset maximum value has not yet been reached. The algorithm also terminates its iteration. In another case, if the number of iterations... The preset maximum number of iterations has been reached. But residual energy Not yet below the preset residual threshold This indicates that the algorithm failed to achieve the required fitting accuracy within the specified maximum number of iterations. In this case, the algorithm uses the least squares solution corresponding to the currently selected set of atoms as the final reconstruction result.

[0052] Preset residual threshold Set as the total energy of the signal segment within the current window. 1%. Total energy of the signal segment within the current window. The calculation method is as follows: The received signal amplitudes of all symbols within the current window are squared and summed, and the sum is used as the total signal energy. A preset residual threshold is set. With total signal energy The relationship is represented as For example, the total energy of the signal segment within the current window. If the value is 1000 units, then the preset residual threshold is... It is 10 units. When the residual energy When the value decreases to below 10 units, the residual condition is met, and the algorithm terminates.

[0053] Preset maximum number of iterations Equal to the total number of candidate tap support cluster tap positions Half of. The number of tap positions contained in the candidate tap support set is denoted as 1 / 2. Then the maximum number of iterations is preset. For example, when the candidate tap support set contains 10 tap positions, the maximum number of iterations is preset. The maximum number of iterations is 5, meaning the algorithm performs a maximum of 5 iterations, selecting at most 5 atoms from the 10 atoms in the candidate tap support set for reconstruction. The maximum number of iterations is preset when the candidate tap support set contains 16 tap positions. The maximum number of iterations is 8. This is the preset maximum number of iterations when the candidate tap support set contains 6 tap positions. It is 3 times. When When it is an odd number, Round down to the nearest integer. When the value is 7, the maximum number of iterations is preset. For 3 times, When the value is 15, the maximum number of iterations is preset. It was 7 times.

[0054] In each iteration, the receiver selects the atom with the largest inner product with the current residual from the atom set corresponding to the candidate tap support set. In the first iteration, the current residual is the initial observation vector. This refers to the received signal segment within the current window. The receiver calculates the observation vector. The receiver calculates the dot product of the atom with each atom in the candidate set, and selects the atom with the largest absolute value of the dot product to add to the selected atom set. Starting from the second iteration, the receiver first uses the selected atom set to analyze the observation vector. Perform least squares projection and calculate the projection residuals. Then calculate the residual. The atom with the largest absolute value of the inner product of the atom and the candidate set is selected and added to the set of selected atoms.

[0055] The total number of atoms in the set of atoms corresponding to the candidate tap support set is As the iterations proceed, the number of selected atoms gradually increases, while the number of remaining atoms in the candidate set gradually decreases. When the number of iterations... achieve At that time, the number of atoms selected was At this point, the algorithm terminates, and even if there are still unselected atoms in the candidate set, these atoms will no longer participate in subsequent calculations. When the residual energy... In the number of iterations Not achieved The time has fallen below the preset residual threshold The algorithm terminates prematurely and stops iterating.

[0056] When the total number of candidate tap support concentrated tap positions When the value is 0, it means that the temporal energy metric value of no tap position in the current window exceeds the preset energy threshold, and the orthogonal matching pursuit algorithm does not perform iteration.

[0057] In a typical implementation scenario, if the candidate tap support set contains 12 tap positions, then the maximum number of iterations is preset. The process is repeated 6 times. The receiving end performs the first iteration, selecting the atom with the largest inner product from the 12 atoms. The second iteration selects the atom with the largest inner product from the remaining 11 atoms. The third, fourth, and fifth iterations each select one atom from the remaining atoms. This continues until the fifth iteration, where the residual energy... The value is 8 units, which is lower than the preset residual threshold. If the residual energy is less than 10 units, the algorithm terminates after the 5th iteration and does not execute the 6th iteration. The value is 15 units, which is still higher than the preset residual threshold. If the number of units reaches 10, the algorithm continues with the 6th iteration, and terminates after the iteration is completed.

