Carrier frequency offset estimation method and system based on carrier chip

Abstract

The present application relates to the field of wireless communication technology, and specifically to a carrier frequency deviation estimation method and system based on a carrier chip. The method includes: obtaining a standard carrier frequency, a communication modulation signal, and a standard modulation signal; obtaining a noise signal according to the signal difference, and obtaining a noise stability weight according to the fluctuation of the noise signal in the frequency domain; obtaining a frequency difference according to the difference between the frequency position and the harmonic order and the standard carrier frequency, obtaining a frequency weight according to the frequency difference, and obtaining a noise interference weight based on this and the noise stability weight; converting the noise signal into the frequency domain, and obtaining the noise interference intensity according to the ratio of the energy amplitude and the noise interference weight; using the noise interference intensity and signal strength at the current moment as input, and obtaining the carrier frequency deviation estimation at the current moment through neural network training. The present application can obtain a more accurate carrier frequency deviation estimation, which can effectively improve the communication quality of the carrier chip.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a carrier frequency offset estimation method and system based on a carrier chip. Background Art

[0002] A carrier chip is a communication chip that can mix baseband signals and carrier signals to generate modulated signals. When the carrier chip communicates, the modulated signal is sent by the carrier chip at the signal sending end, and the modulated signal is received by the carrier chip at the signal receiving end. When modulating a signal, the carrier frequency of the signal needs to be determined at the signal receiving end. Only when the carrier frequency is accurate can a good demodulation effect be obtained.

[0003] However, in the actual transmission process, the modulated signal is easily affected by noise and carrier frequency deviation occurs. For example, in the PLC (power line carrier modulation) communication system using power lines as channels, power lines are used as channels for communication. However, there are a series of factors that cause carrier frequency deviation in the power lines, such as load impedance, thermal noise, and multipath effects. These factors can easily cause frequency deviation of the modulated signal, which in turn causes communication errors. The frequency domain information collected by the FFT algorithm is discrete data information and cannot reflect the continuous signal frequency domain characteristics. Therefore, there is a fence effect, which will lead to inaccurate determination of carrier frequency deviation and cause communication errors. Therefore, in power line carrier communication, there is a carrier frequency deviation phenomenon caused by channel interference. The traditional frequency deviation estimation algorithm based on the FFT algorithm has a fence effect when solving the problem, which is prone to communication errors. Summary of the invention

[0004] In order to solve the technical problem of the fence effect, the present application provides a carrier frequency offset estimation method and system based on a carrier chip, and the technical solutions adopted are as follows:

[0005] In a first aspect, the present application proposes a carrier frequency offset estimation method based on a carrier chip, the method comprising the following steps:

[0006] Preset a standard carrier frequency, obtain a communication modulation signal, take the communication modulation signal and the standard carrier frequency as input, and obtain the standard modulation signal through signal demodulation and signal modulation;

[0007] The noise signal is obtained according to the difference between the communication modulation signal and the standard modulation signal, and the noise signal is evenly divided and then converted into the frequency domain to obtain the noise spectrum vector; the noise stability weight of each frequency position is obtained according to the fluctuation characteristics of the energy amplitude at the same frequency position of all noise spectrum vectors;

[0008] The energy attenuation multiple of the harmonic is obtained according to the preset harmonic attenuation amplitude; the harmonic is obtained based on the frequency position and the order of the harmonic, and the frequency difference of each order of the harmonic is obtained according to the difference between the harmonic and the standard carrier frequency; the frequency weight of each order of the harmonic at each frequency position is obtained according to the positive and negative of the frequency difference; the noise interference weight at each frequency position is obtained according to the frequency weight and energy attenuation multiple of all orders of harmonics at the same frequency position, combined with the noise stabilization weight at each frequency position;

[0009] The noise signal and the standard modulation signal at each moment are converted into the frequency domain to obtain the frequency domain noise vector and the frequency domain signal vector. The energy amplitude at the standard carrier frequency position of the frequency domain signal vector is the energy amplitude of the carrier. The noise interference intensity at each moment is obtained according to the ratio of the energy amplitude at different frequency positions in the frequency domain noise vector to the carrier energy amplitude combined with the noise interference weight at each frequency position;

[0010] The current noise interference intensity and signal strength are used as input, and the carrier frequency offset estimation at the current moment is obtained through neural network training.

