A sub-band full-duplex digital domain shared coefficient self-interference cancellation method

Abstract

The application relates to the technical field of wireless communication and discloses a sub-band full-duplex digital domain shared coefficient self-interference cancellation method, which comprises the following steps: transforming a transceiving baseband signal to a digital frequency domain to obtain a frequency domain observation value and a digital domain reference value; grouping frequency points of a receiving sub-band continuously according to the distance from the sub-band boundary; calculating initial shared complex coefficients and smoothing the initial shared complex coefficients to obtain final shared complex coefficients; performing time domain shared complex coefficient adaptive filtering on the final shared complex coefficients to obtain time domain shared complex coefficients; constructing a residual self-interference estimation value by using the time domain shared complex coefficients and subtracting the residual self-interference estimation value from the receiving frequency domain observation value; and performing cancellation effect monitoring on the output after cancellation and configuring time domain filtering parameters. The application uses fixed continuous grouping to make the shared complex cancellation coefficients of the frequency points in the same group, limits the coefficient mutation by using adjacent smoothing, uses time domain shared complex coefficient adaptive filtering to improve the coefficient stability and cancellation precision by using time correlation, realizes parameter adaptation by using cancellation effect monitoring, reduces the parameter quantity and calculation complexity, and improves the adaptation ability.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, specifically to a method for eliminating self-interference of sub-band full-duplex digital domain shared coefficients. Background Technology

[0002] Subband full-duplex refers to arranging the transmission and reception directions on adjacent or partially adjacent subbands within the same carrier bandwidth, enabling the same node to transmit and receive simultaneously. For the base station side, subband full-duplex does not require complete overlap between the transmission and reception subbands in the frequency domain. However, because they are within the same RF transceiver link system, factors such as nonlinearity of the transmitter power amplifier, limited analog isolation, non-ideal filter transition bands, and errors in digital-to-analog and analog-to-digital conversion can still cause residual self-interference in the reception subband. This residual self-interference, after down-conversion and analog-to-digital conversion at the receiver, typically needs further elimination in the baseband digital domain at the receiver.

[0003] The currently disclosed digital domain self-interference cancellation approaches in sub-band full-duplex mainly involve reconstructing residual self-interference in the receiver's baseband digital domain using the known transmitted baseband signal, and then subtracting it from the received signal. Existing publicly available solutions typically focus on the following paths: first, using a higher-order nonlinear behavior model to jointly model the nonlinearity of the transmitter power amplifier and the self-interference coupling link, and then reconstructing the residual self-interference in the digital domain; second, using data-driven models such as neural networks to fit the residual self-interference; and third, using partial-band reconstruction to reduce the processing overhead of the digital cancellation link. While these methods can improve the residual self-interference suppression effect in sub-band full-duplex scenarios to some extent, they generally suffer from the following problems: First, the model has a large number of parameters, and the training or estimation process is complex, which will bring a high computational burden when real-time processing is required at the receiving end. Secondly, it is highly dependent on the observation length, training samples, or model stability, and the model needs to be readjusted when the subband boundary or service bandwidth changes. Third, the processing effect of some band reconstruction methods depends to a large extent on the accuracy of spectrum reconstruction. When the residual self-interference changes rapidly near the subband boundary, the reconstruction error will affect the elimination result.

[0004] Therefore, without introducing high-order nonlinear full models, relying on complex neural network structures, or employing partial band reconstruction, how to propose a low-complexity digital domain self-interference elimination method in the baseband digital domain of the receiving end, with a clear processing flow, controllable number of parameters, and applicable to the rapidly changing characteristics of residual self-interference near the subband boundary, remains a practical problem to be solved in subband full-duplex technology. Summary of the Invention

[0005] The technical problem to be solved by this invention is: in sub-band full-duplex scenarios, residual self-interference needs to be eliminated in the baseband digital domain at the receiving end, but existing digital domain methods generally have large model parameter scale, complex estimation or training process, and heavy real-time implementation burden. Therefore, a sub-band full-duplex digital domain shared coefficient self-interference elimination method is proposed.

