Method and device for suppressing and demodulating back scattering adjacent channel interference of cellular passive internet of things

Abstract

The embodiment of the invention discloses an adjacent channel interference suppression demodulation method and device for backscatter communication of a cellular passive internet of things, and relates to the technical field of wireless communication. The method comprises the following steps: acquiring an OFDM signal, respectively carrying out signal processing and deconvolution on the OFDM signal to obtain a modulation signal and a sending signal, and training a neural network to obtain a reconstruction model; the method comprises the following steps: receiving a time domain OFDM signal, reconstructing the time domain OFDM signal through a reconstruction model, and obtaining an adjacent band interference signal generated by leakage of the time domain OFDM signal through a band-pass filter; and performing Fourier transform interference elimination on the time domain OFDM signal and the adjacent band interference signal, performing maximum ratio combination, and performing maximum likelihood detection to obtain original information corresponding to the time domain OFDM signal. According to the method, the OFDM signal is effectively processed by constructing the neural network reconstruction model, high-precision reconstruction of the signal and precise identification and elimination of adjacent band interference are realized, and the problems of high error rate and low communication quality of a communication system due to the fact that interference cannot be suppressed in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and in particular to a cellular passive Internet of Things backscattered adjacent frequency interference suppression demodulation method and device. Background Art

[0002] Cellular passive IoT backscatter (backscatter) communication technology plays a key role in 5G-A and 6G communications. Cellular communication systems have efficient spectrum utilization and wide coverage, which are suitable for the deployment of passive IoT devices. Passive backscatter communication uses the signals of cellular base stations for data transmission, which is suitable for large-scale IoT applications. Therefore, the present invention integrates passive IoT backscatter devices with cellular communication systems to provide more efficient and low-energy IoT communication solutions for smart devices and applications. Cellular passive IoT backscatter communication specifically provides a low-power solution for IoT and sensor networks by adjusting the impedance of the backscatter device to reflect the excitation source signal. At the same time, it can also enhance signal coverage, support large-scale connections, and open up possibilities for emerging application areas such as smart cities and smart manufacturing. As an important enabling technology for future communication systems, backscatter technology is expected to promote the realization of "Internet of Everything" and inject new impetus into the development of 5G-A and 6G networks.

[0003] The cellular passive IoT backscatter (backscatter) communication system is a cellular base station that sends a sinusoidal excitation carrier to the backscatter device. The backscatter device uses part of the received signal to collect energy to maintain its normal operation, and uses the excitation signal to adjust its own impedance for reflection, transmitting the backscatter device information back to the base station. RFID technology is a typical representative technology. This technology can free tags from the limitations of batteries, solve the energy problem of IoT sensors, and play an important role in the future IoT.

[0004] However, backscatter (reverse scatter) communication under broadband adjacent-band interference uses the guard band frequency to transmit data, which will be affected by the adjacent-band interference signal leaked from the broadband signal. In addition, since the backscatter device has no power amplifier and the reflected signal undergoes two round-trip fading, its power is much smaller than the signal power of the adjacent-band interference leaked by the uplink user. The signal strength difference can be as high as 20-30dB, which greatly affects the accurate reception and demodulation of the signal.

[0005] Therefore, there is an urgent need for a cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method that can achieve low bit error rate, high efficiency and anti-interference based on the backscatter communication characteristics. Summary of the invention

[0006] The embodiments of the present invention provide a cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method to solve the problem that the prior art cannot achieve the ideal interference suppression effect when facing complex nonlinear distortion and strong adjacent band interference, resulting in a high bit error rate and low communication quality of the communication system. The technical solution is as follows:

[0007] According to one aspect of the present invention, a cellular passive Internet of Things backscattered adjacent-band interference suppression and demodulation method comprises: acquiring an OFDM signal, performing signal processing and deconvolution on the OFDM signal to obtain a modulated signal and a transmitted signal, and using the modulated signal and the transmitted signal to train a neural network to obtain a reconstruction model; receiving a time-domain OFDM signal, reconstructing the time-domain OFDM signal through the reconstruction model, and obtaining an adjacent-band interference signal generated by leakage of the time-domain OFDM signal through a bandpass filter; performing Fourier transform to eliminate interference on the time-domain OFDM signal and the adjacent-band interference signal to obtain a plurality of interference-eliminated signals; and performing maximum ratio combining on the plurality of interference-eliminated signals and then performing maximum likelihood detection to obtain original information corresponding to the time-domain OFDM signal.

[0008] In one of the embodiments, the OFDM signal is processed and deconvolved to obtain a modulated signal and a transmitted signal respectively by the following steps: removing the cyclic prefix, performing fast Fourier transform and decision detection on the OFDM signal to obtain frequency domain symbols, performing I FFT and adding a cyclic prefix on the frequency domain symbols to obtain a modulated signal; estimating the communication channel to obtain channel parameters, and deconvolving the OFDM signal according to the channel parameters to obtain a transmitted signal.

[0009] In one of the embodiments, the use of the modulated signal and the transmitted signal to train a neural network to obtain a reconstruction model is achieved by the following steps: using the modulated signal as input and the transmitted signal as output to train the neural network, the neural network continuously adjusting internal parameters during the training process; obtaining a loss value based on the difference between the output of the neural network and the OFDM signal, and ending the training when the loss value reaches a set threshold to obtain a reconstruction model.

[0010] In one embodiment, the reconstruction model includes an input layer, 5 fully connected layers, an activation function and an output layer, the fully connected layer includes 30 neurons, each of the fully connected layers is connected to one of the activation functions, and the output layer includes 2 neurons.

