CN105556860B - Systems and methods for nonlinear digital self-interference cancellation - Google Patents

Abstract

A system and method for nonlinear digital self-interference cancellation includes a preprocessor that generates a first preprocessed digital transmit signal from a digital transmit signal of a full-duplex radio, a nonlinear converter that converts the first preprocessed digital transmit signal into a nonlinear self-interference signal according to a conversion configuration, a conversion adapter that sets the conversion configuration of the nonlinear converter, and a postprocessor that combines the nonlinear self-interference signal with a digital receive signal of the full-duplex radio.

Description

Systems and methods for nonlinear digital self-interference cancellation

Cross References to Related Applications

This application claims the benefit of US Provisional Application Serial No. 61/864,453, filed August 09, 2013, which is hereby incorporated by reference in its entirety.

technical field

The present invention relates generally to the field of wireless communications, and more particularly to new and useful systems and methods for non-linear digital self-interference cancellation.

background

Conventional wireless communication systems are half-duplex; that is, they are not capable of simultaneously transmitting and receiving signals on a single wireless communication channel. More recently, work in the field of wireless communications has led to advances in the development of full-duplex wireless communications systems; these systems, if successfully implemented, could provide enormous benefits to the field of wireless communications. For example, the use of full-duplex communication by cellular networks can cut spectrum requirements in half. A major obstacle to successful full-duplex communication is the problem of self-interference. While progress has been made in this area, none of the solutions aimed at addressing self-interference have successfully addressed the conversion of the baseband digital signal to the transmitted RF signal (during transmission) and the received RF signal back to baseband The non-linearity caused by the conversion (during reception) of a digital signal. Therefore, there is a need to create new and useful systems and methods for non-linear digital self-interference cancellation in the field of wireless communications. The present invention provides such new and useful systems and methods.

Brief description of the drawings

Figure 1 is a schematic representation of a full-duplex radio including digital and analog self-interference cancellation;

Figure 2 is a schematic representation of the system of the preferred embodiment;

Figure 3 is a schematic representation of the system of the preferred embodiment;

Figure 4 is a schematic representation of the nonlinear converter of the system of the preferred embodiment;

Figure 5 is a schematic representation of the system of the preferred embodiment;

Figure 6 is a schematic representation of the system of the preferred embodiment;

Figure 7 is a schematic representation of the system of the preferred embodiment;

Figure 8A is an example signal representation of nonlinear distortion in a transmitted signal;

Figure 8B is an example signal representation of predistortion in a transmitted signal;

Figure 9 is a flowchart representation of the steps of the method of the preferred embodiment; and

Figure 10 is a flowchart representation of the non-linear transformation steps of the method of the preferred embodiment.

Description of the preferred embodiment

The following description of the preferred embodiments of the invention is not intended to limit the invention to these preferred embodiments, but to enable any person skilled in the art to make and use the invention.

1. Full-duplex wireless communication system

Wireless communication systems have revolutionized the way the world communicates, and the rapid growth of communication using such systems has provided increased economic and educational opportunities in all regions and industries. Unfortunately, the wireless spectrum required for communication is a finite resource, and the rapid growth in wireless communication has made the availability of this resource an even more scarce one. As a result, spectral efficiency has become increasingly important for wireless communication systems.

A promising solution for increasing spectral efficiency is found in full-duplex wireless communication systems; ie: wireless communication systems capable of transmitting wireless signals and receiving wireless signals on the same wireless channel at the same time. This technique allows for a doubling of spectral efficiency compared to standard half-duplex wireless communication systems.

While full-duplex wireless communication systems are of substantial value to the field of wireless communications, such systems are known to face challenges due to self-interference; since reception and transmission occur on the same channel at the same time, The signal may include undesired signal components from the signal transmitted from the transceiver. As a result, full-duplex wireless communication systems typically include analog and/or digital self-interference cancellation circuitry that reduces self-interference.

A full-duplex transceiver preferably samples the transmit output as a baseband digital signal or as an RF analog signal, but the full-duplex transceiver may additionally or alternatively sample the transmit output in any suitable manner. This sampled transmit output can be used by a full-duplex transceiver to cancel interference from received wireless communication data such as RF analog signals or baseband digital signals, for example. In many full-duplex transceivers, the digital cancellation system works by imposing a scaled version of the transmitted digital baseband signal on the received baseband signal, and the analog cancellation system works by imposing the transmitted function with a scaled version of the RF analog signal. This architecture is generally effective at reducing interference when transceiver components operate in a linear regime, but fails to account for signal non-linearities caused by the conversion of data to the transmitted RF signal (and vice versa). These nonlinearities may become more pronounced as transmitter/receiver power is increased; as a result, full-duplex transceivers without effective nonlinear interference cancellation may be limited in power range due to performance issues.

The systems and methods described herein increase the performance of full-duplex transceivers as shown in FIG. 1 (and other applicable systems) by providing non-linear digital self-interference cancellation. Other suitable systems include active sensing systems (eg, radar), wired communication systems, wireless communication systems, and/or any other suitable systems (including communication systems where transmit and receive frequency bands are close in frequency but do not overlap).

2. System for nonlinear digital self-interference cancellation

As shown in FIG. 2 , the system 100 for nonlinear digital self-interference cancellation includes a pre-processor 110 , a nonlinear transformer 120 , a transform adapter 130 and a post-processor 140 . System 100 may additionally or alternatively include linear converter 150 and/or analog signal sampler 160 .

System 100 functions to reduce self-interference in a full-duplex wireless communication system by canceling non-linear components of self-interference present in digital signals caused by received RF transmissions. Nonlinear digital self-interference cancellation can improve the performance of full-duplex wireless communication systems in many modes of operation; particularly in operating modes in which components of the full-duplex wireless communication system are operating in a substantially non-linear state (e.g., designed to maximize mode of operation to optimize transmission power, power efficiency, etc.). System 100 reduces non-linear digital self-interference by passing the digital transmit signal through pre-processor 110 which samples the digital transmit signal in the transmission path and passes the sampled digital transmit signal to nonlinear transformer 120 . The nonlinear transformer 120 generates a nonlinear self-interference cancellation signal based on the input transmit signal and the transformation configuration set by the transformation adapter 130 . The non-linearly canceled signal is then combined with the digital receive signal originating from the RF receiver through a post-processor 140 to remove self-interference in the digital receive signal. If system 100 includes linear transformer 150, linear transformer 150 preferably operates in parallel with nonlinear transformer 120 to remove both linear and nonlinear components of self-interference in the digital received signal. If system 100 includes analog signal sampler 160, the output of analog signal sampler 160 (e.g., a sample of the RF transmit signal passed through the ADC of analog signal sampler 160) may be used as an input to nonlinear converter 120 And/or used to tune the nonlinear converter 120 (preferably via the converter adapter 130).

