A Self-Interference Signal Cancellation Method Based on the CMP Power Amplifier Model

Abstract

The present invention provides a self-interference signal cancellation method based on a CMP power amplifier model, which relates to the technical field of analog domain self-interference cancellation. The method first trains the CMP model, then obtains the reference signal of the first adaptive filter by passing the local modulation signal through the CMP model, and then passes the reference signal and the hydrophone received signal through the first adaptive filter together to complete the digital domain self-interference reconstruction, obtaining the self-interference reconstruction signal. The self-interference reconstruction signal and the received signal are cancelled in the analog domain to obtain the signal after analog domain self-interference cancellation. The obtained signal and the reference signal are passed through the second adaptive filter together to complete the residual digital domain self-interference cancellation, and further demodulate the error of the second adaptive filter to obtain the information of the remote desired signal. The method can significantly improve the self-interference cancellation performance of the in-band full-duplex underwater acoustic communication system.

Description

Technical Field

[0001] The present invention relates to the technical field of analog domain self-interference cancellation, and particularly to a self-interference signal cancellation method based on a CMP power amplifier model. Background Art

[0002] In the process of analog domain self-interference cancellation, it is necessary to focus on how to cancel the non-linear components to improve the performance of analog domain self-interference cancellation. The digital-assisted analog domain self-interference cancellation technology is the mainstream method of current analog domain self-interference cancellation technology. This technology models or extracts the non-linear part in the power amplifier through an auxiliary link, so that the reference signal contains non-linear components. Therefore, this analog domain self-interference cancellation technology can effectively eliminate some non-linear interferences.

[0003] In response to the above problems, Wu Wei from Nanjing University of Posts and Telecommunications published "A Method for Canceling Self-Interference Signals in Wireless Full-Duplex Multipath Based on Piecewise Inverse Transformation" in the Signal Processing Journal, which records an analog-domain self-interference cancellation method applicable to full-duplex devices in a wireless multipath environment. This method is based on the Rayleigh distribution model, directly generates digital-domain cancellation parameters using the method for canceling self-interference signals in wireless full-duplex multipath based on piecewise inverse transformation, and converts the cancellation signal to the analog domain through an additional transmission link for self-interference cancellation. Compared with other analog-domain cancellation schemes assisted by the digital domain, this scheme does not require the establishment of a complex channel model and estimation of channel parameters in the digital domain, nor does it require coupling the transmitter signal, thus simplifying the structure. Zhou Jiaqiong from Harbin Engineering University proposed a digital-assisted analog self-interference cancellation scheme in her master's thesis "Research on Digital and Analog Self-Interference Cancellation Technologies for Full-Duplex Underwater Acoustic Communication", and studied the nonlinear effects of power amplifiers involved in analog self-interference cancellation, using the memory polynomial model to describe the nonlinear effects of power amplifiers. It includes two analog self-interference cancellation schemes in total. The first is to reconstruct the self-interference signal using the power amplifier output and the estimated underwater acoustic channel to achieve analog cancellation; the second is to modulate the signal through the memory polynomial model and the estimated underwater acoustic channel to reconstruct the self-interference signal to achieve analog cancellation. Zhao Yujie from Northeast Electric Power University proposed a time-varying least mean square adaptive filtering algorithm with a gradually decaying convergence parameter in her master's thesis "Research on Self-Interference Cancellation Technology for Simultaneous and Co-Frequency Full-Duplex Systems". By using an improved arctangent function and relying on the relationship between the time parameter and the correlation value of the error signal to coordinate and control the step size, it can still maintain ideal convergence, channel tracking ability, interference cancellation ratio and other characteristics at a lower computational complexity, effectively suppressing the self-interference signals in the radio frequency stage of the simultaneous and co-frequency full-duplex system. The research team of Professor Liu Ying from the University of Electronic Science and Technology published a patent "A Digital-Assisted Analog Domain Interference Cancellation Method and Device". This method includes four steps: signal transmission, signal feedback, signal reconstruction and signal reception. In the signal feedback step, a nonlinear extraction unit is used to estimate the nonlinear parameter vector, and in the signal reconstruction step, a nonlinear reconstruction unit is used to calculate the reconstructed digital transmission signal; this method can effectively suppress the nonlinear interference introduced by the transmit radio frequency channel, can conveniently perform nonlinear estimation and parameter adjustment in the digital domain, can conveniently estimate the multipath channel between the transmit radio frequency channel and the receive radio frequency channel in the digital domain, and adjust the parameters, which is more conducive to being extended to other application scenarios and has stronger robustness. The wireless communication technology laboratory of Huawei proposed the adoption of multi-stage self-interference cancellation technology in the IEEE Global Communication Conference and achieved single-antenna CCFD wireless communication on self-made equipment in the laboratory. In this system, in addition to line-of-sight (LOS) communication, the multipath effect of signal propagation is also considered, and the corresponding cancellation of the multipath interference component is performed in the analog domain.