[0011] In the above scheme, the communication modulation signal is obtained through the signal receiving module, and the communication modulation signal within the past 1 millisecond is obtained to estimate the carrier frequency deviation at the current moment; the noise signal is analyzed to obtain the frequency deviation interference intensity received by the carrier within the past 1 millisecond, and the carrier frequency deviation estimation is completed based on this. By extracting the characteristics of the carrier frequency deviation caused by the noise signal, it is determined whether the carrier frequency is interfered by the noise, and the neural network is trained based on this to obtain the carrier frequency deviation estimate. Compared with the traditional method of estimating the carrier frequency deviation using the FFT algorithm, this embodiment avoids the fence effect of the FFT algorithm while ensuring the real-time demodulation requirements of the carrier chip, and can obtain a more accurate carrier frequency deviation estimate, which can effectively improve the communication quality of the carrier chip.

[0012] In one embodiment, the method of taking the communication modulation signal and the standard carrier frequency as input and obtaining the standard modulation signal through signal demodulation and signal modulation is:

[0013] The communication modulation signal and its actual carrier frequency are used as input, a signal demodulation algorithm is adopted, the modulation mode adopts the PSK mode, and the output is a string of digital bit streams and the carrier frequency of the communication modulation signal. The actual carrier frequency is the sum of the standard carrier frequency and the carrier frequency deviation value at the previous moment. If it is the first moment, the standard carrier frequency is the actual carrier frequency;

[0014] The obtained digital bit stream and the carrier frequency of the communication modulation signal are used as input, and the signal modulation algorithm is used for calculation. The necessary parameter is the modulation mode. The modulation mode adopts the PSK mode, and the output is the standard modulation signal.

[0015] In one embodiment, the method of obtaining a noise signal according to the difference between the communication modulation signal and the standard modulation signal, and obtaining a noise spectrum vector by averaging the noise signal and performing frequency domain conversion is as follows:

[0016] Subtracting the standard modulation signal from the communication modulation signal to obtain a noise signal;

[0017] The noise signal is continuously divided into preset segments, each segment is recorded as a segmented noise signal, and the segmented noise signal is used as the input of the fast Fourier transform algorithm to obtain the noise spectrum vector.

[0018] In one embodiment, the method for obtaining the noise stationary weight at each frequency position according to the fluctuation characteristics of the energy amplitude at the same frequency position of all noise spectrum vectors is:

[0019] The standard deviation of the energy amplitudes of all segmented noise signals at the same frequency position is calculated as the amplitude standard deviation of the frequency position. After normalizing the amplitude standard deviation at the frequency position, the number 1 is subtracted from the normalized amplitude standard deviation to obtain the noise stationary weight of the frequency position.

[0020] In one embodiment, the method of obtaining harmonics based on frequency position and harmonic order, and obtaining the frequency difference of each order harmonic according to the difference between the harmonic and the standard carrier frequency is:

[0021] ,in, represents the frequency corresponding to the mth frequency position, h represents the order of harmonics, HF represents the standard carrier frequency, Indicates the frequency difference corresponding to the h-th harmonic at the m-th frequency position.

[0022] In one embodiment, the method for obtaining the frequency weight of each order harmonic at each frequency position according to the positive or negative value of the frequency difference is:

[0023] , Represents the frequency difference corresponding to the h-order harmonic at the m-th frequency position, Indicates the frequency weight corresponding to the h-order harmonic at the m-th frequency position, represents the linear normalization function.

[0024] In one embodiment, the method of obtaining the noise interference weight at each frequency position according to the frequency weights and energy attenuation multiples of all harmonics at the same frequency position and the noise stationary weight at each frequency position is:

[0025] , represents the noise stationary weight at the mth frequency position, represents the frequency weight of the h-th order harmonic at the m-th frequency position, represents the energy attenuation multiple of the h-th order harmonic, represents the maximum harmonic order, Represents the noise interference weight at the mth frequency position.