[0058] After the iteration terminates, the receiver outputs the channel coefficient estimate updated in the last iteration as the reconstructed local channel coefficients. The tap positions corresponding to these channel coefficients constitute the effective tap support set. The number of taps in the effective tap support set is equal to the actual number of iterations performed, and not greater than the total number of tap positions in the candidate tap support set. 1 / 2 of.

[0059] The equalization compensation constructs the tap weights of the linear equalizer based on the reconstructed local channel coefficients, and performs zero-forcing equalization or minimum mean square error equalization on the signal segments within the current window.

[0060] Specifically, after obtaining the reconstructed local channel coefficients, the receiver constructs the tap weights of the current window linear equalizer based on these coefficients. The number of taps in this linear equalizer is consistent with the number of symbol periods covered by the sliding time-domain window; that is, the length of the equalizer is the window length. Each tap corresponds to a delay of one symbol period.

[0061] The receiver arranges the reconstructed local channel coefficients into a channel vector according to the tap index. In this vector, the coefficients at positions corresponding to non-valid tap support sets are 0, while the coefficients at positions corresponding to valid tap support sets are the reconstructed values ​​output by the orthogonal matching pursuit algorithm. The receiver then uses this channel vector... Constructing the equalizer weight vector .

[0062] In this embodiment, the linear equalizer employs a zero-forcing equalization method. The weight vector for zero-forcing equalization... The construction method is as follows: the receiver first determines the channel vector. Constructing the Toeplitz channel convolution matrix The number of rows in this matrix is ​​equal to the window length. The number of columns is Add 1 less than the channel memory length. The receiver calculates the channel convolution matrix. pseudo-inverse matrix The elements in the center row of the pseudo-inverse matrix are selected as the tap weights of the equalizer. The weight vector The received signal vector of the current window Perform convolution operations and output the symbol sequence after zero-forcing equalization. In zero-forcing equalization mode, the equalizer completely eliminates inter-symbol interference (ISI), making the ISI component at the equalizer output zero.

[0063] In another embodiment, the linear equalizer employs a minimum mean square error (MSE) equalization method. The weight vector for minimum mean square error equalization is shown below. The construction method is as follows: the receiver first determines the channel vector. Constructing the Toeplitz channel convolution matrix Simultaneously estimate the noise power within the current window. The receiver calculates the autocorrelation matrix. ,in for The conjugate transpose of the matrix. Let be the identity matrix, and the dimension of the identity matrix is ​​the same as... The dimensions are the same. The receiver calculates the cross-correlation vector. ,in This is a unit vector, with a 1 at its center and 0s at the other positions. The receiving end solves a system of linear equations. The solution This refers to the tap weights of the minimum mean square error equalizer. This weight vector... The received signal vector of the current window The convolution operation is performed, and the output is a symbol sequence equalized with minimum mean square error. Under the minimum mean square error equalization method, the equalizer achieves a balance between eliminating intersymbol interference and suppressing noise amplification, thus minimizing the mean square error at the output.

[0064] Zero-forcing equalization and minimum mean square error equalization are two parallel equalization schemes, with the receiver selecting one based on system configuration. When the system operates in a high signal-to-noise ratio (SNR) environment, noise amplification has a relatively small impact on equalization performance, so the receiver chooses zero-forcing equalization, which requires less computation and eliminates the need for noise power estimation. When the system operates in a low SNR environment, zero-forcing equalization may amplify noise, leading to performance degradation. In this case, the receiver chooses minimum mean square error equalization, which introduces a noise power estimate to suppress noise amplification during the equalization process.