[0011] In one of the embodiments, the input layer includes an envelope term of the input signal, higher-order terms of the envelope term, and various-order delay terms.

[0012] In one of the embodiments, performing maximum ratio combining on the multiple interference-eliminated signals and then performing maximum likelihood detection to obtain the original information corresponding to the time domain OFDM signal is achieved by the following steps: performing maximum ratio combining on the multiple interference-eliminated signals to obtain multiple enhanced signals, and performing maximum likelihood detection on the multiple enhanced signals to obtain the original information corresponding to the time domain OFDM signal;

[0013] The calculation formula of the maximum likelihood detection is:

[0014] Wherein, x(l) represents the enhanced signal, y(l) represents the time domain OFDM signal, and L represents the length of the signal.

[0015] In one embodiment, the maximum likelihood detection further comprises:

[0016] According to the maximum likelihood decision criterion The decision maker is calculated:

[0017] Wherein, N represents the number of receiving antennas, H1 is the assumption that b(l)=1, and H0 is the assumption that b(l)=0.

[0018] In one embodiment, the probability distribution of x(l) is

[0019] In one of the embodiments, the method further includes the following steps: if the backscattered signal is deployed on the edge resource blocks on both sides of the spectrum, the reconstruction model is used to reconstruct and eliminate the leakage interference of the single-sided cellular signal; if the backscattered signal is deployed on the middle resource block of the spectrum, the reconstruction model is used to reconstruct and eliminate the leakage interference of the cellular signals on both sides respectively.

[0020] According to one aspect of the present invention, a cellular passive Internet of Things backscattered adjacent-band interference suppression and demodulation device comprises: a neural network training module for acquiring an OFDM signal, performing signal processing and deconvolution on the OFDM signal to obtain a modulated signal and a transmitted signal, and using the modulated signal and the transmitted signal to train a neural network to obtain a reconstruction model; a signal reconstruction module for receiving a time-domain OFDM signal, reconstructing the time-domain OFDM signal through the reconstruction model, and obtaining an adjacent-band interference signal generated by leakage of the time-domain OFDM signal through a bandpass filter; an interference elimination module for performing Fourier transform elimination of interference on the time-domain OFDM signal and the adjacent-band interference signal to obtain a plurality of interference-eliminated signals; and a signal demodulation module for performing maximum likelihood detection after maximum ratio combining on the plurality of interference-eliminated signals to obtain original information corresponding to the time-domain OFDM signal.

[0021] The beneficial effects brought by the technical solution provided by the present invention are:

[0022] In the above technical scheme, the present invention first obtains an OFDM signal, performs signal processing and deconvolution on the OFDM signal to obtain a modulated signal and a transmitted signal, uses the modulated signal and the transmitted signal to train a neural network to obtain a reconstruction model, receives a time-domain OFDM signal, reconstructs the time-domain OFDM signal through the reconstruction model, obtains an adjacent-band interference signal generated by leakage of the time-domain OFDM signal through a bandpass filter, performs Fourier transform on the time-domain OFDM signal and the adjacent-band interference signal to eliminate interference, obtains a plurality of signals after eliminating interference, performs maximum ratio combining on the plurality of signals after eliminating interference, and then performs maximum likelihood detection to obtain the original information corresponding to the time-domain OFDM signal, and constructs a neural network. The OFDM signal is effectively processed through the network reconstruction model, achieving high-precision reconstruction of the signal and accurate identification and elimination of adjacent-band interference. The interference signal is separated by a bandpass filter, and the interference is accurately eliminated in the frequency domain in combination with frequency domain processing technology, which significantly improves the signal quality. The maximum ratio combining and maximum likelihood detection techniques are used to recover the original information from multiple interference-eliminated signals, thereby improving the accuracy and reliability of information transmission and providing strong support for the stable communication of the OFDM system in a high-interference environment. This can effectively solve the problem that the existing technology cannot achieve the ideal interference suppression effect when facing complex nonlinear distortion and strong adjacent-band interference, resulting in a high bit error rate and low communication quality in the communication system. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in describing the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without creative work.

[0024] Figure 1 It is a flow chart of a method for suppressing demodulation of backscattered adjacent-channel interference in a cellular passive Internet of Things according to an exemplary embodiment;

[0025] Figure 2 It is a schematic diagram of the implementation environment of a backscatter communication system application scenario in an application scenario;

[0026] Figure 3 yes Figure 2 Schematic diagram of signal deployment of backscatter communication system in corresponding application scenarios;

[0027] Figure 4 yes Figure 2 A schematic diagram of the structure of the reflection device in the corresponding application scenario;

[0028] Figure 5 yes Figure 2 Flow chart of the demodulation method for suppressing backscattered adjacent-channel interference in cellular passive IoT in the corresponding application scenario;

[0029] Figure 6 yes Figure 2 Schematic diagram of the demodulation method for suppressing backscattered adjacent-channel interference in cellular passive IoT in the corresponding application scenario;

[0030] Figure 7 yes Figure 2 Schematic diagram of the structure of the neural network in the corresponding application scenario;

[0031] Figure 8 yes Figure 2 A comparison diagram of OFDM signals in the embodiment and the prior art in the corresponding application scenario;

[0032] Fig. 9 yes Figure 2 The effect diagram of eliminating interference signals in the frequency domain in the corresponding application scenario;

[0033] Fig.10 yes Figure 2 Comparison of bit error rates of backscatter communication systems under adjacent-channel interference in corresponding application scenarios;

[0034] Fig.11 It is a block diagram of a cellular passive Internet of Things backscatter adjacent channel interference suppression demodulation device according to an exemplary embodiment. DETAILED DESCRIPTION

[0035] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0036] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present disclosure refers to the presence of the features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0037] The present invention provides a cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method, which effectively processes OFDM signals by constructing a neural network reconstruction model, realizes high-precision reconstruction of signals and accurate identification and elimination of adjacent band interference, solves the problem that the prior art cannot achieve an ideal interference suppression effect, resulting in a high bit error rate and low communication quality of the communication system. The cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method is suitable for a cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation device, and the cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation device can be an electronic device. The cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method in the embodiment of the present invention can be applied to a variety of scenarios, such as cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation, etc.