[0026] In one embodiment, the method for obtaining the noise interference intensity at each moment according to the ratio of the energy amplitude at different frequency positions in the frequency domain noise vector to the carrier energy amplitude combined with the noise interference weight at each frequency position is:

[0027] , represents the noise interference weight at the mth frequency position, Represents the energy amplitude of the mth frequency position of the frequency domain noise vector, represents the energy amplitude of the carrier, Indicates the noise interference intensity.

[0028] In one embodiment, the method of obtaining the carrier frequency offset estimation at the current moment through neural network training by taking the noise interference intensity and signal strength at the current moment as input is:

[0029] Obtain the noise interference intensity at several moments to form an interference intensity sequence; obtain the signal strength at several moments to form a modulation signal sequence; obtain the carrier frequency label by converting the frequency domain according to the communication modulation signal at the previous moment of each moment, and obtain the carrier frequency labels at several moments as the carrier frequency sequence;

[0030] The interference intensity sequence and the modulation signal sequence are used as the input part of the training data set, and the carrier frequency sequence is used as the label part of the training data set to train the neural network to obtain the trained neural network;

[0031] The noise interference intensity and signal intensity at the current moment are used as inputs to the trained neural network to obtain the frequency deviation carrier frequency, and the frequency deviation carrier frequency is subtracted from the standard carrier frequency to obtain the carrier frequency deviation estimate at the current moment.

[0032] In a second aspect, an embodiment of the present application also provides a carrier frequency offset estimation system based on a carrier chip, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above-mentioned carrier frequency offset estimation methods based on a carrier chip when executing the computer program.

[0033] The beneficial effects of this application are:

[0034] The communication modulation signal is obtained through the signal receiving module, and the communication modulation signal within the past 1 millisecond is obtained to estimate the carrier frequency deviation at the current moment; the noise signal is analyzed to obtain the frequency deviation interference intensity received by the carrier within the past 1 millisecond, and the carrier frequency deviation estimation is completed based on this. By extracting the characteristics that cause the carrier frequency deviation in the noise signal, it is determined whether the carrier frequency is interfered by the noise, and the neural network is trained based on this to obtain the carrier frequency deviation estimate. Compared with the traditional method of estimating the carrier frequency deviation using the FFT algorithm, this embodiment avoids the fence effect of the FFT algorithm while ensuring the real-time demodulation requirements of the carrier chip, and can obtain a more accurate carrier frequency deviation estimate, which can effectively improve the communication quality of the carrier chip. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0036] Figure 1 A flow chart of a carrier frequency offset estimation method based on a carrier chip provided in one embodiment of the present application;

[0037] Figure 2 It is a schematic diagram of communication modulation signal;

[0038] Figure 3 It is a schematic diagram of a standard modulation signal;

[0039] Figure 4 is a schematic diagram of noise signal;

[0040] Figure 5 Schematic diagram of noise spectrum vector. DETAILED DESCRIPTION

[0041] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the carrier frequency offset estimation method and system based on the carrier chip proposed in the present application, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0042] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0043] Carrier frequency offset estimation method embodiment based on carrier chip:

[0044] The specific scheme of the carrier frequency offset estimation method based on the carrier chip provided by the present application is described in detail below with reference to the accompanying drawings.

[0045] See also Figure 1 , which shows a flow chart of a carrier frequency offset estimation method based on a carrier chip provided by an embodiment of the present application, the method comprising the following steps:

[0046] Step S001, preset a standard carrier frequency, obtain a communication modulation signal, take the communication modulation signal and the standard carrier frequency as input, and obtain the standard modulation signal through signal demodulation and signal modulation.

[0047] The current line communication system based on the carrier chip includes a communication modulation module, a signal receiving module, a frequency deviation estimation module, and a signal demodulation module.

[0048] First, a preset carrier frequency is obtained in the signal demodulation module as a standard carrier frequency. The standard carrier frequency is 200K Hz in this embodiment.