[0004] Although the existing technologies can usually achieve good results, the separate digital domain technology alone cannot solve the problem of the failure to obtain the desired signal at the far end caused by the saturation of the collected signal. In an underwater acoustic channel, traditional methods in the radio field (such as antenna polarization, antenna orthogonality, etc.) are no longer applicable. The self-interference cancellation performance in the spatial domain is limited and is affected by the current environment (such as the number of array elements, sea surface fluctuations, sea breeze, etc.). Most of the analog domain self-interference cancellation technologies are circuit simulations, software simulations, etc., lacking the joint debugging design of a physical-based solution and a hardware system. How to jointly perform self-interference cancellation with a lower complexity in the analog domain and the digital domain, design the connection scheme of related hardware devices, and conduct field experiments that can be carried out in real time for verification is an urgent problem to be solved currently. Summary of the Invention

[0005] The purpose of the present invention is to provide a self-interference signal cancellation method based on a CMP power amplifier model. Aiming at the problems in the analog domain self-interference cancellation process, the concept of digital-assisted analog self-interference cancellation is introduced, so that the analog domain self-interference cancellation can eliminate the non-linear components in the self-interference signal and further improve the self-interference cancellation performance.

[0006] To achieve the above purpose, the present invention provides a self-interference signal cancellation method based on a CMP power amplifier model. The method includes the following steps:

[0007] S101. Generate a known broadband initial signal , and obtain a signal after delaying and attenuating the power amplifier output signal , and combine with the broadband initial signal , and through the CMP model training process, calculate the CMP non-linear distortion model parameters of the power amplifier, n being discrete sampling points;

[0008] S102. The local signal source is modulated and the transmitted signal is sent out by the transducer x n , x n passes through the self-interference channel , and is received by the hydrophone, denoted as the hydrophone received signal y n , and the hydrophone received signal y n is collected locally after passing through the low-noise amplifier;

[0009] S103. After the transmitted signal is processed by the CMP model, the reference signal of the first adaptive filter is obtained ;

[0010] S104. The reference signal and the hydrophone received signal​​​​y n Together, they pass through the first adaptive filter to complete the digital domain self-interference reconstruction, obtaining the estimated self-interference channel , and the self-interference reconstruction signal , ;

[0011] S105. Cancel the self-interference reconstruction signal and the received signal in the analog domain to obtain the signal after analog domain self-interference cancellation ;

[0012] S106. It is collected by the system ADC to the local, and passes through both ends of the second adaptive filter together with the reference signal to complete the cancellation of the residual digital domain self-interference, obtaining the estimated residual self-interference channel , and the residual reconstruction signal . The error e n of the second adaptive filter is the remote communication signal to be demodulated. At this time, the relationship between the error and the residual reconstruction signal is:

[0013] ;

[0014] S107. Demodulate the error of the second adaptive filter to obtain the information of the remote desired signal.

[0015] Furthermore, the expression of the hydrophone received signal y n is:

[0016]

[0017] Among them, is the local self-interference signal received by the hydrophone at the local receiving end, , is the remote communication signal received by the hydrophone at the local receiving end, is the ambient noise received by the local receiving end.

[0018] Furthermore, the CMP model training process specifically includes the steps:

[0019] S201. Initialize, determine and , set the model parameters , and , represents the order of the signal and polynomial memory interaction term, represents the number of delay samples, ​​​Represents the order of the interaction term between the signal and the lag envelope, Represents the memory depth of the signal, Represents the number of sampling points of the lag envelope, Represents the order of the interaction term between the signal and the lead envelope, Represents the memory depth of the signal, Represents the number of sampling points of the lead envelope;

[0020] S202. Determine the CMP model expression. The CMP model expression is as follows:

[0021]

[0022]

[0023]

[0024] Estimate the model parameters by fitting the model 、 and , k Represents the non - linear order, l Represents the delay, m Represents the delay samples of the positive and negative cross - period time offset;

[0025] S203. Define the loss function;

[0026] S204. Construct the design matrix and define the design matrix X and the parameter vector β Then there is:

[0027]

[0028] Represents the true output. Each column of the design matrix X corresponds to a CMP model term, and each row corresponds to a time sample n. Expand all the coefficients 、 and to obtain the parameter vector β The dimension of the parameter vector corresponds to the number of columns of the design matrix;

[0029] S205. Estimate the parameter vector using the least - squares method. The expression is:

[0030]

[0031] Obtain After that, decompose it in order into 、 and as the final result of the CMP model;

[0032] S206. Substitute 、 and into the CMP model, calculate the reference signal , and adjust , and according to the loss function to obtain the finally output CMP model.