[0038] See also Figure 1 , an embodiment of the present invention provides a cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method, which is applicable to electronic equipment.

[0039] In the following method embodiments, for ease of description, the execution subject of each step of the method is taken as an electronic device as an example for illustration, but this does not constitute a specific limitation.

[0040] like Figure 1 As shown, the method may include the following steps:

[0041] Step 110, acquiring an OFDM signal, performing signal processing and deconvolution on the OFDM signal to obtain a modulated signal and a transmitted signal, and using the modulated signal and the transmitted signal to train a neural network to obtain a reconstruction model.

[0042] In one possible implementation, the OFDM signal is subjected to cyclic prefix removal, fast Fourier transform and decision detection to obtain frequency domain symbols, the frequency domain symbols are subjected to IFFT and cyclic prefix addition to obtain a modulated signal, the communication channel is estimated to obtain channel parameters, and the OFDM signal is deconvolved according to the channel parameters to obtain a transmitted signal.

[0043] In one possible implementation, the modulated signal is used as input and the transmitted signal is used as output to train the neural network. During the training process, the neural network continuously adjusts its internal parameters, and a loss value is obtained based on the difference between the output of the neural network and the OFDM signal. When the loss value reaches a set threshold, the training ends and a reconstructed model is obtained.

[0044] In one possible implementation, the reconstruction model includes an input layer, 5 fully connected layers, an activation function and an output layer, the fully connected layer includes 30 neurons, each fully connected layer is connected to an activation function, and the output layer includes 2 neurons.

[0045] In a possible implementation, the input layer includes an envelope term of the input signal, higher-order terms of the envelope term, and various-order delay terms.

[0046] Step 130 , receiving a time domain OFDM signal, reconstructing the time domain OFDM signal through a reconstruction model, and obtaining an adjacent band interference signal generated by leakage of the time domain OFDM signal through a bandpass filter.

[0047] Step 150: Perform Fourier transform on the time domain OFDM signal and the adjacent band interference signal to eliminate interference, and obtain multiple interference-eliminated signals.

[0048] Step 170: Perform maximum ratio combining on the multiple interference-eliminated signals and then perform maximum likelihood detection to obtain original information corresponding to the time-domain OFDM signal.

[0049] In a possible implementation, multiple interference-eliminated signals are combined by maximum ratio combining to obtain multiple enhanced signals, and maximum likelihood detection is performed on the multiple enhanced signals to obtain original information corresponding to the time-domain OFDM signal.

[0050] Specifically, the calculation formula of maximum likelihood detection is:

[0051]

[0052] Wherein, x(l) represents the enhanced signal, y(l) represents the time domain OFDM signal, and L represents the length of the signal.

[0053] In a possible implementation, the maximum likelihood detection further includes:

[0054] According to the maximum likelihood decision criterion The decision maker is calculated:

[0055]

[0056] Wherein, N represents the number of receiving antennas, H1 is the assumption that b(l)=1, and H0 is the assumption that b(l)=0.

[0057] In one possible implementation, the probability distribution of x(l) is

[0058] In one possible implementation, if the backscattered signal is deployed on the edge resource blocks on both sides of the spectrum, the reconstruction model is used to reconstruct and eliminate the leakage interference of the cellular signal on one side; if the backscattered signal is deployed on the middle resource block of the spectrum, the reconstruction model is used to reconstruct and eliminate the leakage interference of the cellular signals on both sides.

[0059] Through the above process, the present invention first obtains an OFDM signal, performs signal processing and deconvolution on the OFDM signal respectively to obtain a modulated signal and a transmitted signal, uses the modulated signal and the transmitted signal to train a neural network to obtain a reconstruction model, receives a time-domain OFDM signal, reconstructs the time-domain OFDM signal through the reconstruction model, obtains an adjacent-band interference signal generated by leakage of the time-domain OFDM signal through a bandpass filter, performs Fourier transform on the time-domain OFDM signal and the adjacent-band interference signal to eliminate interference respectively, obtains a plurality of signals after eliminating interference, performs maximum ratio combining on the plurality of signals after eliminating interference, and then performs maximum likelihood detection to obtain original information corresponding to the time-domain OFDM signal, and constructs a neural network. The network reconstruction model effectively processes OFDM signals, realizes high-precision reconstruction of signals and accurate identification and elimination of adjacent-band interference. It separates interference signals through bandpass filters and accurately eliminates interference in the frequency domain in combination with frequency domain processing technology, significantly improving signal quality. It uses maximum ratio combining and maximum likelihood detection technology to recover the original information from multiple interference-eliminated signals, thereby improving the accuracy and reliability of information transmission and providing strong support for the stable communication of OFDM systems in high-interference environments. It can effectively solve the problem that the existing technology cannot achieve the ideal interference suppression effect when facing complex nonlinear distortion and strong adjacent-band interference, resulting in high bit error rate and low communication quality of the communication system.