System 100 may be implemented using a general purpose processor, digital signal processor, application specific integrated circuit (ASIC), field programmable gate array (FPGA), and/or any suitable processor(s) or circuit(s). System 100 preferably includes memory to store configuration data, but may additionally or alternatively be configured using externally stored configuration data or in any suitable manner.

System 100 is preferably implemented using full-duplex radios. Additionally or alternatively, the system 100 may be implemented as an active sensing system (e.g., radar), a wired communication system, a wireless communication system, and/or any other suitable system (including one in which the transmit and receive bands are close in frequency) but not overlapping communication systems).

The pre-processor 110 functions to sample the digital transmit signal for further processing by a nonlinear transformer 120 as shown in FIG. 3 . The digital transmit signal sampled by pre-processor 110 preferably comprises a digital signal originating from the electronic device destined for an RF transmitter of a full-duplex radio (or other full-duplex wireless communication system). The digital transmit signal sampled by pre-processor 110 may additionally or alternatively include a digital transmit signal from analog signal sampler 160 or from any other suitable source.

The digital transmit signal sampled by the pre-processor 110 is preferably encoded for conversion to an analog signal by the RF transmitter (e.g., encoded by PSK, QAM, OFDM, etc.), but may additionally or alternatively be encoded in any Appropriate way to encode.

Pre-processor 110 preferably samples the digital transmit signal corresponding to the local sampling rate; ie, pre-processor 110 preferably passes all digital transmit data to nonlinear transformer 120 . Additionally or alternatively, pre-processor 110 may sample a subset of the digital transmit signal data; for example, if the digital transmit signal has a native sampling rate of 40 MHz, pre-processor 110 may sample Previously corresponding to a sampling rate of 20MHz every other sample was dropped (whereas the RF transmitter could still receive all samples corresponding to a sampling rate of 40MHz). The pre-processor 110 may additionally or alternatively interpolate the digital transmit signal to increase or decrease the sampling rate. In one example, the pre-processor 110 modifies the sampling rate of the digital transmit signal to match the sampling rate of the RF receiver of the full-duplex radio.

When sampling digital transmit data, preprocessor 110 may perform preprocessing to prepare the sampled digital transmit signal for processing by nonlinear transformer 120 . Pre-processor 110 may include various operators for pre-processing such as scaling, shifting, and/or otherwise modifying the digital transmit signal.

In one implementation, pre-processor 110 modifies the sampled digital transmit signal by removing information that is unlikely to materially affect the output of nonlinear transformer 120 . This may include, for example, discarding samples if they do not represent a change from previous samples above some change threshold. As another example, if the digital transmit signal corresponds to a particular amplitude of the output analog signal, then only digital signal data corresponding to amplitudes above certain amplitude thresholds may be passed to the nonlinear transformer 120 .

If the pre-processor 110 receives digital transmit signals from more than one source (e.g., from both the transmit line preceding the RF transmitter and the analog signal sampler 160), the pre-processor 110 may additionally or alternatively use Any suitable way to combine the signals, or to select one signal over another. For example, preprocessor 110 may pass the average of the two signals to nonlinear transformer 120 . As another example, the pre-processor 110 may prefer analog signal sampler origination over a transmit path digital transmission above a certain transmitter power (and vice versa, at or below that transmitter power). Signal digital transmit signal. The selection and combining of the two signals may depend on any suitable condition.

If preprocessor 110 passes the sampled digital transmit signal to more than one input (e.g., to both nonlinear transformer 120 and linear transformer 150), preprocessor 110 may pass different versions of the sampled digital Transmit signals are provided to different inputs. As a first example, preprocessor 110 may pass the same signal to both nonlinear transformer 120 and linear transformer 150 . As a second example, preprocessor 110 may pass every fourth sample of the digital signal to nonlinear transformer 120 and pass every sample of the digital signal to linear transformer 150 (if the signal is more nonlinearly distorted than This may be useful if the linear distortion varies more slowly). As a third example, preprocessor 110 may separate the sampled digital signal into "linear" and "nonlinear" components, where the "linear" and "nonlinear" components respectively correspond to The linear distortion in the self-interference and the non-linear distortion in the received self-interference contribute to the digital signal components.

The nonlinear transformer 120 functions to transform the sampled digital transmit signal into a nonlinear self-interference signal; ie, a signal representing the assumed contribution of the nonlinear self-interference to the received digital signal. The contribution of nonlinear self-interference may arise from various sources, including components (eg, mixers, power amplifiers, ADCs, DACs, etc.) in both the RF receiver and RF transmitter of a full-duplex radio. Furthermore, the contribution of nonlinear self-interference may vary randomly or with environmental or input conditions (eg transmit power, ambient temperature, etc.).

The nonlinear transformer 120 preferably transforms the sampled digital transmit signal by using a mathematical model suitable for modeling the contribution of nonlinear self-interference from RF transmitters, RF receivers, and/or other sources. Examples of mathematical models that may be used by nonlinear transformer 120 include generalized memory polynomial (GMP) models, Volterra models, and Wiener-Hammerstein models; nonlinear transformer 120 may additionally or alternatively use any combination of models or gather.

The non-linear transformer 120 may additionally or alternatively generate a method for correcting the non-linear A mathematical model for modeling the contribution of self-interference. These models can be generated from previously known models or can be created using neural network and/or machine learning techniques.

Many mathematical models suitable for use in nonlinear converter 120, including the GMP model, model the contribution of nonlinear self-interference as a sum or product of signals with different orders; for example, the general form of GMP is as follows:

where the input signal is represented by x[n], and c _nk represents the coefficient of GMP. The first sum of GMP captures nonlinear self-interference effects that occur based on the current value of the input signal, while the second two terms capture nonlinear self-interference effects determined by the past value of the input signal (called memory effects) .

For these mathematical models, the bandwidth of the k-order term is usually k times larger than the bandwidth of the input signal; for example, if the input signal x[n] has a bandwidth of 40MHz, the third-order term (e.g., x[n]|x[nm] | ² ) will occupy a bandwidth of 120MHz. To avoid problems caused by aliasing, the input signal is preferably sampled at a sampling rate of 120 MHz (three times greater than the original Nyquist sampling rate of 40 MHz). As the number of terms increases, so does the ability to model nonlinear self-interference effects, but so does the minimum sampling rate to avoid aliasing. This presents another problem; the RF transmitter may also have to match this increased sampling rate in order to subtract non-linear digital interfering signals from the received signal. For example, if the GMP model uses seventh order terms, then for the same 40MHz transmit signal, the RF receiver may have to sample the received signal at a rate of 280MHz to avoid aliasing problems (and likewise, the transmit signal may have to be samples at the same rate).

In one embodiment of the invention, the nonlinear transformer 120 works by separating the model used to generate the nonlinear interference signal into components, (each component corresponding to an output order (e.g., a component containing x[n] terms, Another component contains x[n]|x[nm]| ² terms)) to solve these problems. A preferred consequence of this separation is that the sampling rate necessary for each model component to avoid aliasing is known as a function of the component signal order. In this embodiment, the nonlinear transformer includes several transform paths 121 , each of which may include an upsampler 122 , a model component 123 , a filter 124 and a downsampler 125 , as shown in FIG. 4 . Each transform path 121 corresponds to a model component 123 of a particular order; sampling.