[0065] In a typical implementation scenario, the window length The value is 32. The reconstructed local channel coefficients contain three valid tap support set positions, located at tap indices 2, 5, and 9, respectively, with corresponding reconstructed channel coefficient values ​​of 32. , and The coefficients at the remaining tap positions are all 0. The receiver arranges this channel coefficient sequence into a channel vector of length 32. The second element is The 5th element is The 9th element is All other elements are 0. The receiver uses this channel vector. Constructing the equalizer weight vector Weight vector The length is 32. Equalizer weight vector. With the current window received signal vector Perform convolution operations, and the output of the convolution is an equalized sequence of symbols.

[0066] Equalizer weight vector Once the reconstructed local channel coefficients based on the current window are completed, they will be used for all channels within the current window. Equalization processing for one symbol period. After the current window slides to the next position, the receiver reconstructs a new equalizer weight vector based on the reconstructed local channel coefficients of the new window, and the equalizer weight vector of the previous window is no longer used.

[0067] Perform continuity checks on the valid tap support set between two adjacent sliding windows: If the number of overlapping taps between the current window's valid tap support set and the previous window's valid tap support set is greater than or equal to a preset percentage threshold of the total number of taps in the previous window's valid tap support set, then the current window's valid tap support set remains unchanged. Otherwise, a relocation process is triggered, which includes lowering a preset energy threshold and recalculating the time-domain energy metric.

[0068] Specifically, after the receiver completes local channel reconstruction and outputs a valid tap support set in the current window, it compares the valid tap support set of the current window with that of the previous window to determine the continuity of the channel multipath structure between adjacent windows. This continuity check is performed after the orthogonal matching pursuit algorithm iteration of the current window terminates and outputs a valid tap support set, and before entering the next window processing cycle.

[0069] The valid tap support set of the current window is denoted as The effective tap support set of the previous window is denoted as . and This is a set of tap indices, where each element corresponds to a tap position. The receiving end calculates... and The intersection, that is , The element in the table represents the tap position that appears in both windows simultaneously. Size is denoted as This refers to the number of taps shared by both windows. Size is denoted as That is, the total number of taps in the valid tap support set of the previous window.

[0070] Receiver calculation occupy proportion The receiving end will Compared with the preset ratio threshold A comparison is made. In this embodiment, The value is 0.7. When When the value is greater than or equal to 0.7, it indicates that the effective tap support set of the current window highly overlaps with that of the previous window, the channel multipath structure remains basically unchanged between adjacent windows, and the receiver determines that the continuity check has passed. When the value is less than 0.7, it indicates that the overlap between the effective tap support set of the current window and the previous window is insufficient, the channel multipath structure has changed significantly, and the receiver determines that the continuity check fails.

[0071] If the continuity check passes, the receiver maintains the valid tap support set of the current window unchanged, i.e., it does not trigger a relocation process, and the current window continues to use the same valid tap support set in the processing cycle of the next window. At the same time, the local channel coefficients of the current window also remain unchanged. In the processing cycle of the next window, the valid tap support set and local channel coefficients of the current window are passed to the next window as data from the previous window.

[0072] If the continuity check fails, the receiver triggers a relocation process. The relocation process is executed as follows: the receiver sets a preset energy threshold... Adjust to the preset energy threshold used in the previous window. of ,Right now The receiver uses a reduced preset energy threshold. Within the current window, re-perform the threshold comparison of the time-domain energy metric, and select metric values ​​exceeding the threshold. The tap positions are then re-marked as a new candidate tap support set. Subsequently, the receiver re-performs orthogonal matched pursuit reconstruction based on the new candidate tap support set, outputting a new effective tap support set and new local channel coefficients.

[0073] In a typical implementation scenario, the effective tap support set of the previous window There are 4 tap positions, including tap positions {2, 5, 9, 14}. The current window's valid tap support set is 4. Includes tap positions {2, 5, 9, 16}. The receiver calculates... and The intersection, The expression is {2, 5, 9}. The value is 3. The receiving end calculates... If the value is greater than the preset ratio threshold of 0.7, the continuity check passes, and the receiving end maintains the current window's valid tap support set {2, 5, 9, 16} unchanged.