[0006] To address the aforementioned technical problems, this invention provides a method for eliminating self-interference of shared coefficients in the sub-band full-duplex digital domain, comprising the following steps: Step S1: Obtain the received baseband signal and the known transmitted baseband signal of the current processing block, and transform both to the digital frequency domain to obtain the received frequency domain observation value and the digital domain reference value obtained from the known transmitted baseband signal of the local end; Step S2: Group the received subband frequency points into fixed, continuous groups according to their distance from the subband boundary, from closest to furthest. Step S3: For each group, calculate the initial shared complex coefficients corresponding to that group; Step S4: Perform fixed-adjacency smoothing on the initial shared complex coefficients to obtain the final shared complex coefficients corresponding to each group; Step S5: Perform time-domain shared coefficient adaptive filtering on the final shared complex coefficients to obtain the time-domain shared complex coefficients corresponding to each group; Step S6: For any frequency point, construct the residual self-interference estimate of that frequency point using the time-domain shared complex coefficients of the group, and subtract it from the received frequency domain observation to obtain the eliminated received sub-band frequency domain output. Step S7: Monitor the cancellation effect of the frequency domain output of the canceled receiving sub-band, and configure the time domain filtering parameters of the subsequent processing block according to the monitoring results.

[0007] Further, in step S1, the received baseband signal and the known transmitted baseband signal at this end are respectively processed... The point-based discrete Fourier transform is calculated using the following formula: in, The received frequency domain observation value, The reference value for the digital field is... For the received baseband discrete signal of the current processing block, To transmit baseband discrete signals at this end. =0,1,…, -1, This represents the number of sampling points in the current processing block.

[0008] Further, in step S1, the set of frequency points corresponding to the received sub-band is: in, The starting frequency point in the receive sub-band that is closest to the boundary between the transmit and receive sub-bands. This represents the total number of frequency points included in the receiving subband.

[0009] Further, in step S2, starting from the starting frequency point, continuous grouping is performed along the receiving sub-band direction to obtain continuous and non-overlapping groups. Each group is determined according to the following formula: Obtain continuous and non-overlapping groupings: ... in, From before removal The remaining frequency points after the group constitute the group. The length of the first group, For the first Group length, Group number This represents the total number of groups.

[0010] Further, in step S3, for any group, a least-squares estimate is constructed using the received frequency domain observations and digital domain reference values ​​of all frequency points within the group, and the initial shared complex coefficients corresponding to the group are calculated according to the following formula: in, To receive frequency domain observations, For reference values ​​in the numeric field, express conjugate, Here is the regularization constant. For the first Groups, For the first The initial shared complex coefficients of the nth group, wherein the initial shared complex coefficients represent the initial shared complex coefficients of the nth group. The residual self-interference equivalent digital domain coefficients shared by multiple frequency points within the group.

[0011] Further, in step S4, the initial shared complex coefficients undergo a fixed-adjacent smoothing process, with a smoothing factor of 1. Perform the following steps: in, As a smoothing factor, For the first The initial shared complex coefficients of each group are smoothed to obtain the final shared complex coefficients of each group. .

[0012] Furthermore, in step S5, the time-domain shared coefficient adaptive filtering adopts a time-domain recursive filtering method, which is executed according to the following formula: in, This is the sequence number of the current processing block. The final shared complex coefficients obtained by the current processing block in step S4. These are the time-domain shared complex coefficients of the previous processing block after time-domain filtering. For time-domain recursion factors, For the current processing block Time-domain shared complex coefficients for each group.