[0060] In one application scenario, Figure 2 A schematic diagram of the implementation environment of a backscatter communication system application scenario is shown. Figure 3 The signal deployment diagram of the backscatter communication system in this application scenario is shown. Figure 4 The schematic diagram of the reflection device is shown. Figure 5 The flowchart of the demodulation method for suppressing adjacent-channel interference in cellular passive IoT backscattering is shown. Figure 6 A schematic diagram showing the demodulation method for suppressing adjacent-channel interference with backscattering in cellular passive IoT is shown. Figure 7 The schematic diagram of the neural network structure is shown. Figure 8 A comparison diagram of OFDM signals in the embodiment of the present invention and the prior art is shown. Fig. 9 The effect diagram of eliminating interference signals in the frequency domain is shown. Fig.10 A comparison chart of the bit error rates of backscatter communication systems under adjacent channel interference is shown.

[0061] like Figure 2 As shown, the backscatter communication system includes an access point (such as an RFID system reader or an access point of other types of backscatter communication systems) and a reflection device, and takes into account the influence of out-of-band interference generated by broadband cellular signals of adjacent frequency users.

[0062] like Figure 3As shown, the backscattered uplink signal occupies resource block n, and the uplink cellular signal is transmitted in the adjacent frequency band. The out-of-band interference signal generated by the nonlinear distortion of the power amplifier will seriously affect the accurate reception of the backscattered signal.

[0063] like Figure 4 As shown, each reflection device includes: a backscatter antenna module: used to receive and reflect signals from environmental access points; a backscatter modulation module: changing the load impedance of the antenna according to the information symbol to achieve backscatter modulation; a microcontroller module: used to control the communication process of the reflection device; a signal processor module: used for the reflection device to perform basic signal processing, such as decoding of control signals, etc.; a RF energy collector and battery module: used to collect energy from the incident signal and charge the battery to power all modules; other modules, including storage, sensing, clock and other units.

[0064] The reflection device modulates the incident signal it receives by switching the load impedance to change the amplitude and / or phase of its backscattered signal, and the backscattered signal is received and finally decoded by the full-duplex access point.

[0065] like Figure 5 As shown, the cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method may include the following steps:

[0066] Step S1: The base station sends a downlink excitation signal.

[0067] Step S2: The reflective device collects energy and backscatters the received signal according to the information it carries.

[0068] Step S3: The base station receives the broadband cellular signal and generates training data to train the deep neural network.

[0069] Step S4: the base station receives and generates out-of-band interference generated by the broadband cellular signal of the adjacent frequency user, and eliminates the interference in the frequency domain.

[0070] Step S5, performing maximum ratio combining and maximum likelihood detection on the signal obtained by eliminating interference in sequence to achieve signal demodulation.

[0071] The signal in step S1 is in the form of a pure carrier signal deployed on resource block n. The power reflection coefficient of the reflection device in step S2 is configured as a fixed constant known to the access point. The reflection device absorbs the signal transmitted by the access point and obtains energy therefrom. The reflection device reflects the incident signal back to the access point with different amplitudes and phases by continuously changing the antenna impedance, thereby modulating the incident signal and achieving the purpose of communicating with the access point.

[0072] In addition, since the base station downlink excitation signal is deployed on resource block n, the backscattering device transmits the reflected signal on the frequency band of these resource blocks. However, the spectrum leakage of adjacent-band broadband cellular signals due to the nonlinear distortion of the power amplifier will cause strong adjacent-band interference signals to enter these resource blocks. The difference in strength between the leaked broadband interference signal and the reflected signal can be as high as 20-30dB, which will seriously affect the demodulation performance.

[0073] In step S3, the reflection device uses the broadband cellular signal generated by demodulation and remodulation at the receiving end before and after being affected by the nonlinear distortion of the power amplifier to train a deep neural network, fit the nonlinear behavior of the power amplifier at the transmitting end, and use it for the subsequent interference reconstruction method.

[0074] In step S4, the base station generates a time-domain broadband cellular signal after judgment detection of the received signal, and uses the neural network trained in step S3 to reconstruct the time-domain cellular signal sent by the user end, and further obtains the leaked adjacent-band interference signal through filtering. The time-domain signal received by the antenna and the reconstructed adjacent-band interference signal are Fourier transformed respectively to eliminate interference in the frequency domain.

[0075] It is worth mentioning that in step S2, the backscattered uplink signal occupies resource block n, and its adjacent resource blocks all transmit uplink broadband cellular signals.

[0076] OFDM signals are composed of the superposition of multiple orthogonal subcarriers, each of which carries a portion of data. When the phases of multiple subcarriers are the same, they can form a large peak, which makes the OFDM signal have a high peak-to-average power ratio (PAPR). High PAPR means that the peak power of the signal is much higher than the average power, which makes the amplifier easily enter the nonlinear working area in the power amplifier, especially when the power amplifier tries to amplify these high peaks. The nonlinear effect not only affects the original frequency components of the signal, but also generates new harmonics and intermodulation products in the spectrum. These new frequency components may appear outside the original OFDM band, resulting in out-of-band emissions. This out-of-band emission can interfere with adjacent channels, especially in a crowded spectrum environment, which will have a significant impact on other communication systems.

[0077] In addition, since the backscattering device has no power amplifier and the reflected signal undergoes two round-trip fading, its power is much smaller than the signal power of the adjacent band interference leaked by the uplink user, so an efficient interference elimination algorithm is required at the receiving end.

[0078] like Figure 6As shown in the figure, during the training stage of the neural network, on the one hand, the base station sequentially removes the cyclic prefix, performs FFT and decision detection on the OFDM signal it receives to obtain the originally transmitted frequency domain symbols, and then obtains the originally transmitted time domain OFDM modulated signal as the input of the neural network through operations such as I FFT and adding cyclic prefix; on the other hand, the base station deconvolves the received OFDM signal with the estimated channel to obtain the time domain OFDM signal sent by the user-end antenna as the output of the neural network to remove interference from other factors and ensure accurate fitting of the nonlinear distortion of the power amplifier.