[0033] Further, step S105 specifically includes the following steps:

[0034] S301. Output the self-interference reconstruction signal through the NI-DAQ device. When outputting, adjust the amplitude so that the amplitude of the self-interference reconstruction signal is within the same range as the amplitude of the received signal, and perform analog output to cancel the self-interference reconstruction signal and the received signal in the analog domain;

[0035] S302. By adjusting the time delay, make the amplitude of the output signal waveform of the subtractor the lowest to obtain the optimal delay, and obtain the signal after self-interference cancellation in the analog domain .

[0036] Compared with the prior art, the beneficial effects of the present invention are:

[0037] A self-interference signal cancellation method based on a CMP power amplifier model provided by the present invention obtains a reconstructed power amplifier output signal after passing the local modulation signal through the trained CMP model, and uses it as a reference signal, and sequentially performs digital domain self-interference reconstruction, analog domain self-interference cancellation, and residual digital domain self-interference cancellation to complete digital-assisted analog domain self-interference cancellation. Using the method, the self-interference cancellation performance of the in-band full-duplex underwater acoustic communication system can be significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only the preferred embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0039] Figure 1 is the overall flowchart of a self-interference signal cancellation method based on a CMP power amplifier model provided by an embodiment of the present invention.

[0040] Figure 2 is the schematic diagram of the sea trial experimental scenario provided by an embodiment of the present invention.

[0041] Figure 3 It is the constellation diagram of the power amplifier output signal reconstructed through the CMP model provided by the embodiment of the present invention.

[0042] Figure 4 It is a schematic diagram showing the relationship between the reconstructed signal NMSE of the class T self-made power amplifier and the non-linear order and memory depth provided by the embodiment of the present invention.

[0043] Figure 5 It is a schematic diagram showing the relationship between the reconstructed signal NMSE of the power amplifier of model BK2713 and the non-linear order and memory depth provided by the embodiment of the present invention.

[0044] Figure 6 It is a schematic diagram for comparing the frequency domain of the power amplifier output signal and the reconstructed signal provided by the embodiment of the present invention.

[0045] Figure 7 It is a schematic diagram of the NMSE curve of the analog domain self-interference cancellation of different reference signals under the condition of the existence of a remote communication signal provided by the embodiment of the present invention. Detailed implementation manners

[0046] The principles and features of the present invention will be described below in conjunction with the accompanying drawings. The listed embodiments are only used to explain the present invention and are not intended to limit the scope of the present invention.

[0047] Refer to Figure 1 , this embodiment provides a self-interference signal cancellation method based on a CMP power amplifier model, and the method includes the following steps:

[0048] S101. Generate a known broadband initial signal , obtain a signal after delaying and attenuating the power amplifier output signal , and combine with the broadband initial signal , through the CMP model training process, calculate the CMP non-linear distortion model parameters of the power amplifier, n being discrete sampling points.

[0049] Exemplarily, the broadband initial signal can be an LFM signal, or an HFM signal, or other broadband signals.

[0050] S102. Modulate the local signal source, and transmit the transmitted signal through the transducer x n , x n passes through the self-interference channel , and is received by the hydrophone, denoted as the hydrophone received signal y n , the hydrophone received signal y n ​​​​After passing through the low-noise amplifier, it is collected locally.

[0051] In this embodiment, the received signal y n can be expressed as:

[0052]

[0053] where is the local self-interference signal received by the local receiving hydrophone, , is the remote communication signal received by the local receiving hydrophone, is the ambient noise received by the local receiving end. When the remote communication signal (also known as the remote desired signal) is not transmitted, the received signal does not contain the term.

[0054] S103. After the transmitted signal is processed by the CMP model, the reference signal of the first adaptive filter is obtained.

[0055] S104. The reference signal and the hydrophone received signal y n are passed through the first adaptive filter together to complete the digital-domain self-interference reconstruction, obtaining the estimated self-interference channel , and the self-interference reconstruction signal , .