[0013] Further, in step S5, the time-domain recursive factor Measured by the difference in normalized coefficients between the current processing block and the previous processing block. Determine, calculate according to the following formula: in, Let be the time-domain sensitivity constant. This is a regularization constant; when the coefficient difference is small, i.e., the channel change is gradual, The value approaches 0; when the coefficient difference is large, i.e., the channel changes rapidly. The value approaches 1.

[0014] Further, in step S5, when the difference in the normalized coefficients is measured... Greater than the preset threshold season = Skip the time-domain recursive filtering and use the final shared complex coefficients of the current processing block as the time-domain shared complex coefficients.

[0015] Further, in step S6, for any frequency point, using the time-domain shared complex coefficients of the group to which the frequency point belongs, the residual self-interference estimate of the frequency point is constructed according to the following formula and digital domain cancellation is performed: in, Frequency point number, For frequency point The group number it belongs to, For the current processing block Time-domain shared complex coefficients of each group, The reference value for the digital field is... The received frequency domain observation value, For frequency point The residual self-interference estimate, The received frequency domain value after elimination; after all frequency points have undergone the above processing, the eliminated frequency domain output of the received sub-band is obtained. .

[0016] Further, in step S7, the elimination gain is calculated according to the following formula: in, The received frequency domain observations before elimination, The received frequency domain value after elimination. For receiving sub-band frequency points, To eliminate gain; according to the said gain elimination The comparison result with the preset threshold configures the temporal filtering parameters of subsequent processing blocks: when When the value is less than the first preset threshold, maintain the time-domain sensitivity constant. Unchanged; when When the value is not less than the first preset threshold and not greater than the second preset threshold, decrease. ;when When it exceeds the second preset threshold, increase The parameter adjustment is performed after the comparison results have been continuously maintained for a preset number of processing blocks.

[0017] Furthermore, this is deployed in the baseband digital domain processing flow at the receiving end, and is executed after completing conventional RF isolation, analog suppression, down-conversion, and analog-to-digital conversion; the eliminated receiving sub-band frequency domain output... It is sent into the channel estimation, equalization, demodulation or decoding process.

[0018] The advantages of this invention compared to the prior art are: 1. This invention estimates a large number of frequency points within the receiving subband by compressing them into a small number of shared complex coefficients. This results in a significant reduction in the number of parameters to be calculated compared to higher-order nonlinear models or data-driven models, and the calculation path is more suitable for real-time processing in the baseband digital domain at the receiving end, thus reducing processing complexity.

[0019] 2. This invention achieves a processing method that does not rely on training, iterative search, or partial extrapolation by adopting a single path of frequency domain transformation, fixed continuous grouping, group coefficient calculation, adjacent smoothing, and group elimination. The processing flow is clear and easy to implement directly into the digital processing flow of the receiving end.

[0020] 3. By using smaller groups near the boundary region and larger groups far from the boundary region, this invention achieves a better balance between rapidly changing regions near the boundary and gently changing regions far from the boundary without significantly increasing the number of parameters, and is more targeted at residual self-interference near the sub-band boundary.

[0021] 4. This invention achieves a more stable coefficient result by having each shared complex coefficient supported by multiple frequency points within the corresponding group and performing a fixed smoothing process between groups. This is more effective than completely independent and unrelated local estimation in suppressing coefficient anomalies caused by local frequency fluctuations.

[0022] 5. This invention introduces time-domain shared coefficient adaptive filtering between each processing block and utilizes the slowly varying time correlation of the self-interference channel between adjacent processing blocks to perform cross-block time-domain recursion on the shared complex coefficients. This achieves the effect of further suppressing the random noise of single-block estimation in the time dimension. Compared with block-by-block independent processing, the stability of coefficients and the elimination accuracy are further improved.

[0023] 6. By introducing elimination effect monitoring and adaptive configuration, this invention dynamically adjusts the time-domain filtering parameters according to the actual elimination gain, achieving the effect of adaptive adjustment of system parameters with channel and interference environment, avoiding the mismatch problem of fixed parameters in different scenarios, and making the system more robust.