[0079] Furthermore, in step S4, when the backscatter signal is deployed in the resource blocks on both sides, it is necessary to use a neural network to reconstruct and eliminate the interference of the unilateral cellular signal leakage. When the backscatter signal is deployed in the middle resource block, it is only necessary to reconstruct and eliminate the interference of the cellular signal leakage on both sides.

[0080] The resource block occupancy is set as follows: the backscatter communication uplink occupies RB0 and RB1 resource blocks, and the corresponding baseband frequency range is f0~f1; the OFDM signal occupies RB2~RB N Resource blocks, the center frequency of RB0 and RB1 resource blocks is f I =(f0+f1) / 2, so it is necessary to shift the OFDM signal after the PA fitted by the neural network to the left in the frequency domain by f I , and then design the cut-off frequency to be f c =(f1-f0) / 2 to obtain the equivalent baseband signal of its leakage interference, and then move it to the f0~f1 frequency band in the frequency domain to obtain the interference y of the OFDM signal leaking in the backscatter link frequency band. I (l), namely:

[0081]

[0082] Among them, y I [l] is the out-of-band interference of the OFDM signal after the nonlinear PA reconstructed by the receiving end. The interference simulated by the receiving end after passing through the channel can be expressed as:

[0083]

[0084] in, The channel response of the uplink cellular signal estimated by the receiving end.

[0085] The signal received by the receiving end of the backscatter communication system can be expressed as: y(l) = y u (l)+y i (l)+z(l),y u (l) represents the useful signal containing the modulation information of the reflecting device, y i(l) is the out-of-band interference leaked from the adjacent band cellular user, z(l) is the power Additive Gaussian noise, The receiving end first uses the results of the neural network fitting Eliminate out-of-band interference signals in the frequency domain, and then perform operations such as channel estimation and equalization, decoding and decision making.

[0086] Finally, a simulation experiment is used to verify the interference reconstruction and elimination effect of the present invention. Without loss of generality, the simulation verification here uses a memory polynomial baseband model to simulate the nonlinear effect of the power amplifier (PA) on the OFDM signal:

[0087]

[0088] like Figure 7 As shown, in order to improve the modeling accuracy of the power amplifier nonlinearity, the network model used in the simulation here is an enhanced real-valued time-delay neural network, which has 20 feature inputs and 2 feature outputs. The model uses two orthogonal components of the input signal I in and Q in To predict the two components I corresponding to the output signal out and Q out In addition to the real and imaginary parts of the signal, the network input also increases the envelope term of the input signal and the higher-order terms of the envelope term to improve the nonlinear processing capability of the model, and increases the delay terms of each order to simulate the memory effect of the power amplifier.

[0089] The internal structure of the network contains five fully connected layers, each containing 30 neurons, and each linear layer is followed by a Leaky ReLU activation layer. The characteristic of Leaky ReLU is that it allows small negative gradients to pass through, that is, when the input value is negative, the output is a small proportion of the negative input (here 0.01). This can avoid the "dead neuron" problem of the ReLU activation function, that is, the problem that some neurons only output zero values ​​during training.

[0090] like Figure 8As shown in the figure, a comparison of OFDM signals before and after the power amplifier and a performance graph of deep neural network modeling are shown. The black line represents the power spectral density of the OFDM signal that is not affected by the nonlinearity of the power amplifier. At around 10MHz, the signal power of its sidelobe leakage will rapidly decay from -80dBm to -130 to -140dBm. The red line represents the power spectral density of the OFDM signal after PA nonlinear distortion. At around 10MHz, the signal power of its sidelobe leakage will only decay from -80dBm to -90 to -100dBm, which shows that it has produced serious out-of-band interference. The blue line represents the nonlinear distortion effect fitted by neural network modeling. It can be seen that the two have a high degree of overlap, indicating that the neural network can fit the out-of-band distortion part very well.

[0091] Define the normalized mean square error (NMSE) calculation formula:

[0092]

[0093] Among them, y i represents the actual observed value, Represents the predicted value. NMSE is used to measure the error of the model prediction value relative to the actual observed value. Its standardization makes the result independent of the scale of the data and is suitable for comparison of different data sets. The calculated NMSE between the OFDM distortion signal fitted by the neural network and the actual value is -40dB, which shows that the deep neural network can well fit the influence of PA nonlinear distortion.

[0094] like Fig. 9 As shown in the figure, the effect of eliminating interference signals in the frequency domain at the receiving end of the backscattering system is shown, where the black cross solid line is the power spectrum of the signal received at the receiving end, which includes adjacent-band interference leaked from the OFDM signal output by the nonlinear power amplifier, label backscattered signals and noise; the black asterisk solid line is the power spectrum of the signal after the adjacent-band interference leaked from the OFDM signal output by the nonlinear power amplifier reconstructed by the neural network and the adjacent-band interference is eliminated in the frequency domain; the black circle solid line is the power spectrum of the background noise.

[0095] Depend on Fig. 9 It can be seen intuitively that the received signal is affected by the leakage interference of adjacent band users, and the peak power spectral density of the received signal is about 45dBm / Hz higher than the noise floor. After the neural network reconstructs the interference leakage of adjacent frequency band users and eliminates the interference in the frequency domain, the peak power spectral density of the received signal can be reduced by about 45dBm / Hz. At this time, the received signal after eliminating the interference is on par with the noise floor power, which means that the interference is almost completely eliminated.