[0074] In another typical implementation scenario, the effective tap support set of the previous window There are 4 tap positions, including tap positions {2, 5, 9, 14}. The current window's valid tap support set is 4. There are 3 tap positions, including tap positions {3, 6, 10}. The receiver calculates... and The intersection, It is an empty set. The value is 0. The receiver calculates... If this value is less than the preset ratio threshold of 0.7, the continuity check fails, and the receiver triggers a relocation process. The receiver will then set the preset energy threshold. Adjust to the preset energy threshold used in the previous window. Then, re-execute the energy metric, candidate support set labeling, and orthogonal matching tracing reconstruction.

[0075] During the continuity check, if the previous window's valid tap support set If the previous window is an empty set (i.e., the previous window is a completely noisy window), then the current window will not undergo continuity verification and will directly trigger the relocation process, independently determining the candidate tap support set based on the energy measurement results of the current window itself.

[0076] In window length and sliding step size Once determined, the preset ratio threshold for continuity verification is set. It is preset according to the system configuration. In this embodiment... Using 0.7 is suitable for scenarios requiring higher stability. A value of 0.8 is used in scenarios requiring higher sensitivity. Take 0.6. The value of is determined during system initialization and remains unchanged during subsequent operation.

[0077] When the relocation process is triggered: The preset energy threshold is adjusted to half of the preset energy threshold used in the previous window, the time-domain energy metric is recalculated, and the candidate tap support set is updated. The adjustment stops when the preset energy threshold is reduced to the preset lower limit.

[0078] Specifically, when the receiving end triggers a relocation process after failing the continuity check, it performs an adjustment operation on a preset energy threshold. The triggering conditions for the relocation process have been described in detail; this embodiment explains the specific execution process of threshold adjustment during the relocation process.

[0079] When the receiving end triggers the relocation process, it reads the preset energy threshold used in the previous window and records this threshold as... The receiver calculates a new preset energy threshold. The calculation formula is: This means reducing the preset energy threshold used in the previous window to half of its original value. The receiving end then updates the preset energy threshold of the current window to... Using the updated threshold, re-perform the threshold comparison of the time-domain energy metric within the current window, and re-apply the metric value exceeding the threshold. The tap positions are marked as new candidate tap support sets. Subsequently, the receiver re-performs orthogonal matched pursuit reconstruction based on the new candidate tap support sets, outputting a new effective tap support set and new local channel coefficients. The new effective tap support set and local channel coefficients are used for subsequent continuity check and equalization processing.

[0080] During the threshold adjustment process, the receiver monitors the rate of decrease in the preset energy threshold. The receiver pre-sets a preset lower limit. This preset lower limit is used to restrict the threshold from decreasing indefinitely. In this embodiment, the preset lower limit... Based on the noise power estimate within the current window Confirmed, specifically The preset lower limit is four times the noise power estimate. This value ensures that the threshold is always higher than the noise floor, thus avoiding misjudging noise taps as valid signal taps.

[0081] After each triggering of the relocation process and reduction of the preset energy threshold, the receiver will update the threshold. With preset lower limit Comparison. When Greater than or equal to At that time, the receiving end uses As the preset energy threshold for the current window, the threshold comparison and candidate support set marking operations in the normal repositioning process are executed.

[0082] when Less than At that time, the receiver sets the preset energy threshold of the current window to... It will no longer decrease. At this point, the receiving end uses... Perform threshold comparison and candidate support set labeling operations if using If no temporal energy metric value at any tap position is detected to exceed the threshold, it is determined that there is no effective multipath component in the current window, and the full noise window processing strategy is executed.

[0083] In a typical implementation scenario, the preset energy threshold used in the previous window... The noise power estimate for the current window is 0.8. The lower limit is 0.05. Calculated as The first time a relocation is triggered, , Greater than The receiver uses 0.4 as the preset energy threshold for the current window. If the continuity check still fails after using 0.4, a second repositioning trigger will be initiated. , equal The receiver uses 0.2 as the preset energy threshold for the current window. If no tap position exceeding the threshold is detected after using 0.2, it is determined that there is no valid multipath component within the current window.