[0024] 7. This invention belongs to the internal processing method of the baseband digital domain of the receiving end. It can be executed after conventional radio frequency isolation, analog suppression and analog-to-digital conversion. It achieves the effect of not requiring additional control signaling or changes to the basic transceiver structure of subband full-duplex, and does not rely on new protocols or hardware modifications. Attached Figure Description

[0025] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application.

[0026] In the attached diagram: Figure 1 This is a flowchart of a method for eliminating self-interference of shared coefficients in a sub-band full-duplex digital domain according to the present invention.

[0027] Figure 2 This is a flowchart of the time-domain shared coefficient adaptive filtering in the sub-band full-duplex digital domain shared coefficient self-interference cancellation method of the present invention.

[0028] Figure 3This is a flowchart of the self-interference cancellation method for sub-band full-duplex digital domain shared coefficients according to the present invention, which includes monitoring and adaptive configuration of cancellation effect. Detailed Implementation

[0029] The following detailed description of the embodiments is intended to exemplify the principles of this application, but should not be used to limit the scope of this application. That is, the self-interference cancellation method of sub-band full-duplex digital domain shared coefficients in this application is not limited to the described embodiments.

[0030] The present invention will be further described below with reference to embodiments.

[0031] Example 1: Self-interference cancellation of digital domain sharing coefficients applied to a 100 MHz sub-band full-duplex base station This embodiment targets a sub-band full-duplex test base station. The base station operates with a 100MHz carrier bandwidth and uses adjacent sub-bands for transmission and reception at the same time. The transmission sub-band occupies 75MHz and the reception sub-band occupies 25MHz. The base station already has conventional transmit / receive isolation and analog suppression capabilities at the radio frequency end. However, residual self-interference caused by nonlinearity and limited isolation at the transmitter still exists in the baseband digital domain at the receiver end. This method is deployed in the baseband digital domain processing flow at the receiver end. After completing conventional radio frequency isolation, analog suppression, down-conversion, and analog-to-digital conversion, the residual self-interference in the receiver sub-band is reconstructed and subtracted in the digital domain.

[0032] like Figure 1 As shown, in this embodiment, each processing block corresponds to one OFDM symbol, using a 4096-point Discrete Fourier Transform. Since the receiving subband occupies one-quarter of the total bandwidth, the receiving subband corresponds to 1024 consecutive frequency points. The receiving frequency point closest to the boundary between the transmitting and receiving subbands is denoted as... Then the set of receiving sub-band frequency points is denoted as .

[0033] Step S1: Obtain the received baseband discrete signal of the current OFDM symbol. and the known baseband discrete signal transmitted at this end ,in =0,1,…,4095, perform 4096-point Discrete Fourier Transform on both, and obtain the received frequency domain observation value according to the following formula. and reference frequency domain value : Step S2: Perform fixed continuous grouping of the 1024 receiving frequency points in order of distance from the sub-band boundary from near to far, starting from the frequency point. Start by continuously grouping along the direction of the receiving sub-band to obtain continuous and non-overlapping groups.

[0034] The reason for adopting the above-mentioned fixed grouping rule in this invention is that the residual self-interference changes faster near the boundary region of the received subband, requiring finer grouping, while the changes are relatively gradual far from the boundary region, allowing for larger grouping to reduce the number of parameters. The length of the first group is taken as... =16 and using the doubling grouping rule, we get 6 groups. Including the 1st to 16th receiving frequencies, Including receiving frequencies from the 17th to the 48th, There are a total of 6 groups, with lengths of 16, 32, 64, 128, 256 and 528 respectively. After this grouping, this embodiment only needs to solve 6 shared complex coefficients in each OFDM symbol.