[0096] Furthermore, the interference rejection combining (I RC) technology is used as a comparison method to compare with the embodiment of the present invention. The core of the I RC technology is to effectively distinguish and suppress interference signals by using the multi-antenna system of the receiver while retaining useful signal components. It is achieved by applying specific weight combinations to the received signal vectors, and these weights are intended to maximize the desired signal to interference and noise ratio (SI NR).

[0097] Assuming the received signal is y(n), it can be expressed as:

[0098] y(n)=Hs(n)+H i s i (n)+n,

[0099] According to the principle of IRC algorithm, the transmitted signal using maximum likelihood estimation is:

[0100]

[0101] in, is the target signal estimate, H is the channel response matrix, Q is the interference noise covariance matrix, and H H It represents the deconjugated transpose of the matrix. For the traditional I RC algorithm, to achieve the effect of interference suppression, the estimation of the covariance matrix of the interference noise and the channel response is the key.

[0102] like Fig.10 As shown, a comparison chart of the bit error rate of the backscatter communication system under adjacent channel interference is shown, wherein the black solid line is the reference bit error rate performance without interference and with known channel response, the blue solid line is the bit error rate performance of the IRC method, and the red solid line is the bit error rate performance of the neural network interference reconstruction and elimination method of the present invention.

[0103] Depend on Fig. 9 It can be seen that under given conditions, the bit error rate performance of the embodiment of the present invention is significantly better than that of the I RC method, and at a bit error rate of , the performance of the neural network method is improved by about 6dB compared with the I RC method, and the performance is degraded by about 2.5dB compared with the ideal situation of known channels and no interference.

[0104] Through the above process, the embodiment of the present invention realizes efficient processing of adjacent channel interference in the cellular passive Internet of Things backscatter communication system through the broadband cellular signal adjacent channel interference reconstruction and elimination technology based on deep neural network, which can significantly reduce the impact of adjacent channel interference on the backscatter signal and improve the signal reception quality. Through the precise reconstruction and elimination of adjacent channel interference, the method significantly reduces the demodulation bit error rate of the backscatter information. Compared with the traditional interference suppression combining (IRC) technology, it has better performance at the same bit error rate, indicating that it can effectively improve the accuracy of signal demodulation.

[0105] In another application scenario, the backscatter communication system includes a full-duplex base station and a reflection device. The full-duplex base station is equipped with N+1 receiving antennas, one of which is used to send single-tone excitation signals and receive backscatter signals deployed in the protection band of the cellular system. The backscatter device switches different impedances according to the information bits to achieve different backscatter coefficients for backscattering. The N antennas of the full-duplex base station are used for cellular system communication. However, the backscatter communication detection deployed in the protection band is easily affected by the interference of the adjacent frequency leakage of the uplink broadband signal. The present invention designs an adjacent band interference reconstruction and elimination method based on deep learning, and proposes a corresponding signal model and signal detection algorithm to achieve low bit error rate backscatter information demodulation.

[0106] Specifically, the backscatter communication reflected signal is deployed on the resource block n (RB n ), since the base station will not only receive the reflected signal y u , and will also receive strong interference signals y from the uplink broadband signal leaking into the guard band frequency band i , assuming that the lth sampling point of the base station sending the single tone signal is s(l), and the lth sampling point of the reflected device's own signal is

[0107] Where p is the transmit power, b(l) is the Manchester coded BPSK modulation symbol, and f c is the frequency at which the impedance of the reflective device switches, T s is the sampling frequency. The base station not only receives the reflected signal y u , and receives the adjacent band leakage strong interference signal y of the uplink cellular terminal broadband signal i , so the baseband signal received by the base station is:

[0108]

[0109] Among them, g b The downlink channel for the base station to send a single tone signal to the backscatter device, g f The uplink channel between the reflection device and the base station receiver, is the phase noise, ⊙ is the Hadamard product, y u (l) is the effective signal received from the reflection device, y i (l) is adjacent band interference, For background noise.

[0110] Among them, the backscatter device is equipped with an antenna for signal reflection. The base station sends a downlink single-tone signal. The reflection device selects different backscatter coefficients for backscattering according to the information bit. The base station reconstructs and eliminates interference and detects the signal of the reflection device.

[0111] Specifically, the base station first sends a downlink excitation signal in the form of a pure carrier signal deployed on resource block n. Then the reflection device collects energy and adjusts the impedance according to the information it carries to achieve backscattering of the received signal. The power reflection coefficient is a fixed constant. The device receives the excitation signal transmitted by the base station and obtains energy from it. By changing the antenna impedance, the device reflects the signal back to the access point with different amplitudes and phases to achieve signal modulation and communication with the base station. Since the base station downlink excitation signal is deployed on resource block 0 and resource block 1, the backscattering device reflected signal is also located in the frequency band of the resource block. In addition, the adjacent broadband cellular signal will generate leaked adjacent band interference, and the nonlinear distortion of the power amplifier will significantly increase the power of the leaked adjacent band interference signal.

[0112] Furthermore, in order to reconstruct the leaked adjacent band interference signal, the neural network needs to be trained first to reconstruct and eliminate the interference using a neural network at the base station. During the training phase, on the one hand, the base station sequentially removes the cyclic prefix, performs FFT and decision detection on the OFDM signal it receives to obtain the originally transmitted frequency domain symbols, and then obtains the originally transmitted time domain OFDM modulated signal through I FFT and adding a cyclic prefix as the input of the neural network; on the other hand, the base station deconvolves the received OFDM signal with the estimated channel to obtain the time domain OFDM signal sent by the user-end antenna as the output of the neural network.