[0084] In another typical implementation scenario, the preset energy threshold used in the previous window... The noise power estimate for the current window is 0.4. The lower limit is 0.08. Calculated as The first time a relocation is triggered, , Less than The receiving end directly sets the preset energy threshold of the current window to... That is, 0.32, and will not be further reduced to 0.2, in order to avoid the threshold falling below the noise floor.

[0085] After the repositioning process is completed, the new candidate tap support set and the corresponding local channel coefficients are output to the equalization compensation module. If the continuity check of the next window still fails after the repositioning process, the receiver continues to perform the above threshold adjustment operation until the threshold is reduced to the preset lower limit or the positioning is successful.

[0086] Parallel sliding monitoring of the same received signal sequence is performed using a first sliding window and a second sliding window of different lengths. The length of the first sliding window is greater than the length of the second sliding window, and the two sliding windows slide synchronously with the same length M. When the size of the intersection of the first candidate tap support set located by the first sliding window and the second candidate tap support set located by the second sliding window exceeds a preset ratio threshold of the size of the second candidate tap support set, the intersection is used as the final candidate tap support set for subsequent channel reconstruction. Otherwise, the second candidate tap support set is used as the final candidate tap support set, and the window length of the first sliding window is reduced by M symbols.

[0087] Specifically, the receiving end uses a first sliding window and a second sliding window of different lengths to perform parallel sliding monitoring of the same received signal sequence. The length of the first sliding window is denoted as... The length of the second sliding window is denoted as ,and In this embodiment, The value is 64 symbol periods. The value is 32 symbol periods. The two sliding windows have the same duration. Synchronous sliding, in this embodiment Values of That is, 16 symbol periods. Both the first and second windows start from the beginning of the baseband digital signal sequence and slide forward simultaneously with the same sliding step size. At each window position, channel estimation, energy measurement, threshold screening, and candidate tap support set localization operations are performed independently.

[0088] At each window position, the first sliding window locates the first candidate tap support set. The second sliding window locates the second candidate tap support set. . and Both are sets of tap indices. The receiving end calculates... and The intersection, that is , The element in the table is the tap index that is simultaneously identified as a candidate tap position by both windows. Size is denoted as This refers to the number of candidate taps jointly marked by the two windows. Size is denoted as That is, the total number of candidate tap support sets located by the second sliding window.

[0089] Receiver calculation occupy proportion The receiving end will Compared with the preset ratio threshold A comparison is made. In this embodiment, The value is 0.6, which is 60%. When When the value is greater than or equal to 0.6, it indicates that the candidate support set positioning results of the two windows are highly overlapping, the positioning results are highly reliable, and the receiver determines that the intersection is valid.

[0090] Under the condition that the intersection is valid, the receiving end will As the final candidate tap support set, it is used for subsequent orthogonal matching tracking reconstruction and equilibrium compensation. It retains the combined judgment results of the first window (stable) and the second window (sensitive), while eliminating isolated noise taps that are falsely detected by any single window.

[0091] Under the condition that the intersection is invalid, that is A value less than 0.6 indicates a significant discrepancy between the positioning results of the two windows, possibly due to rapid channel changes or sudden interference in the current window region. In this case, the positioning result of the second window (shorter and more sensitive) is more reliable, and the receiver will... This serves as the final candidate tap support set. Simultaneously, the receiver reduces the window length of the first sliding window. A symbol, namely ,in This is the sliding step size. In this embodiment, The number of symbols is 16, and the length of the first window is reduced from 64 to 48.