[0035] Step S3: Calculate the initial shared complex coefficients for each group. For any group... Construct a least-squares estimate using the received observations and reference values ​​for all frequencies within the group, and set the regularization constant as follows: =10 To prevent the denominator from being too small, the initial shared complex coefficients for this group are calculated using the following formula: in, express The conjugate of the initial shared complex coefficients Indicates the first The residual self-interference equivalent digital domain coefficients shared by multiple frequency points within a group are used to obtain the initial shared complex coefficients for that group. .

[0036] Step S4: To avoid non-physical abrupt changes in coefficients between adjacent groups, a fixed-adjacency smoothing process is performed on the 6 initial shared complex coefficients obtained in Step S3. Let the smoothing factor be... =0.8, execute according to the following formula: After smoothing, the final shared complex coefficients of each group are obtained. Smoothed This serves as the final shared complex coefficient corresponding to this OFDM symbol.

[0037] Step S5: As Figure 2 As shown, time-domain shared coefficient adaptive filtering is performed on the final shared complex coefficients of this OFDM symbol, and the time-domain sensitivity constant is taken. =0.05. Let the time-domain shared complex coefficients of the previous OFDM symbol, after time-domain filtering, be... For the first OFDM symbol, there is no historical information, let = Proceed directly to step S6 without time-domain recursion. For subsequent OFDM symbols, calculate the inter-block normalization coefficient difference measure according to the following formula: in, As the regularization constant, the adaptive time-domain recursive factor is then calculated using the following formula: Then, time-domain recursive filtering is performed according to the following formula: Obtain the time-domain shared complex coefficients of each group of this OFDM symbol. And store it for recursive use in the time domain of the next OFDM symbol. Meanwhile, when Greater than the preset threshold At that time, the preset threshold For example, let 0.5, = Skip this time-domain recursion and directly use the final shared complex coefficients of the current OFDM symbol as the time-domain shared complex coefficients.

[0038] Step S6: Reconstruct residual self-interference by group and perform digital domain cancellation for any frequency point. The residual self-interference estimate for this frequency point is constructed using the time-domain shared complex coefficients of the group, and calculated according to the following formula: in, Frequency point number For frequency point The group number it belongs to, For the current processing block Time-domain shared complex coefficients of each group, For reference frequency domain values, To receive frequency domain observations, For frequency point The residual self-interference estimate, The digital domain output is then eliminated; this is then subtracted from the received observations to obtain the eliminated digital domain output. After performing this process on all 1024 received frequency points, the eliminated frequency domain output of the received subband is obtained. The output is then fed into the base station’s regular receive link for subsequent channel estimation, equalization and demodulation.

[0039] Step S7: As Figure 3As shown, the cancellation effect is monitored and parameters are configured for the frequency domain output after cancellation. The cancellation gain of this OFDM symbol is calculated according to the following formula: in, The received frequency domain observations before elimination, The received frequency domain value after elimination. This is the set of receiving sub-band frequencies. Set the elimination effect threshold. =-15dB =-5dB: When Less than At the same time, maintain the time-domain sensitivity constant =0.05 remains unchanged; when ≤ ≤ At that time, Adjusted to ×0.9; when Greater than At that time, Adjusted to ×1.5, and set simultaneously The upper limit constraint is that the parameter adjustment is triggered after the elimination effect condition has been maintained for a preset number of OFDM symbols, such as 5.

[0040] In this embodiment, the present invention does not introduce a high-order nonlinear full model, neural network training process, or partial band reconstruction step. Instead, it performs boundary-guided parameter compression modeling of the received subband in a fixed continuous grouping manner within the baseband digital domain of the receiving end, so that the frequency points in the same group share a complex elimination coefficient. At the same time, by smoothing adjacent coefficient abrupt changes, a digital domain self-interference elimination process with clear computational load, clear steps, and easy real-time implementation is formed. For the 100MHz subband full-duplex base station, this method only needs to process 6 shared complex coefficients in each OFDM symbol, without relying on training, iterative search, or partial band extrapolation, which is convenient to implement in the baseband digital processing module of the base station receiver.