[0113] Furthermore, in the actual deployment stage, since the nonlinear behavior of the power amplifier is mainly determined by its physical and electrical characteristics, such as the characteristics of the transistor, the design of the amplifier, etc. These characteristics usually do not change significantly in a short period of time, so this trained model can be used to model the actual received signal in subsequent communications. The base station generates a time domain OFDM signal after judgment detection of the received signal, and uses the generated neural network to reconstruct the time domain OFDM signal sent by the user end. The adjacent band interference signal generated by the leakage of the time domain OFDM signal is obtained through a bandpass filter, and the time domain signal received by the antenna and the reconstructed adjacent band interference signal are Fourier transformed respectively to eliminate interference in the frequency domain;

[0114] Furthermore, the signal obtained by eliminating interference is subjected to maximum ratio combining and maximum likelihood detection in sequence to achieve signal demodulation.

[0115] Specifically, in maximum likelihood detection, the background noise follows a complex Gaussian distribution,

[0116] Maximum Likelihood Decision Criterion The decision maker is calculated:

[0117] in, Assume that b(l)=1. Assuming b(l) = 0, the probability distribution of x(l) is:

[0118]

[0119] Through the above process, the present invention effectively processes OFDM signals by constructing a neural network reconstruction model, realizes high-precision reconstruction of signals and accurate identification and elimination of adjacent-band interference, separates interference signals through a bandpass filter, and accurately eliminates interference in the frequency domain in combination with frequency domain processing technology, significantly improving signal quality, and utilizes maximum ratio combining and maximum likelihood detection technology to recover original information from multiple interference-eliminated signals, thereby improving the accuracy and reliability of information transmission, and providing strong support for the stable communication of OFDM systems in high-interference environments, thereby effectively solving the problem that the prior art cannot achieve an ideal interference suppression effect when facing complex nonlinear distortion and strong adjacent-band interference, resulting in a high bit error rate and low communication quality in the communication system.

[0120] The following is an embodiment of the device of the present invention, which can be used to perform the cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method involved in the present invention. For details not disclosed in the embodiment of the device of the present invention, please refer to the method embodiment of the cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method involved in the present invention.

[0121] See also Fig.11 In an embodiment of the present invention, a cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation device 800 is provided.

[0122] The cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation device 800 includes but is not limited to: a neural network training module 810, a signal reconstruction module 830, an interference elimination module 850 and a signal demodulation module 870.

[0123] Among them, the neural network training module 810 is used to obtain the OFDM signal, perform signal processing and deconvolution on the OFDM signal to obtain a modulated signal and a transmitted signal, and use the modulated signal and the transmitted signal to train the neural network to obtain a reconstruction model.

[0124] The signal reconstruction module 830 is used to receive a time-domain OFDM signal, reconstruct the time-domain OFDM signal through a reconstruction model, and obtain an adjacent-band interference signal generated by leakage of the time-domain OFDM signal through a bandpass filter.

[0125] The interference elimination module 850 is used to perform Fourier transform on the time domain OFDM signal and the adjacent band interference signal to eliminate interference, so as to obtain multiple interference-eliminated signals.

[0126] The signal demodulation module 870 is used to perform maximum likelihood detection after maximum ratio combining of multiple interference-eliminated signals to obtain original information corresponding to the time-domain OFDM signal.

[0127] It should be noted that, in the cellular passive Internet of Things backscatter adjacent channel interference suppression demodulation provided in the above-mentioned embodiment, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the cellular passive Internet of Things backscatter adjacent channel interference suppression demodulation device will be divided into different functional modules to complete all or part of the functions described above.

[0128] In addition, the cellular passive Internet of Things backscatter adjacent channel interference suppression demodulation device provided in the above-mentioned embodiments and the cellular passive Internet of Things backscatter adjacent channel interference suppression demodulation method embodiments belong to the same concept, and the specific manner in which each module performs the operation has been described in detail in the method embodiments and will not be repeated here.

[0129] Compared with the related art, the present invention has the following beneficial effects:

[0130] 1. The present invention first obtains an OFDM signal, performs signal processing and deconvolution on the OFDM signal to obtain a modulated signal and a transmitted signal, uses the modulated signal and the transmitted signal to train a neural network to obtain a reconstruction model, receives a time-domain OFDM signal, reconstructs the time-domain OFDM signal through the reconstruction model, obtains an adjacent-band interference signal generated by leakage of the time-domain OFDM signal through a bandpass filter, performs Fourier transformation to eliminate interference on the time-domain OFDM signal and the adjacent-band interference signal, obtains a plurality of signals after eliminating interference, performs maximum ratio combining on the plurality of signals after eliminating interference, and then performs maximum likelihood detection to obtain original information corresponding to the time-domain OFDM signal, and constructs a neural network The reconstruction model effectively processes OFDM signals, realizes high-precision reconstruction of signals and accurate identification and elimination of adjacent-band interference. It separates interference signals through bandpass filters and accurately eliminates interference in the frequency domain in combination with frequency domain processing technology, significantly improving signal quality. It uses maximum ratio combining and maximum likelihood detection technology to recover the original information from multiple interference-eliminated signals, thereby improving the accuracy and reliability of information transmission and providing strong support for the stable communication of OFDM systems in high-interference environments. It can effectively solve the problem that the existing technology cannot achieve the ideal interference suppression effect when facing complex nonlinear distortion and strong adjacent-band interference, resulting in high bit error rate and low communication quality of the communication system.

[0131] 2. The present invention is highly adaptable: the nonlinear distortion of power amplifiers is modeled using neural networks, which can cope with power amplifiers with different physical and electrical characteristics. Since these characteristics usually do not change significantly in a short period of time, the trained neural network model can be effectively applied to actual communications for a long time.