Upsampler 122 functions to increase the number of samples contained within the digital transmit signal in order to reduce aliasing effects. It should be noted that upsampling may not be necessary for the first order terms of this model. Upsampler 122 preferably increases the number of samples contained within the digital transmit signal according to linear interpolation, but any suitable method may additionally or alternatively be used. In one example, the upsampler 122 works by creating a sequence containing raw samples separated by L-1 (where L is the upsampling factor) zero values and then passing the new signal through finite impulse response (FIR) low-pass filtering device to upsample the digital transmit signal. In another example, the upsampler 122 creates a sequence containing the original samples separated from each other by L-1 new samples, where each new sample is modeled with respect to how the DAC converts the digital samples into an analog signal ( For example, if the output of the DAC is not perfectly linear from output to output). For a transmit path 121 (and model component 123) of order k, the upsampler 122 preferably upsamples the digital transmit signal by an upsampling factor of k, but may additionally or alternatively by any suitable factor The digital transmit signal is up-sampled.

Model component 123 represents the portion of the model that produces an output of a particular signal order; for example, model component 123 corresponding to order 3 of the GMP model may be represented as

c _n3 x[n]|x[n]| ² +c _n3 x[n]|x[nm]| ²

The model component 123 preferably includes only single-order model terms, but may additionally or alternatively include model terms of more than one order. The model component 123 preferably contains a set of expressions from a generalized memory polynomial (GMP) model, a Volterra model, a Wiener-Hammerstein model, or a neural network model, but may additionally or alternatively contain any suitable model or combination of models part or all.

The model component 123 preferably takes as input the corresponding upsampled digital transmit signal and outputs a non-linear interference signal component.

Filter 124 functions to reduce the bandwidth of the non-linear interfering signal component in preparation for combination with the digital signal received from the RF receiver (or other suitable source). Filter 124 is preferably a digitally implemented FIR low-pass filter, but may additionally or alternatively be any suitable type of filter (e.g., infinite impulse response (IIR) filter, Fourier transform-based filtering device). Filter 124 preferably reduces the bandwidth of the non-linear interfering signal component to match the bandwidth of the digital baseband signal received from the RF transmitter, but may additionally or alternatively act to cap at a frequency lower than that produced by model component 123 The maximum bandwidth of all nonlinear interfering signal components is the bandwidth of the nonlinear interfering signal component for any value. Filter 124 preferably functions to prepare non-linear interfering signal components for downsampling and to remove non-linear interfering signal components not found in the received baseband signal (e.g., if the RF receiver has corresponding low-pass filter or potentially with a corresponding band-pass filter for RF signals).

Downsampler 125 functions to reduce the number of samples contained within the non-linear interference signal component generated by model component 123 (and preferably filtered by filter 124). The downsampler 125 preferably downsamples the non-linear interfering signal components by simply removing the signal at specific time intervals (e.g., throwing away one every other sample to halve the number of samples), but may additionally Or alternatively the non-linear interfering signal component is down-sampled by any suitable method. Downsampler 125 preferably downsamples the non-linear interfering signal components to match the sampling rate of the received digital baseband signal, but may additionally or alternatively downsample the non-linear interfering signal components to any suitable sampling rate.

The non-linear interfering signal components are preferably combined by the non-linear transformer 120 before being sent to the post-processor 140; additionally or alternatively, the non-linear transformer 120 may pass the non-linear interfering signal components to the post-processor 140 without merging them. The nonlinear transformer 120 combines the nonlinear interfering signal components, preferably by adding them, but may additionally or alternatively combine them in any suitable manner (e.g., scaling the components before adding them and/or The components are combined multiplicatively).

The transformation adapter 130 functions to set the transformation configuration of the nonlinear transformer 120 . Transformation adapter 130 may additionally set the transformation configuration of linear transformer 150 (if present); details discussed below for the transformation configuration of nonlinear transformer 120 preferably also apply to the transformation configuration of linear transformer 150 (unless otherwise stated).

The transformation configuration preferably includes the or types of models used by the nonlinear transformer 120 and configuration details related to each model (each individual model is a model type paired with a particular set of configuration details). For example, a transform configuration may configure nonlinear transformer 120 to use a GMP model with a particular set of coefficients. If the model type is static, the transformation configuration may simply include model configuration details; for example, if the model is always a GMP model, the transformation configuration may only include coefficients for the model rather than data specifying the model type.

The transform configuration may additionally or alternatively include other configuration details related to the nonlinear transformer 120 . For example, if the nonlinear transformer 120 includes multiple transformation paths 121, the transformation adapter 130 can set the number of these transformation paths 121, the model order to which their respective model components 123 correspond, the type of filtering used by the filter 124 and/or any other suitable details. In general, the transform configuration may include any details related to the computation or structure of the nonlinear transformer 120 .

The transformation configuration is preferably selected and/or generated by transformation adapter 130 . Transformation adapter 130 may set an appropriate transformation configuration by selecting from stored static configurations, from dynamically generated configurations, or by any other suitable means or combination of means. For example, transform adapter 130 may select from three static transform configurations based on their suitability for particular signal and/or environmental conditions (the first for low transmitter power, the second for moderate transmit transmitter power and the third is suitable for high transmitter power). As another example, the transformation adapter 130 may dynamically generate configurations based on signal conditions and/or environmental conditions; the coefficients of the GMP model are set by formulas that take transmitter power, temperature, and receiver power as inputs.

The transformation adapter 130 preferably sets the transformation configuration (whether the transformation configuration is selected from a static set of configurations or generated from a formula or model) based on various input data. The input data used by the transformation adapter 130 may include static environmental and system data (e.g., receiver operating characteristics, transmitter operating characteristics, receiver elevation above sea-level), dynamic environmental and system data (e.g., , current ambient temperature, current receiver temperature, average transmitter power, ambient humidity) and/or system configuration data (e.g., receiver/transmitter settings), signal data (e.g., digital transmit signal, RF transmit signal, RF receive signal, digital receive signal). If system 100 uses analog signal sampler 160, transform adapter 130 may additionally or alternatively use the output of analog signal sampler 160 as input for setting the transform configuration. Transformation adapter 130 may additionally or alternatively generate and/or use models based on the input data to set transformation configurations; for example, a transmitter manufacturer may give a model to predict the internal temperature of a transmitter based on transmitter power, and The transform adapter 130 can use the output of this model (given the transmitter power) as input data for setting the transform configuration.

The transformation adapter 130 may set the transformation configuration at any time, but preferably even transforms the configuration in response to a time threshold or other input data threshold being crossed. For example, transformation adapter 130 may reset the transformation configuration every ten seconds based on changed input data values. As another example, the transition adapter 130 may reset the transition configuration whenever a transmitter power threshold is crossed (e.g., whenever the transmitter power is increased by ten percent from the last transition configuration setting or whenever the transmitter power is increased by more than some static value).