[0092] After the window length is adjusted, the first sliding window will be monitored with the new length in subsequent sliding operations. Continue running, while maintaining the same length. Continue parallel sliding monitoring. If insufficient overlap occurs again at subsequent window positions, the length of the first window continues to decrease. The process continues until the length of the first window is reduced to be equal to the length of the second window. At this point, it will no longer decrease.

[0093] In a typical implementation scenario, the first window length The length of the second window is 64. The step size is 32. The value is 16. At a certain window position, the first candidate tap support set located by the first window. Given 5 tap positions {2, 5, 9, 14, 20}, the second candidate tap support set is located by the second window. The values ​​are {2, 5, 9, 16}, with a total of 4 tap positions. and intersection The expression is {2, 5, 9}. It is 3. It is 4. If the ratio is greater than the preset threshold of 0.6, the receiving end will use the intersection {2, 5, 9} as the final candidate tap support set.

[0094] In another typical implementation scenario, at a certain window position, the first window is positioned... Given {2, 5, 9, 14, 20}, the second window locates... Given the set {3, 6, 10, 16}, the intersection of the two sets is an empty set. =0, The value is less than the preset ratio threshold of 0.6. The receiving end will... That is, {3, 6, 10, 16} is used as the final candidate tap support set, and the length of the first window is reduced from 64 to 48 by 16. Subsequent windows continue to slide in parallel with lengths of 48 and 32.

[0095] When the length of the first window is reduced to 48, if the intersection of the positioning results of the two windows is insufficient again in subsequent window positions, the length of the first window is further reduced by 16 to 32. At this point, the length of the first window is equal to the length of the second window, the two windows are the same size, and the length is no longer reduced when the intersection is insufficient.

[0096] When the time-domain energy metric value exceeds the preset energy threshold if no tap position is detected within the sliding time-domain window: If it is determined that there are no valid multipath components in the current window, the valid tap support set and corresponding local channel coefficients of the previous window are directly used for equalization compensation in the current window.

[0097] Specifically, after the receiver completes the calculation of the temporal energy metric value and the comparison with the preset energy threshold in the current window, it counts the number of tap positions in the current window whose temporal energy metric values ​​exceed the preset energy threshold. When this number is 0, that is, when no tap position in the current window has a temporal energy metric value exceeding the preset energy threshold, the receiver determines that there is no effective multipath component in the current window, and the current window is marked as a full noise window.

[0098] When the criteria for determining the full noise window are met, the receiver does not perform the candidate tap support set marking operation, nor does it perform the orthogonal matching pursuit reconstruction operation. The receiver reads the valid tap support set of the previous window stored in the memory unit. and the corresponding local channel coefficients The receiving end will and It is directly used for equalization compensation in the current window. That is, it uses the effective tap support set of the previous window and the local channel coefficients to construct the linear equalizer tap weights of the current window, performs zero-forcing equalization or minimum mean square error equalization on the signal segment of the current window, and outputs the equalized symbol sequence.

[0099] In a typical implementation scenario, there are a total of 32 tap positions within the current window. The receiver calculates 32 time-domain energy metrics, which are 0.05, 0.03, 0.04, 0.06, 0.02, 0.05, ..., with a preset energy threshold of 0.15. The receiver compares each of these 32 metrics with 0.15. Since all 32 metrics are less than 0.15, and no tap position's metric exceeds the threshold, the receiver determines that there are no valid multipath components within the current window. The receiver then reads the valid tap support set from the previous window. and the corresponding local channel coefficients The receiver uses the four tap positions and corresponding channel coefficients to construct the equalizer for the current window, and performs equalization compensation on the signal segments of the current window.

[0100] When the criteria for a full noise window are met, the candidate tap support set for the current window is denoted as... The effective tap support set is also denoted as The current window does not generate new candidate tap support sets or new local channel coefficients. The effective tap support sets and local channel coefficients of the previous window are directly reused in the current window, but these reused support sets and channel coefficients are not passed to the next window as new outputs of the current window. When the next window processes the data, the effective tap support sets and local channel coefficients of its previous window (i.e., the current window) are the reused data passed from the previous window, not the data generated by the current window itself.