[0056] S105. The self-interference reconstruction signal and the received signal are cancelled in the analog domain to obtain the signal after analog-domain self-interference cancellation.

[0057] S106. is collected locally by the system ADC and passed through both ends of the second adaptive filter together with the reference signal to complete the residual digital-domain self-interference cancellation, obtaining the estimated residual self-interference channel , and the residual reconstruction signal , , and the error e of the second adaptive filter n is the remote communication signal to be demodulated. At this time, the relationship between the error e n and the residual reconstruction signal is:

[0058]

[0059] S107. For the error ​​​Demodulation is performed to obtain the remote desired signal information.

[0060] As a possible implementation, the CMP model training process specifically includes the steps of:

[0061] S201. Initialization, determine and , set the model parameters , and . represents the order of the interaction term between the signal and the polynomial memory, represents the number of delayed samples, represents the order of the interaction term between the signal and the lag envelope, represents the memory depth of the signal, represents the number of sampling points of the lag envelope, represents the order of the interaction term between the signal and the leading envelope, represents the memory depth of the signal, represents the number of sampling points of the leading envelope.

[0062] S202. Determine the CMP model expression, and the CMP model expression is as follows:

[0063]

[0064]

[0065]

[0066] Estimate the model parameters 、 and , k represents the non-linear order, l represents the delay, m represents the delayed samples of the positive and negative cross-period time offset. represents the current sample, represents the previous delayed sample, represents the backward delayed sample.

[0067] S203. Define the loss function. To estimate 、 and , it is necessary to minimize the error between the predicted output and the true output .

[0068] In this implementation, the least squares error As the loss function:

[0069]

[0070] N is the number of data samples.

[0071] S204. Construct a design matrix and define the design matrix X and the parameter vector β , then there is:

[0072]

[0073] represents the true output. Each column of the design matrix X corresponds to a CMP model term. For example, each row corresponds to a time sample n. Expand all the coefficients 、 and to obtain the parameter vector β . The dimension of the parameter vector corresponds to the number of columns of the design matrix.

[0074] S205. Estimate the parameter vector by the least squares method. The expression is:

[0075]

[0076] After obtaining , decompose it into 、 and in sequence as the final result of the CMP model;

[0077] S206. Substitute 、 and into the CMP model to calculate the reference signal . Adjust , and according to the loss function to obtain the final output CMP model.

[0078] As another possible implementation manner, step S105 specifically includes the following steps:

[0079] S301. Output the self-interference reconstruction signal through the NI-DAQ device. When outputting, adjust the amplitude so that the amplitude of the self-interference reconstruction signal is within the same range as the amplitude of the received signal, and perform analog output to cancel the self-interference reconstruction signal and the received signal in the analog domain.

[0080] S302. Adjust the time delay to minimize the amplitude of the output signal waveform of the subtractor, so as to obtain the optimal delay and obtain the signal after analog domain self-interference cancellation. .

[0081] Using the analog self-interference cancellation method for power amplifier output reconstruction based on the CMP model provided in this embodiment can significantly improve the self-interference cancellation performance of the in-band full-duplex underwater acoustic communication system. The present invention makes a detailed analysis of the influence of reconstructing the power amplifier output signal for the CMP model on the self-interference cancellation performance, and completes the sea trial experiment based on the Figure 2 scheme shown.

[0082] Figure 3 The constellation diagram of the power amplifier output signal reconstructed by the CMP model is shown. It can be seen from Figure 3 that after the reconstruction by the CMP model, the constellation diagrams of the output signals are basically fitted, and the power amplifier output signal after being fitted by the CMP model has nonlinear characteristics.

[0083] Figure 4 , Figure 5 It shows that for the self-made class T power amplifier, in the power amplifier output reconstruction based on the CMP model, when the minimum NMSE is reached, the nonlinear order of the model is 1 and the memory depth is 11, where the minimum NMSE is -21.89 dB. For the BK2713 power amplifier, in the power amplifier output reconstruction based on the CMP model, when the minimum NMSE is reached, the nonlinear order of the model is 1 and the memory depth is 11, and the minimum NMSE is -24.84 dB. It can be seen from this that for the power amplifier output reconstruction based on the CMP model, a very high nonlinear order is not required, and the NMSE changes relatively stably with the memory depth. And the self-made class T power amplifier has a relatively high nonlinear influence on the self-interference signal.

[0084] Figure 6 It shows the comparison of the output signals of the BK2713 model / self-made class T power amplifier reconstructed by the CMP model in the frequency spectrum. It can be seen from this figure that the power amplifier output after being fitted by the CMP model also has a good fitting effect in the system frequency band.