If the system 100 is connected to a full-duplex wireless communication system that also has an analog canceller, the transform adapter 130 can cooperate with the analog canceller (e.g., set the transform configuration based on data from the analog canceller or make the transform configuration setup time analog canceller coordination) to reduce overall self-interference (or for any other suitable reason).

Transformation adapter 130 preferably adjusts the transformation configuration and/or the transformation configuration generation algorithm (ie, the algorithm that dynamically generates the transformation configuration) to reduce self-interference for a given set of transmit signals and system/environmental conditions. Transformation adapter 130 may use analytical methods, online gradient descent methods (eg, LMS, RLMS), and/or any other suitable method to tune the transformation configuration and/or the transformation configuration generation algorithm. Adjusting the transformation configuration preferably includes changing the transformation configuration based on learning. In the case of a neural network model, this may include changing the structure and/or weightings of the neural network based on test inputs. In the case of a GMP polynomial model, this may include optimizing the GMP polynomial coefficients according to a gradient descent method.

Transformation adapter 130 may be based on testing input scenarios (e.g., scenarios where the signal received by the RF receiver is known), scenarios where there is no input (e.g., the only signal received at the RF receiver is the one transmitted by the RF transmitter signal) or scenarios where the received signal is unknown to adjust the transform configuration. In cases where the received signal is an unknown signal, the transform adapter 130 may adjust the transform configuration based on historically received data (eg, what the signal looked like ten seconds ago) or any other suitable information. Transformation adapter 130 may additionally or alternatively adjust the transformation configuration based on the content of the transmitted signal; for example, if the transmitted signal is modulated in a particular way, transformation adapter 130 may look for this in the self-interfering signal. the same modulation; more specifically, the transform adapter 130 can adjust the transform configuration such that when the self-interfering signal is combined with the digital received signal, the remaining modulation (as an indicator of self-interference) is reduced (compared to the previous transform configuration ).

The post-processor 140 functions to combine the nonlinear self-interference signal generated by the nonlinear transformer 120 with the digital signal received by the RF receiver, as shown in FIG. 5 . The post-processor 140 preferably combines the nonlinear self-interference signal from the nonlinear transformer 120 with the digital receive signal from the RF receiver of the full-duplex wireless communication system. Optionally or alternatively, the post-processor 140 may combine the linear self-interference signal from the linear transformer 150 with the digital receive signal from the RF receiver of the full-duplex wireless communication system. Post-processor 140 may additionally or alternatively combine the linear or non-linear self-interference signal with any suitable digital received signal. The digital received signal entering the post-processor 140 is preferably encoded, (e.g., encoded by PSK, QAM, OFDM, etc.) for conversion to an analog signal by the RF transmitter, but may additionally or alternatively be encoded in any Appropriate way to encode.

Post-processor 140 may perform post-processing to prepare the self-interfering signal for combination with the digital received signal; this may include scaling, shifting, filtering, and/or otherwise modifying the self-interfering signal. For example, post-processor 140 may include a low-pass filter designed to filter out high-frequency components of the generated self-interfering signal (eg, to match the corresponding bandwidth of the RF transmitter). If system 100 includes linear transformer 150, the post-processor may combine the outputs of linear transformer 150 and nonlinear transformer 120 before combining (possibly weighted) the two signals with the combination of the digital received signal. Post-processor 140 preferably matches the sampling rate of the self-interfering signal output by nonlinear transformer 120 and linear transformer 150 to the sampling rate of the output of the RF receiver through upsampling and/or downsampling as previously described, but The sampling rate of the self-interfering signal may additionally or alternatively be left unchanged or set to a different sampling rate than the sampling rate of the RF receiver output.

Post-processor 140 may combine the output from nonlinear transformer 120 and the output from linear transformer 150 (if present) in any suitable manner (including combining nonlinear or linear self-interfering signal components). For example, post-processor 140 may combine the output from nonlinear transformer 120 and the output from linear transformer 150 as a weighted sum. As another example, post-processor 140 may select the output from one of the two transformers (or may select a subset of the outputs from either or both; for example, two of the five non-linear self-interfering signal components indivual). If the pre-processor 110 separates the digital transmit signal between the linear transformer 150 and the nonlinear transformer 120, the post-processor 140 can build on this separation (e.g., by performing a join operation, which is the approximate inverse of the separation operation). ) to recombine the digital signals.

The linear transformer 150 functions to transform the sampled digital transmit signal into a linear self-interference signal; ie, a signal representing the assumed contribution of the linear self-interference to the received digital signal. As with nonlinear self-interference, contributions from linear self-interference can arise from various sources. Nonlinearity often arises from the nonlinear behavior of typical wireless transmitter components, while real wireless channels can often be very linear in response. While it is possible to model the entire self-interfering signal with a larger non-linear model, for performance reasons it is often advantageous to use a hybrid model in which the transmitter (and possibly the receiver) The nonlinearity is explained by a smaller nonlinear model and the wireless channel response is explained by a linear model (in system 100, these models may be implemented in nonlinear transformer 120 and linear transformer 150, respectively). Separating the models in this way may allow for a significant reduction in the number of calculations performed to generate the self-interfering signal. Additionally, while nonlinearities in a transmitter or receiver can be relatively stable over time (e.g., over a period of minutes or hours), linear self-interference effects that exist in wireless channels can change very rapidly (eg, over a period of milliseconds). Using separate models allows the simpler linear self-interference model to be tuned and tuned at a fast rate without having to also tune and tune the computationally more complex non-linear self-interference model. This concept can be extended to have separate nonlinear models for the transmitter and receiver, as shown in FIG. 6 .

Linear transformer 150 preferably transforms the sampled digital transmit signal by using a mathematical model suitable for modeling the contribution of linear self-interference from RF transmitters, RF receivers, wireless channels, and/or other sources. Examples of mathematical models that may be used by linear transformer 150 include generalized memory polynomial (GMP) models, Volterra models, and Wiener-Hammerstein models; nonlinear transformer 120 may additionally or alternatively use any combination or collection of models .

The linear converter 150 may additionally or alternatively generate a function for linear self-interference based on a comparison of the sampled digital transmit signal with the received signal (from the analog signal sampler 150, the receive path, or any other suitable source). Contributions are modeled in mathematical models. These models can be generated from previously known models, or can be created using neural network and/or machine learning techniques.

Analog signal sampler 160 functions to provide digital signals to system 100 that are converted from RF transmit signals (and/or baseband or intermediate frequency analog signals). The digital signal differs from the digital transmit signal in that it may contain non-linearities caused by the conversion of the digital transmit signal to the RF transmit signal (and/or baseband or IF analog transmit signal) and back again, but is also different from RF receives the signal in that it results from a different signal path (eg analog signal is sampled before reaching the antenna). Accordingly, analog signal sampler 160 may be used to provide information to nonlinear converter 120 , linear converter 150 and/or conversion adapter 130 that may not be contained in the digital transmit signal. Analog signal sampler 160 output is preferably directed by preprocessor 110 to a suitable source, but analog signal sampler 160 may additionally or alternatively output to any suitable portion of system 100 (including nonlinear transformer 120 and/or or conversion adapter 130).