[0101] In another typical implementation scenario, multiple consecutive windows are all identified as full-noise windows. When the first window is identified as a full-noise window, the receiver uses the effective tap support set and local channel coefficients of the 0th window (i.e., the window preceding the 1st window) for equalization compensation, while the effective tap support set and local channel coefficients of the 1st window are stored as a copy of the 0th window's data. When the second window is identified as a full-noise window, the receiver reads the effective tap support set and local channel coefficients of the 1st window (i.e., the copy of the 0th window's data) from the storage unit and uses them for equalization compensation in the 2nd window. This process continues, and during consecutive full-noise windows, the receiver continuously uses the effective tap support set and local channel coefficients of the most recent non-full-noise window for equalization compensation.

[0102] When the criteria for determining the full noise window are met, the receiver does not trigger the relocation process. The relocation process is only triggered when the continuity check fails. During the full noise window, the continuity check is not performed, nor is the threshold adjustment operation performed. Instead, the fallback processing strategy of the full noise window is directly executed.

[0103] A communication channel equalization compensation system, comprising: The time-domain sampling module is used to sample the received signal in the time domain to obtain the baseband digital signal sequence; The sliding window truncation module is used to truncate the baseband digital signal sequence segment by segment according to the symbol reception order using a sliding time-domain window, and obtain the signal segment within the current window; The channel estimation module is used to estimate the channel impulse response within the current window based on the signal segment and the known training sequence, and to obtain the channel impulse response coefficients at each tap position. The tap support set positioning module is used to determine the time-domain energy metric value of each tap position within the window based on the channel impulse response coefficient, and to mark the tap positions whose time-domain energy metric value exceeds a preset energy threshold as candidate tap support sets. The local channel reconstruction module is used to take the received signal segments within the range of the candidate tap support set as observation vectors, the time-shifted version of the transmitted signal as dictionary atoms, and use the orthogonal matching pursuit algorithm to select matching atoms from the atom set corresponding to the candidate tap support set and reconstruct the sparse representation of the channel impulse response to obtain the reconstructed local channel coefficients and effective tap support set. The equalization compensation output module is used to perform equalization compensation on the signal segment within the current window based on the reconstructed local channel coefficients, and output the equalized symbol sequence.

[0104] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0105] In the description of this application, it should be noted that the terms "first", "second", and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0106] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0107] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0108] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0109] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0110] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the technical scope disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

[0111] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

Claims

1. A communication channel equalization compensation method, characterized in that, Includes the following steps: The receiving end performs time-domain sampling on the received signal to obtain the baseband digital signal sequence; The baseband digital signal sequence is segmented according to the symbol reception order using a sliding time-domain window to obtain the signal segment within the current window; Based on the signal segment within the current window and the known training sequence, estimate the channel impulse response within the window and obtain the channel impulse response coefficients at each tap position; Based on the channel impulse response coefficient, determine the time-domain energy metric value for each tap position within the window; The tap positions where the time-domain energy metric value exceeds a preset energy threshold are marked as candidate tap support sets; The received signal segments within the candidate tap support set are used as observation vectors, and the time-shifted version of the transmitted signal is used as dictionary atoms. The orthogonal matching pursuit algorithm is used to select matching atoms from the atom set corresponding to the candidate tap support set and reconstruct the sparse representation of the channel impulse response to obtain the reconstructed local channel coefficients and effective tap support set. Based on the reconstructed local channel coefficients, equalization compensation is performed on the signal segments within the current window, and the equalized symbol sequence is output.

2. The communication channel equalization compensation method according to claim 1, characterized in that, The sliding time-domain window covers L consecutive symbol periods. The sliding time-domain window slides forward once every M new symbols received, where M is no greater than 1 / 2 of L.