[0132] 3. The present invention can improve communication efficiency: By optimizing the signal processing flow, including operations such as removing cyclic prefix, FFT, decision detection, I FFT, adding cyclic prefix, etc., the method can improve communication efficiency while ensuring communication quality. This helps to achieve more efficient data transmission in IoT devices and sensor networks.

[0133] 4. The present invention supports large-scale connections: Backscatter technology, as an important enabling technology for future communication systems, has the characteristics of low power consumption, enhanced signal coverage and support for large-scale connections. This demodulation method further improves the performance of the backscatter communication system, helps promote the realization of "Internet of Everything", and injects new impetus into the development of 5G-A and 6G networks.

[0134] 5. The present invention can improve signal coverage and reliability: the backscatter device adjusts its own impedance to reflect the signal, and combined with the signal processing algorithm of the base station, it can enhance the signal coverage and improve the reliability of signal transmission. This is of great significance for the application of IoT devices and sensor networks in complex environments.

[0135] 6. The present invention can reduce energy consumption: Backscattering technology communicates by adjusting impedance to reflect the excitation source signal, which has lower energy consumption than traditional communication methods. This is of great significance for the long-term operation and energy management of IoT devices.

[0136] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0137] The above descriptions are only some embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method, characterized in that: The method comprises: Acquire an OFDM signal, perform signal processing and deconvolution on the OFDM signal to obtain a modulated signal and a transmitted signal, and use the modulated signal and the transmitted signal to train a neural network to obtain a reconstruction model; Receive a time domain OFDM signal, reconstruct the time domain OFDM signal through the reconstruction model, and obtain an adjacent band interference signal generated by leakage of the time domain OFDM signal through a bandpass filter; Performing Fourier transformation on the time-domain OFDM signal and the adjacent-band interference signal to eliminate interference, respectively, to obtain a plurality of interference-eliminated signals; The multiple interference-eliminated signals are subjected to maximum ratio combining and then maximum likelihood detection to obtain original information corresponding to the time-domain OFDM signal.

2. The cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method according to claim 1, characterized in that: The performing signal processing and deconvolution on the OFDM signal to obtain a modulated signal and a transmitted signal respectively includes: Performing cyclic prefix removal, fast Fourier transform and decision detection on the OFDM signal to obtain frequency domain symbols, and performing I FFT and adding cyclic prefix on the frequency domain symbols to obtain a modulated signal; The communication channel is estimated to obtain channel parameters, and the OFDM signal is deconvolved according to the channel parameters to obtain a transmission signal.

3. The cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method according to claim 1, characterized in that: The method of using the modulated signal and the transmitted signal to train a neural network to obtain a reconstruction model includes: The modulation signal is used as input and the transmission signal is used as output to train a neural network, and the neural network continuously adjusts internal parameters during the training process; A loss value is obtained according to the difference between the output of the neural network and the OFDM signal. When the loss value reaches a set threshold, the training ends and a reconstruction model is obtained.

4. The cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method according to claim 1, characterized in that: The reconstruction model includes an input layer, 5 fully connected layers, an activation function and an output layer, the fully connected layer includes 30 neurons, each of the fully connected layers is connected to one of the activation functions, and the output layer includes 2 neurons.

5. The cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method as claimed in claim 4, characterized in that: The input layer includes the envelope term of the input signal, the higher-order terms of the envelope term and various-order delay terms.

6. The cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method according to claim 1, characterized in that: The step of performing maximum likelihood detection on the multiple interference-eliminated signals and then performing maximum ratio combining to obtain original information corresponding to the time domain OFDM signal includes: Performing maximum ratio combining on the multiple interference-eliminated signals to obtain multiple enhanced signals, and performing maximum likelihood detection on the multiple enhanced signals to obtain original information corresponding to the time-domain OFDM signal; The calculation formula of the maximum likelihood detection is: Wherein, x(l) represents the enhanced signal, y(l) represents the time domain OFDM signal, and L represents the length of the signal.

7. The cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method as claimed in claim 6, characterized in that: The maximum likelihood detection also includes: According to the maximum likelihood decision criterion The decision maker is calculated: Wherein, N represents the number of receiving antennas, H1 is the assumption that b(l)=1, and H0 is the assumption that b(l)=0.

8. The cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method as claimed in claim 6, characterized in that: The probability distribution of x(l) is 9. The cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation method according to claim 1, characterized in that: The method further comprises: If the backscattered signal is deployed on the edge resource blocks on both sides of the spectrum, the reconstruction model is used to reconstruct and eliminate the leakage interference of the cellular signal on one side; If the backscattered signal is deployed on the middle resource block of the spectrum, the reconstruction model is used to reconstruct and eliminate the leakage interference of the cellular signals on both sides.

10. A cellular passive Internet of Things backscatter adjacent frequency interference suppression demodulation device, characterized in that: The device comprises: A neural network training module is used to obtain an OFDM signal, perform signal processing and deconvolution on the OFDM signal to obtain a modulated signal and a transmitted signal, and use the modulated signal and the transmitted signal to train a neural network to obtain a reconstruction model; A signal reconstruction module, used for receiving a time domain OFDM signal, reconstructing the time domain OFDM signal through the reconstruction model, and obtaining an adjacent band interference signal generated by leakage of the time domain OFDM signal through a bandpass filter; An interference elimination module, used for performing Fourier transform on the time domain OFDM signal and the adjacent band interference signal to eliminate interference, so as to obtain a plurality of interference-eliminated signals; The signal demodulation module is used to perform maximum likelihood detection after maximum ratio combining of the multiple interference-eliminated signals to obtain original information corresponding to the time domain OFDM signal.

Images