In a variation of the preferred embodiment, the system 100 includes a digital predistortion circuit (DPD) 170, as shown in FIG. Since most of the nonlinearities in full-duplex wireless communication systems arise from the components of the RF transmitter, and these nonlinearities may contribute to the reduced power efficiency of the RF transmitter, reducing the nonlinear component of the RF transmit signal may is advantageous (both from the perspective of increasing transmitter efficiency and in order to reduce the amount of non-linear self-interference cancellation required). An example of non-linear distortion that occurs when converting a digital transmit signal to an RF transmit signal is shown in FIG. 8A. One way of doing this involves pre-distorting the digital transmit signal so that the distortion in the digital transmit signal is used to correct for the distortion introduced by the RF transmitter when converting the digital transmit signal to an RF transmit signal, as in Figure 8B shown.

DPD 170 preferably employs samples (which may be digital or analog) from the output of the RF transmitter to measure the nonlinearity inherent in the output of the RF transmitter. DPD 170 preferably receives samples from analog signal sampler 160, but may additionally or alternatively receive them from any suitable source. Based on the RF transmitter output samples, DPD 170 transforms the digital transmit signal to create an "inverse" nonlinearity in the signal (as shown in Figure 8B). This "inverse" nonlinearity reduces the nonlinearity present in the final RF transmit signal when further transformed by the RF transmitter (during the conversion of the digital transmit signal to an RF transmit signal).

Predistortion (or other linearization techniques) provided by DPD 170 or other suitable sources may be utilized to reduce the complexity of digital self-interference cancellation. By placing DPD 170 (as shown in FIG. 7 ) after preprocessor 110 in the digital transmit signal path, non-linearities in the receive signal path are reduced, and additionally, non-linear converter 120 does not need to be sensitive to the digital transmit signal. The transformation is done to remove the non-linearity introduced by DPD 170 (as it may be required if DPD 170 is present before pre-processor 110 in the digital transmit signal path).

2. Method for nonlinear digital self-interference cancellation

As shown in FIG. 9, the method 200 for nonlinear digital self-interference cancellation includes receiving a digital transmission signal S210, transforming the digital transmission signal into a nonlinear self-interference signal according to a transformation configuration S220, and transforming the nonlinear self-interference signal Combining with the digital received signal S230. The method 200 may additionally include preprocessing S215 the digital transmission signal, transforming the digital transmission signal into a linear self-interference signal S225, dynamically adjusting the transformation configuration S240 and/or performing digital predistortion S250 on the digital transmission signal.

Method 200 functions to reduce self-interference in a full-duplex wireless communication system by canceling non-linear components of self-interference present in digital signals caused by received RF transmissions. Nonlinear digital self-interference cancellation can improve the performance of full-duplex wireless communication systems in many modes of operation; in particular, operating modes in which the components of the full-duplex wireless communication system operate in a substantially non-linear state (e.g., designed to maximize mode of operation to maximize transmit power, power efficiency, etc.). The method 200 reduces nonlinear digital self-interference in a full-duplex wireless communication system by sampling the digital transmission signal (step S210). The received digital transmission signal may be preprocessed (possibly into linear and nonlinear components) during step S215, after which the digital transmission signal may be transformed into a nonlinear self-interference signal according to the transformation configuration (step S220), And optionally transformed into a linear self-interference signal (step S225). The self-interference signal is then combined with the digital received signal of the full-duplex wireless communication system (step S230) in order to reduce the self-interference signal present in the signal received by the wireless communication system. The method 200 may also include dynamically adjusting the transform configuration (step S240) in order to increase the effectiveness of self-interference reduction due to the non-linear self-interference transform and/or digitally predistort the digital transmit signal prior to the transmitter of the wireless communication signal (step S250) in order to reduce the amount of non-linear self-interference present in the received digital transmit signal (this may reduce the computing power required for the calculations in steps S220, S225 and/or S240).

Method 200 is preferably implemented by system 100, but may additionally or alternatively be implemented by any suitable system for non-linear digital self-interference cancellation for use with a full-duplex wireless communication system. Additionally or alternatively, method 200 may be employed using an active sensing system (e.g., radar), a wired communication system, a wireless communication system, and/or any other suitable system (including where the transmit and receive bands are similar in frequency but non-overlapping communication systems).

Step S210 includes receiving a digital transmission signal. Step S210 functions to provide a digital signal intended for transmission by a full-duplex wireless communication system so that the signal can be used to remove self-interference at a full-duplex wireless communication system receiver. The digital transmission signal received at S210 preferably comprises a digital signal originating from the electronic device and destined for an RF transmitter of a full-duplex radio (or other full-duplex wireless communication system). The digital transmit signal received in S210 may additionally or alternatively comprise a digital transmit signal converted from an analog transmit signal (e.g., an RF transmit signal of an RF transmitter of a full-duplex radio) or from any other suitable source . The digital transmission signal received in S210 is preferably encoded (e.g., by PSK, QAM, OFDM, etc. for conversion to an analog signal by the RF transmitter, but may additionally or alternatively be done in any suitable manner coding.

Step S215 includes preprocessing the digital transmission signal. Step S215 functions to perform initial processing (if necessary) of the digital transmission signal received in S210. Step S215 preferably includes preprocessing all data received in S210; additionally or alternatively, S215 may include sampling a subset of the digital transmit signal data; for example, if the digital transmit signal has a native sampling rate of 40 MHz , then S215 may include discarding every other sample as part of preprocessing corresponding to a sampling rate of 20MHz. Step S215 may additionally or alternatively include up-sampling or down-sampling the digital transmission signal to increase or decrease the sampling rate. In one example, S215 includes modifying the sampling rate of the digital transmit signal to match the sampling rate of the RF receiver of the full-duplex radio.

In one implementation, S215 includes modifying the digital transmit signal by removing information that is unlikely to substantially affect the result of the nonlinear transformation. This may include, for example, removing a signal component if the signal component does not represent a change above a certain change threshold on a previous signal component. As another example, if the digital transmit signal corresponds to a particular amplitude of the output analog signal, digital signal data corresponding to amplitudes below a certain amplitude threshold may be removed.

If multiple digital transmit signals are received in S210 (e.g., from both the transmit line and the analog signal sampler prior to the RF transmitter), S215 may include combining the signals in any suitable manner or may be superior to Another signal selects a signal. For example, two signals may be combined by taking the average of the two signals or summing them. As another example, S215 may include selecting a simulation of a digital transmit signal originating superior to a transmit path digital transmit signal above a certain transmitter power (and vice versa at or below that transmitter power). Signal sampler.