[0041] This embodiment further introduces time-domain shared coefficient adaptive filtering in step S5 between each OFDM symbol. It utilizes the slowly varying time correlation of the self-interference channel between adjacent OFDM symbols to perform cross-symbol time-domain recursion on the six shared complex coefficients, further suppressing single-symbol estimation noise in the time dimension, making the coefficients more stable and the cancellation accuracy higher. Simultaneously, through the cancellation effect monitoring in step S7, the time-domain sensitivity constant is dynamically adjusted based on the actual cancellation gain, allowing the system parameters to adapt to the channel and interference environment.

[0042] It should also be noted that the time-domain shared coefficient adaptive filtering only requires storing 6 additional complex numbers in each OFDM symbol. These 6 complex numbers are the time-domain shared complex coefficients of the previous OFDM symbol. It also requires performing 6 complex multiplications and additions and 1 difference measurement calculation. The elimination effect monitoring only requires calculating the power ratio once. The storage and computational overhead brought by both methods is negligible compared to the original processing, and does not change the characteristics of low complexity and easy real-time implementation of this method.

[0043] It should be noted that the combination of the technical features in this case is not limited to the combination methods described in the claims of this case or the combination methods described in the specific embodiments. All technical features described in this case can be freely combined or combined in any way, unless they contradict each other.

[0044] It should also be noted that the embodiments listed above are merely specific embodiments of the present invention. Obviously, the present invention is not limited to the above embodiments, and similar changes or modifications made thereto are those that can be directly derived or easily conceived by those skilled in the art from the content disclosed in the present invention, and should all fall within the protection scope of the present invention.

Claims

1. A method for eliminating self-interference of shared coefficients in a sub-band full-duplex digital domain, characterized in that, Includes the following steps: Step S1: Obtain the received baseband signal and the known transmitted baseband signal of the current processing block, and transform both to the digital frequency domain to obtain the received frequency domain observation value and the digital domain reference value obtained from the known transmitted baseband signal of the local end; Step S2: Group the received subband frequency points into fixed, continuous groups according to their distance from the subband boundary, from closest to furthest. Step S3: For each group, calculate the initial shared complex coefficients corresponding to that group; Step S4: Perform fixed-adjacency smoothing on the initial shared complex coefficients to obtain the final shared complex coefficients corresponding to each group; Step S5: Perform time-domain shared coefficient adaptive filtering on the final shared complex coefficients to obtain the time-domain shared complex coefficients corresponding to each group; Step S6: For any frequency point, construct the residual self-interference estimate of that frequency point using the time-domain shared complex coefficients of the group, and subtract it from the received frequency domain observation to obtain the eliminated received sub-band frequency domain output. Step S7: Monitor the cancellation effect of the frequency domain output of the canceled receiving sub-band, and configure the time domain filtering parameters of the subsequent processing block according to the monitoring results.

2. The method for eliminating self-interference of sub-band full-duplex digital domain shared coefficients according to claim 1, characterized in that, In step S1, the received baseband signal and the known transmitted baseband signal at this end are respectively processed. The point-based discrete Fourier transform is calculated using the following formula: ； in, The received frequency domain observation value, The reference value for the digital field is... For the received baseband discrete signal of the current processing block, To transmit baseband discrete signals at this end. =0,1,…, -1, This represents the number of sampling points in the current processing block.

3. The method for eliminating self-interference of sub-band full-duplex digital domain shared coefficients according to claim 1, characterized in that, In step S1, the set of frequency points corresponding to the received sub-band is as follows: ;in, The starting frequency point in the receive sub-band that is closest to the boundary between the transmit and receive sub-bands. This represents the total number of frequency points included in the receiving subband.