3. The communication channel equalization compensation method according to claim 1, characterized in that, The method for determining the time-domain energy metric is as follows: The amplitudes of the channel impulse response coefficients at each tap position within the current window are squared and accumulated, and the accumulated result is used as the time-domain energy metric for that tap position.

4. The communication channel equalization compensation method according to claim 1, characterized in that, The iteration termination condition of the orthogonal matching pursuit algorithm is: The residual energy is lower than the preset residual threshold or the number of iterations reaches the preset maximum number of iterations; the preset maximum number of iterations is equal to 1 / 2 of the total number of tap positions in the candidate tap support set.

5. The communication channel equalization compensation method according to claim 1, characterized in that, The equalization compensation constructs the tap weights of the linear equalizer based on the reconstructed local channel coefficients, and performs zero-forcing equalization or minimum mean square error equalization on the signal segments within the current window.

6. The communication channel equalization compensation method according to claim 1, characterized in that, Perform continuity checks on the valid tap support set between two adjacent sliding windows: If the number of overlapping taps between the current window's valid tap support set and the previous window's valid tap support set is greater than or equal to a preset percentage threshold of the total number of taps in the previous window's valid tap support set, then the current window's valid tap support set remains unchanged. Otherwise, a relocation process is triggered, which includes lowering a preset energy threshold and recalculating the time-domain energy metric.

7. The communication channel equalization compensation method according to claim 6, characterized in that, When the relocation process is triggered: The preset energy threshold is adjusted to half of the preset energy threshold used in the previous window, the time-domain energy metric is recalculated, and the candidate tap support set is updated. The adjustment stops when the preset energy threshold is reduced to the preset lower limit.

8. The communication channel equalization compensation method according to claim 2, characterized in that, Parallel sliding monitoring of the same received signal sequence is performed using a first sliding window and a second sliding window of different lengths. The length of the first sliding window is greater than the length of the second sliding window, and the two sliding windows slide synchronously with the same length M. When the size of the intersection of the first candidate tap support set located by the first sliding window and the second candidate tap support set located by the second sliding window exceeds a preset ratio threshold of the size of the second candidate tap support set, the intersection is used as the final candidate tap support set for subsequent channel reconstruction. Otherwise, the second candidate tap support set is used as the final candidate tap support set, and the window length of the first sliding window is reduced by M symbols.

9. The communication channel equalization compensation method according to claim 1, characterized in that, When the time-domain energy metric value exceeds the preset energy threshold if no tap position is detected within the sliding time-domain window: If it is determined that there are no valid multipath components in the current window, the valid tap support set and corresponding local channel coefficients of the previous window are directly used for equalization compensation in the current window.

10. A communication channel equalization compensation system, characterized in that, For performing the method according to any one of claims 1 to 9, comprising: The time-domain sampling module is used to sample the received signal in the time domain to obtain the baseband digital signal sequence; The sliding window truncation module is used to truncate the baseband digital signal sequence segment by segment according to the symbol reception order using a sliding time-domain window, and obtain the signal segment within the current window; The channel estimation module is used to estimate the channel impulse response within the current window based on the signal segment and the known training sequence, and to obtain the channel impulse response coefficients at each tap position. The tap support set positioning module is used to determine the time-domain energy metric value of each tap position within the window based on the channel impulse response coefficient, and to mark the tap positions whose time-domain energy metric value exceeds a preset energy threshold as candidate tap support sets. The local channel reconstruction module is used to take the received signal segments within the range of the candidate tap support set as observation vectors, the time-shifted version of the transmitted signal as dictionary atoms, and use the orthogonal matching pursuit algorithm to select matching atoms from the atom set corresponding to the candidate tap support set and reconstruct the sparse representation of the channel impulse response to obtain the reconstructed local channel coefficients and effective tap support set. The equalization compensation output module is used to perform equalization compensation on the signal segment within the current window based on the reconstructed local channel coefficients, and output the equalized symbol sequence.

Images