Step S215 may additionally or alternatively perform any pre-processing to prepare the digital transmit signal for nonlinear and/or linear transformation. This may include scaling, shifting and/or otherwise modifying the digitally transmitted signal. If the digital transmit signal is subjected to more than one type of transformation (eg, both linear and non-linear) as part of method 200, then S215 may include providing different versions of the digital transmit signal for different transformation types. As a first example, S215 may include providing the same signal for linear transformation and nonlinear transformation. As a second example, S215 may include providing every fourth sample of the digital signal for nonlinear transformations and every sample of the digital signal for linear transformations (if the nonlinear distortion of the signal varies more slowly than the linear distortion, then This may be useful). As a third example, S215 may include separating the sampled digital signal into "linear" components and "nonlinear" components, where the "linear" and "nonlinear" components respectively correspond to Linear distortions in interference and non-linear distortions in received self-interference contribute to the digital signal components.

Step S220 includes transforming the digital transmit signal into a non-linear self-interfering signal according to the transform configuration. Step S220 functions to transform the digital transmitted signal into a non-linear self-interference signal; ie, a signal representing the assumed contribution of non-linear self-interference to the received digital signal. The contribution of nonlinear self-interference can arise from various sources, including components (eg, mixers, power amplifiers, ADCs, DACs, etc.) in both the RF receiver and the RF transmitter of a full-duplex radio. Furthermore, the contribution of nonlinear self-interference may vary randomly or with environmental or input conditions (eg transmit power, ambient temperature, etc.).

Step S220 preferably includes transforming the digitally transmitted signal by using mathematical models substantially similar to those described in the description of system 100, but may additionally or alternatively be transformed according to any model or set of models The digital transmit signal is transformed. Step S220 may additionally or alternatively include generating a contribution to nonlinear self-interference based on a comparison of the sampled digital transmit signal with the received signal (from analog signal sampling, the receive path, or any other suitable source) The model to model.

In one embodiment of the present invention, S220 transforms the digital transmission signal by using an order separation model similar to the one described in the system 100, as shown in FIG. 10 . In this embodiment, S220 may include transforming the digital transmission signal according to transform paths, where each transform path corresponds to a component of an order separation model. Step S220 preferably includes transforming the digital transmit signals simultaneously in parallel for each transform path, but they may additionally or alternatively be transformed in series and/or at different times. While S220 preferably includes transforming the same digital transmit signal for each transform path, the digital transmit signal may additionally or alternatively be processed in any suitable manner (e.g., S220 may include separating the digital transmit signal for each component and pass a separate component to each transform path).

For each transformation path, S220 preferably includes upsampling S221 the digital transmit signal, transforming S222 the digital transmit signal with a model component, filtering S223 the transformed signal, and downsampling S224 the transformed signal. Additionally or alternatively, S220 may only include transforming S222 the digital transmission signal with a model component and/or filtering S223 the transformed signal (eg, with a first-order model component). Step S220 preferably also includes combining the transformed signals to form a single nonlinear self-interference signal, but may additionally or alternatively not combine the transformed signals. Step S220 combines nonlinear interfering signal components preferably by adding them, but may additionally or alternatively combine them in any suitable manner (e.g., scaling the components before adding them and/or multiplicatively combine the components).

Upsampling the digital transmit signal S221 acts to increase the number of samples contained within the digital transmit signal in order to reduce aliasing effects. It should be noted that upsampling may not be necessary for the first order terms of this model. Step S221 increases the number of samples contained within the digital transmit signal, preferably according to linear interpolation, but any suitable method may additionally or alternatively be used. In one example, S221 consists of creating a sequence containing raw samples separated by L-1 (where L is the upsampling factor) zero values and then passing the new signal through a finite impulse response (FIR) low-pass filter. Upsampling of the digital transmit signal. In another example, S221 includes creating a sequence comprising original samples separated from each other by L-1 new samples, where each new sample is modeled with respect to how the DAC converts the digital samples into an analog signal (e.g., if The output of the DAC is not perfectly linear between outputs. For a transmit path of order k (and model components), S221 preferably includes upsampling the digital transmit signal with an upsampling factor of k, but may additionally or alternatively The digitally transmitted signal is upsampled by any suitable factor.

Transform digital transmit signal with model component S222 functions to transform the digital transmit signal into nonlinear interfering signal components based on the portion of the model that produces an output of a particular signal order; for example, a model component corresponding to order 3 of the GMP model may be marked as

c _n3 x[n]|x[n]| ² +c _n3 x[n]|x[nm]| ²

The model component preferably includes only single-order model terms, but may additionally or alternatively include model terms of more than one order. The model component preferably contains a set of expressions from a generalized memory polynomial (GMP) model, a Volterra model, a Wiener-Hammerstein model or a neural network model, but may additionally or alternatively contain part of any suitable model or combination of models or all.

Filtering the transformed signal S223 functions to reduce the bandwidth of the non-linear interfering signal components in preparation for combining with the digital signal received from the RF receiver (or other suitable source). Filtering is preferably implemented using a digitally implemented FIR low-pass filter, but any suitable type of filter (e.g. infinite impulse response (IIR) filter, Fourier transform based filter) may additionally or alternatively be used ). Step S223 preferably involves reducing the bandwidth of the non-linear interfering signal component to match the bandwidth of the digital baseband signal received from the RF transmitter, but may additionally or alternatively act to limit it below that of all non-linear interfering signal components produced by the model components. Maximum bandwidth of linear interfering signal component Bandwidth of non-linear interfering signal component for any value. Step S223 preferably functions both to prepare nonlinear interfering signal components for downsampling and to remove non-linear interfering signal components not found in the received baseband signal (e.g. if the RF receiver has a corresponding low-pass filter, or potentially a corresponding band-pass filter for RF signals).

Downsampling Transformed Signal S224 functions to reduce the number of samples contained within the non-linear interfering signal component. Step S224 preferably involves downsampling the non-linear interfering signal components by simply removing signals at specific time intervals (e.g., throwing away one every other sample to halve the number of samples), but may additionally or The non-linear interfering signal component may alternatively be down-sampled by any suitable method. Step S224 preferably down-samples the nonlinear interfering signal component to match the sampling rate of the received digital baseband signal, but may additionally or alternatively down-sample the non-linear interfering signal component to any suitable sampling rate.

Step S225 includes transforming the digital transmission signal into a linear self-interference signal. Step S225 functions to transform the sampled digital transmit signal into a linear self-interference signal; ie, a signal representing the assumed contribution of linear self-interference to the received digital signal. As with nonlinear self-interference, contributions from linear self-interference can arise from various sources. Nonlinearity often arises from the nonlinear behavior of typical wireless transmitter components, while real wireless channels can often be very linear in response. While it is possible to model the entire self-interfering signal with a larger nonlinear model, it is often advantageous for performance reasons to use a hybrid model in which the nonlinearity of the transmitter (and potentially the receiver) is determined by the larger A small nonlinear model is explained and the wireless channel response is explained by a linear model. Separating the models in this way may allow for a significant reduction in the number of calculations performed to generate the self-interfering signal. Additionally, while nonlinearities in a transmitter or receiver can be relatively stable over time (e.g., over a period of minutes or hours), linear self-interference effects that exist in wireless channels can change very rapidly (eg, over a period of milliseconds). Using the split mode allows the simpler linear self-interference model to be tuned and tuned at a very fast rate without having to also tune and tune the computationally more complex non-linear self-interference model. This concept can be extended to have separate nonlinear models for the transmitter and receiver, as shown in FIG. 6 .