[0085] From Figure 7 it can be seen that the analog interference cancellation scheme based on the CMP power amplifier model output reconstruction has a better cancellation effect than the conventional analog interference cancellation method, and the self-interference cancellation performance is improved by 3.5 dB, and the remote communication signal is well retained.

[0086] In summary, to address the problems in the analog-domain self-interference cancellation process, this embodiment introduces the concept of digital-assisted analog self-interference cancellation, and proposes a digital-assisted analog self-interference cancellation method for power amplifier output reconstruction based on the CMP model, enabling the analog-domain self-interference cancellation to eliminate the non-linear components in the self-interference signal and further improving the self-interference cancellation performance.

[0087] The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A self-interference signal cancellation method based on a CMP power amplifier model, characterized in that: The method comprises: S101, generate a known broadband initial signal , the amplifier output signal is delayed and attenuated to obtain the signal ,Will Joint Broadband Initial Signal , after the CMP model training process, the CMP nonlinear distortion model parameters of the power amplifier are calculated, n are discrete sampling points; S102: The local signal source is modulated and the transducer sends out a transmission signal x [ n ], x [ n ]Through the self-interference channel , received by the hydrophone, recorded as the hydrophone received signal y [ n ], the hydrophone receives the signal y [ n ] After passing through a low noise amplifier, the data is collected locally; S103: After the transmitted signal is processed by the CMP model, a reference signal of the first adaptive filter is obtained. ; S104: reference signal Receiving signals with hydrophone y [ n ] pass through the first adaptive filter together to complete the digital domain self-interference reconstruction and obtain the estimated self-interference channel , and the self-interference reconstruction signal , ; S105, canceling the self-interference reconstruction signal and the received signal in the analog domain to obtain a signal after self-interference cancellation in the analog domain ; S106, It is collected locally by the system ADC and passes through the two ends of the second adaptive filter together with the reference signal to complete the residual digital domain self-interference cancellation and obtain the estimated residual self-interference channel , and the residual reconstructed signal , the error of the second adaptive filter e [ n ] is the remote communication signal to be demodulated. At this time, the relationship between the error and the residual reconstructed signal is: ； S107, error of the second adaptive filter Demodulate to obtain the desired signal information at the far end; The CMP model training process includes the following steps: S201, initialization, confirmation and , set the model parameter , and , represents the order of the interaction between the signal and the polynomial memory, represents the number of delayed samples, represents the order of the interaction term between the signal and the lagged envelope, Indicates the memory depth of the signal, represents the number of sampling points of the hysteresis envelope, represents the order of the interaction term between the signal and the leading envelope, Indicates the memory depth of the signal, Indicates the number of sampling points of the leading envelope; S202, determine a CMP model expression, the CMP model expression is as follows: By fitting the model, we can estimate the model parameters 、 and , k represents the nonlinear order, l Indicates delay, m Delayed samples representing positive and negative intertemporal time offsets; S203, define a loss function; S204. Construct design matrix and define design matrix X and parameter vector β , then: represents the real output, the design matrix X Each column corresponds to a CMP model term, and each row corresponds to a time sample n. 、 and Expand to obtain the parameter vector β , the dimension of the parameter vector corresponds to the number of columns of the design matrix; S205. Estimate the parameter vector using the least squares method. The expression is: get Then, decompose it into 、 and , as the final result of the CMP model; S206, will 、 and Substitute into the CMP model and calculate the reference signal , according to the loss function , and Adjustments are made to obtain the final output CMP model.

2. The method for canceling self-interference signals based on a CMP power amplifier model according to claim 1, characterized in that: Hydrophone receiving signal y [ n The expression of ] is: in, is the local self-interference signal received by the local receiving hydrophone, , is the remote communication signal received by the local receiving hydrophone, is the environmental noise received by the local receiving end.

3. The method for canceling self-interference signals based on a CMP power amplifier model according to claim 1, characterized in that: Step S105 specifically includes the following steps: S301, outputting the self-interference reconstruction signal through the NI-DAQ device, adjusting the amplitude during output so that the amplitude of the self-interference reconstruction signal and the amplitude of the received signal are in the same range, and performing analog output so that the self-interference reconstruction signal and the received signal are offset in the analog domain; S302, by adjusting the delay, the amplitude of the output signal waveform of the subtractor is minimized to obtain the corresponding delay, and obtain the signal after the analog domain self-interference is cancelled. .

Images