4. The method for eliminating self-interference of sub-band full-duplex digital domain shared coefficients according to claim 3, characterized in that, In step S2, starting from the starting frequency point, continuous grouping is performed along the receiving sub-band direction to obtain continuous and non-overlapping groups. Each group is determined according to the following formula: ； Obtain continuous and non-overlapping groupings: ；…… in, From before removal The remaining frequency points after the group constitute the group. The length of the first group, For the first Group length, Group number This represents the total number of groups.

5. The method for eliminating self-interference of sub-band full-duplex digital domain shared coefficients according to claim 1, characterized in that, In step S3, for any group, a least-squares estimate is constructed using the received frequency domain observations and digital domain reference values ​​of all frequency points within the group, and the initial shared complex coefficients corresponding to the group are calculated according to the following formula: ;in, To receive frequency domain observations, For reference values ​​in the numeric field, express conjugate, Here is the regularization constant. For the first Groups, For the first The initial shared complex coefficients of the nth group, wherein the initial shared complex coefficients represent the initial shared complex coefficients of the nth group. The residual self-interference equivalent digital domain coefficients shared by multiple frequency points within the group.

6. The method for eliminating self-interference of sub-band full-duplex digital domain shared coefficients according to claim 1, characterized in that, In step S4, the initial shared complex coefficients undergo a fixed-adjacent smoothing process, with the smoothing factor set to 1. Perform the following steps: ； in, As a smoothing factor, For the first The initial shared complex coefficients of each group are smoothed to obtain the final shared complex coefficients of each group. .

7. The method for eliminating self-interference of sub-band full-duplex digital domain shared coefficients according to claim 3, characterized in that, In step S6, for any frequency point, using the time-domain shared complex coefficients of the group to which that frequency point belongs, the residual self-interference estimate for that frequency point is constructed according to the following formula, and digital domain cancellation is performed: ;in, Frequency point number For frequency point The group number it belongs to, For the current processing block Time-domain shared complex coefficients of each group, The reference value for the digital field is... The received frequency domain observation value, For frequency point The residual self-interference estimate, The received frequency domain value after elimination; after all frequency points have undergone the above processing, the eliminated frequency domain output of the received sub-band is obtained. .

8. The method for eliminating self-interference of sub-band full-duplex digital domain shared coefficients according to claim 1, characterized in that, Deployed in the baseband digital domain processing flow at the receiving end, it is executed after completing conventional RF isolation, analog suppression, downconversion, and analog-to-digital conversion; the eliminated receiving subband frequency domain output is sent to the channel estimation, equalization, demodulation, or decoding processing flow.

9. The method for eliminating self-interference of sub-band full-duplex digital domain shared coefficients according to claim 1, characterized in that, In step S5, the time-domain recursive filtering is performed according to the following formula: ;in, This is the sequence number of the current processing block. The final shared complex coefficients obtained by the current processing block in step S4. These are the time-domain shared complex coefficients of the previous processing block after time-domain filtering. For time-domain recursion factors, For the current processing block The time-domain shared complex coefficients of the group; the time-domain recursive factor Measured by the difference in normalized coefficients between the current processing block and the previous processing block. Determine, calculate according to the following formula: ； in, Let be the time-domain sensitivity constant. The normalization constant is the regularization constant; the difference in the normalization coefficient is measured. The value of is related to the time-domain recursive factor. The values ​​of are positively correlated.

10. The method for eliminating self-interference of sub-band full-duplex digital domain shared coefficients according to claim 1, characterized in that, In step S7, the elimination gain is calculated according to the following formula: ;in, The received frequency domain observations before elimination, The received frequency domain value after elimination. For receiving sub-band frequency points, To eliminate gain; according to the said gain elimination The comparison result with the preset threshold configures the time-domain sensitivity constant of subsequent processing blocks. :when When it is less than the first preset threshold, maintain Unchanged; when When the value is not less than the first preset threshold and not greater than the second preset threshold, decrease. ;when When it exceeds the second preset threshold, increase The parameter adjustment is performed after the comparison results have been continuously maintained for a preset number of processing blocks.

Images