Step S225 preferably includes transforming the sampled digital transmit signal by using a mathematical model suitable for modeling the contribution of linear self-interference from RF transmitters, RF receivers, wireless channels and/or other sources. Examples of mathematical models that may be used include generalized memory polynomial (GMP) models, Volterra models, and Wiener-Hammerstein models; S225 may additionally or alternatively include using any combination or collection of models.

Step S230 includes combining the nonlinear self-interference signal with the digital received signal. Step S230 functions to combine the non-linear self-interference signal with the digital signal received by the RF receiver. Step S230 preferably includes combining the non-linear self-interference signal with the digital received signal from the RF receiver of the full-duplex wireless communication system; additionally or alternatively, S230 may include combining the linear self-interference signal with the digital reception signal from the full-duplex wireless communication system. The digital received signals of the RF receivers of the wireless communication system are combined. Step S230 may additionally or alternatively comprise combining the linear or non-linear self-interference signal with any suitable digital received signal.

Step S230 may include performing post-processing to prepare the self-interfering signal for combination with the digital received signal; this may include scaling, shifting, filtering and/or otherwise modifying the self-interfering signal. For example, S230 may include processing the self-interfering signal with a low pass filter designed to filter out high frequency components (eg, to match the corresponding bandwidth of the RF transmitter). Step S230 may include matching the sampling rate of the self-interfering signal to that of the output of the RF receiver by up-sampling and/or down-sampling as described above, but may additionally or alternatively not change the sampling rate of the self-interfering signal rate or set the sampling rate of the self-interfering signal to a sampling rate different from that of the RF receiver output.

Step S230 may include combining the linear self-interference signal and the nonlinear self-interference signal (including combining the nonlinear self-interference signal component or the linear self-interference signal component) in any suitable manner. For example, S230 may include combining the linear self-interference signal and the nonlinear self-interference signal into a weighted sum.

Step S240 includes dynamically adjusting the transformation configuration. Step S240 functions to update and/or change the transform configuration (and potentially also the parameters of the linear transform) used in the nonlinear transform based on changes in signal or environmental conditions. The transformation configuration is preferably as described in the system 100 description, but may additionally or alternatively contain any parameter or set of parameters corresponding to the nonlinear or linear signal transformation performed in S220 and S225.

Dynamically adjusting the transformation configuration S240 may include setting an updated transformation configuration by selecting from a stored static configuration, from generating a new transformation configuration, or by any other suitable means or combination of means. For example, S240 may include selecting from three static transform configurations based on their applicability to specific signal and/or environmental conditions (the first for low transmitter power, the second for medium transmitter power and the third is suitable for high transmitter power). As another example, S240 may include generating an updated configuration based on signal and/or environmental conditions; if the GMP model is used in a nonlinear transformation, the coefficients of the GMP model may be determined by using transmitter power, temperature and receiver power as Enter the formula to calculate.

Step S240 may include setting the transformation configuration (whether the transformation configuration is selected from a set of static configurations or generated from a formula or model) based on various input data. Input data can include static environmental and system data (e.g., receiver operating characteristics, transmitter operating characteristics, receiver altitude), dynamic environmental and system data (e.g., current ambient temperature, current receiver temperature, average transmit machine power, ambient humidity) and/or system configuration data (e.g., receiver/transmitter settings), signal data (e.g., digital transmit signal, RF transmit signal, RF receive signal, digital receive signal).

Step S240 may be performed at any time, but is preferably performed in response to a time threshold or other input data threshold being crossed. For example, S240 may adjust the transformation configuration every ten seconds according to the changed input data value. As another example, the change configuration may be reset whenever a transmitter power threshold is crossed (e.g., whenever the transmitter power is increased by ten percent from the last change configuration setting, or whenever the transmitter power is increased by more than a certain static value).

Step S240 may additionally or alternatively include coordinating with the analog cancellation method of the full-duplex radio (if present); for example, adjusting the transformation configuration based on the analog cancellation data, or coordinating the transformation configuration setting times based on the analog cancellation configuration to To reduce overall self-interference (or for any other suitable reason).

Step S240 preferably adjusts the transform configuration to reduce self-interference for a given set of transmitted signals and system/environmental conditions. Step S240 may use analytical methods, online gradient descent methods, least mean square (LMS) methods, recursive least squares (RLS) methods, regularization and constraint solver methods (such as LASSO) and/or any other suitable method to Adjust the transform configuration and/or the transform configuration generation algorithm. LMS methods can include regularization, an example LMS method includes leaky LMS; RLS methods can also include regularization. Adjusting the transformation configuration preferably includes changing the transformation configuration based on learning. In the case of a neural network model, this may include changing the structure and/or weightings of the neural network based on test inputs. In the case of a GMP polynomial model, this may include optimizing the GMP polynomial coefficients according to a gradient descent method.

Step S240 may additionally or alternatively include scenarios based on test inputs (e.g., scenarios when the signal received by the RF receiver is known), scenarios in which there is no input (e.g., when the only signal received by the RF receiver The signal is the signal transmitted by the RF transmitter) or scenarios where the received signal is unknown adjust the transform configuration. In cases where the received signal is an unknown signal, the transformation configuration may be adjusted based on historically received data (eg, what the signal looked like ten seconds ago) or any other suitable information. The transformation configuration may additionally or alternatively be updated based on the content of the transmitted signal.

Step S250 includes using a digital pre-distortion circuit to perform digital pre-distortion on the digital transmission signal. Step S250 functions to increase transmitter efficiency and/or reduce the amount of nonlinear self-interference cancellation required by full-duplex radios. Since most of the nonlinearities in full-duplex wireless communication systems arise from the components of the RF transmitter, and these nonlinearities may contribute to the reduced power efficiency of the RF transmitter, reducing the nonlinear component of the RF transmit signal may is advantageous (both from the perspective of increasing transmitter efficiency and in order to reduce the amount of non-linear self-interference cancellation required). An example of non-linear distortion that occurs when converting a digital transmit signal to an RF transmit signal is shown in FIG. 8A. One way of doing this involves pre-distorting the digital transmit signal so that the distortion in the digital transmit signal is used to correct for the distortion introduced by the RF transmitter when converting the digital transmit signal to an RF transmit signal, as in Figure 8B shown.

Step S250 preferably includes using samples (which may be digital or analog) from the output of the RF transmitter to measure the nonlinearity inherent in the output of the RF transmitter. Based on the RF transmitter output samples, the digital transmit signal is transformed to create an "inverse" nonlinearity in the signal (as shown in Figure 8B). This "inverse" nonlinearity reduces the nonlinearity present in the final RF transmit signal when further transformed by the RF transmitter (during the conversion of the digital transmit signal to an RF transmit signal).

Predistortion (or other linearization techniques) can be utilized to reduce the complexity of digital self-interference cancellation. By performing predistortion after preprocessing in the signal path, as shown in Figure 7, the nonlinearity in the receive signal path is reduced, and in addition, the nonlinear transformation does not require transformation of the digital transmit signal to remove the Non-linearity introduced by predistortion.

Step S250 may additionally or alternatively include adjusting the digital predistortion circuit to account for varying RF transmit signal distortion characteristics. Adjusting the digital predistortion circuitry is preferably done using techniques substantially similar to those used to update the transform configuration, but may additionally or alternatively be performed using any suitable technique or system. The predistortion characteristics of the digital predistortion circuit are preferably adjusted based on samples of RF transmit signals of the full duplex wireless communication system, but may additionally or alternatively be adjusted based on any suitable input.

The methods of the preferred embodiments and their variations may be at least partially embodied and/or implemented as a machine configured to receive a computer-readable medium storing computer-readable instructions. The instructions are preferably executed by computer-executable components, preferably integrated with the system for non-linear self-interference cancellation. The computer readable medium may be stored on any suitable computer readable medium such as RAM, ROM, flash memory, EEPROM, optical device (CD or DVD), hard drive, floppy disk drive, or any suitable device. The computer-executable components are preferably general-purpose or special-purpose processors, but any suitable special-purpose hardware or hardware/firmware combination device may alternatively or additionally execute the instructions.

As will be appreciated by those skilled in the art from the foregoing detailed description and from the accompanying drawings and claims, modifications and changes may be made to the preferred embodiments of the invention without departing from the invention as defined in the appended claims the scope of the invention.

Claims (14)

1. a kind of system eliminated for non-linear, digital self-interference, including：

● preprocessor, it is communicatively coupled to the digitally transmitted signals of full-duplex wireless communication systems, the preprocessor From the digitally transmitted signals of the pretreatment of digitally transmitted signals generation first and the digitally transmitted signals of the second pretreatment；

● non-linear converter, its according to alternate arrangement by first digitally transmitted signals pre-processed be transformed to it is non-linear from Interference signal；Wherein, the non-linear converter includes the first transform path and the second transform path；

Wherein, first transform path includes doing certainly from the digitally transmitted signals generation first order nonlinear of the described first pretreatment Disturb the model component of component of signal；

Wherein, second transform path include to described first pretreatment digitally transmitted signals by three times up-sampled with Create the up-sampler of the signal of up-sampling, the mould from the signal generation third-order non-linear self-interference signal component of the up-sampling Type component, remove from the third-order non-linear self-interference signal component low pass filter of undesirable high fdrequency component and to institute State down-sampler of the third-order non-linear self-interference signal component by 1/3rd times of progress down-samplings；

● conversion adapter, it sets the alternate arrangement of the non-linear converter；And

● preprocessor, it is by the digital received signals of the non-linear self-interference signal and the full-duplex wireless communication systems Merge；

Wherein, the first order nonlinear self-interference signal component and the filtered and third-order non-linear self-interference of down-sampling letter Number component is merged to create the non-linear self-interference signal.

2. system according to claim 1, wherein, the non-linear converter is from the first order nonlinear self-interference signal Component and the described filtered and weighted sum of the third-order non-linear self-interference signal component of down-sampling create described non-linear Self-interference signal.

3. system according to claim 1, in addition to linear quantizer, the linear quantizer is according to alternate arrangement by institute The digitally transmitted signals for stating the second pretreatment are transformed to linear self-interference signal.

4. system according to claim 3, wherein, the digitally transmitted signals of first pretreatment and the second pre- place The digitally transmitted signals of reason are identical signals.

5. system according to claim 3, wherein, the digitally transmitted signals are separated into non-linear by the preprocessor Component and linear component, wherein, the digitally transmitted signals of first pretreatment are made up of the nonlinear component, and described The digitally transmitted signals of second pretreatment are made up of the linear component.

6. system according to claim 1, wherein, the non-linear converter uses General Memory multinomial model.

7. system according to claim 6, wherein, the alternate arrangement includes General Memory multinomial coefficient, and institute General Memory multinomial coefficient is stated to be adjusted according to gradient optimization algorithm by the conversion adapter.

8. system according to claim 1, in addition to digitally transmitted signals execution digital pre-distortion and by advance The digitally transmitted signals of distortion be delivered to the emitter of the full-duplex wireless communication systems digital predistortion circuit and will be through The sampling for the RF transmission signals that the emitter generates from the digitally transmitted signals of the predistortion is supplied to the mould of the system Intend signal sampler, wherein, the sampling of the RF transmission signals is used to adjust the digital predistortion circuit.

9. a kind of method eliminated for non-linear, digital self-interference, including：

● receive the digitally transmitted signals of full-duplex wireless communication systems；

● the digitally transmitted signals are transformed to by non-linear self-interference signal according to alternate arrangement；Wherein, the numeral is converted Transmission signal includes：The first non-linear self-interference signal component and the second non-linear self-interference signal component are generated, and is merged The first non-linear self-interference signal component and the second non-linear self-interference signal component with create it is described it is non-linear from Interference signal；Wherein, generating the first non-linear self-interference signal component includes handling the numeral with first order modeling component Transmission signal, and generate the described second non-linear self-interference signal component and include：The digitally transmitted signals are entered by three times Row up-sampling handles the signal generation third-order non-linear of the up-sampling certainly to create the signal of up-sampling, with third-order model component Interference signal component, the third-order non-linear self-interference signal component is filtered, and it is non-to filtered three rank Linear self-interference signal component is by 1/3rd times of progress down-samplings；And

● the non-linear self-interference signal and the digital received signals of the full-duplex wireless communication systems are merged.

10. according to the method for claim 9, in addition to based on the non-linear self-interference signal and the digital received believe Number dynamically adjust the alternate arrangement.

11. according to the method for claim 10, wherein, adjusting the alternate arrangement includes being adjusted according to gradient descent algorithm The whole alternate arrangement.

12. it is according to the method for claim 9, in addition to by the digitally transmitted signals pretreatment linear digital transmitting letter Number and non-linear, digital transmission signal, in addition to the linear digital transmission signal is transformed to linear self-interference signal, wherein, The digitally transmitted signals are transformed into non-linear self-interference signal includes the non-linear, digital transmission signal being transformed to institute State non-linear self-interference signal.

13. according to the method for claim 9, also the digitally transmitted signals are entered including the use of digital predistortion circuit Row digital pre-distortion.

14. according to the method for claim 13, wherein, the digital predistortion circuit is led to according to the full duplex radio The samplings of the RF transmission signals of letter system is adjusted.