[0049] The signal sender sends the signal through the communication modulation module, and the signal is obtained by the signal receiving module. The signal obtained by the signal receiving module is recorded as the communication modulation signal.

[0050] Among them, since the frequency of power line carrier communication is usually between 50k and 450k Hz, according to the Nyquist sampling theorem, the sampling frequency is at least twice the signal frequency, so the sampling frequency of the historical communication signal is 900k Hz; when the carrier frequency offset of the communication modulation signal is estimated in this embodiment, the data of the past 1 millisecond is obtained to estimate the frequency offset of the signal at the current moment, so the vector length N=900. In this embodiment, the communication modulation signal is as follows Figure 2 shown.

[0051] The present application confirms the carrier frequency offset estimate of all sampling points at the next moment through the carrier frequency offset estimate of all sampling points at the previous moment, calculates the average of the carrier frequency offset estimates of all sampling points at the previous moment, takes the average value as the carrier frequency offset value of the communication modulation signal, adds the carrier frequency offset value of the communication modulation signal at the previous moment to the standard carrier frequency, and obtains the actual carrier frequency of the communication modulation signal at the current moment; when the system is initialized, there is no carrier frequency offset estimate of the previous moment, and the standard carrier frequency is used as the actual carrier frequency.

[0052] Furthermore, the communication modulation signal and its actual carrier frequency are used as input, and a signal demodulation algorithm is used. The necessary parameter is the demodulation mode. The modulation mode of this embodiment adopts the PSK (phase shift keying) mode, and the output is a series of digital bit streams and the carrier frequency of the communication modulation signal, wherein the digital bit stream is the information transmitted by the signal within the past 1 millisecond. The signal demodulation algorithm is a commonly used technology in the field of communication.

[0053] In order to extract the noise characteristics from the communication modulation signal, it is necessary to subtract the non-noise part of the communication modulation signal, that is, it is necessary to obtain the signal when the communication modulation signal is not interfered by noise. Therefore, the obtained digital bit stream and the carrier frequency of the communication modulation signal are used as input, and the signal modulation algorithm is used for calculation. The necessary parameter is the modulation mode. The modulation mode of this embodiment adopts the PSK mode, and the output is a modulation signal, which is recorded as a standard modulation signal. The standard modulation signal is a signal with a length of N, which is a communication modulation signal without noise interference; the standard modulation signal is as follows Figure 3 shown.

[0054] So far, the communication modulation signal, the standard modulation signal and the standard carrier frequency have been acquired.

[0055] Step S002, obtain a noise signal based on the difference between the communication modulation signal and the standard modulation signal, divide the noise signal evenly and then perform frequency domain conversion to obtain a noise spectrum vector; obtain the noise stationary weight of each frequency position based on the fluctuation characteristics of the energy amplitude at the same frequency position of all noise spectrum vectors.

[0056] There are many kinds of noises in power line communication, among which the more random noises include: white noise, random pulse noise, etc. This kind of noise is relatively unstable, while the carrier has stable frequency domain characteristics, so this unstable noise has little effect on the carrier frequency deviation. Another relatively stable noise includes: colored background noise, narrowband noise, etc. This kind of noise is usually caused by components inside the power line. The source of this noise is stable and has relatively stable characteristics in the frequency domain. When its frequency is near the standard carrier frequency, it may be confused with the carrier, causing the carrier frequency deviation.

[0057] Therefore, the present application determines whether the corresponding frequency position is stable noise by the stability of the noise. Since stable noise has a stable period within a certain period of time, the corresponding amplitude value of stable noise in the frequency domain is relatively stable, and stable noise can be identified by the frequency domain characteristics of the communication modulation signal.

[0058] Subtract the standard modulation signal from the current communication modulation signal to obtain the noise signal. This step removes the non-noise part of the communication modulation signal and further analyzes the noise signal. Figure 4 shown.

[0059] The noise signal is continuously divided into K segments, and each segment is recorded as a segmented noise signal. In this embodiment, K is 9, that is, the length of each segmented noise signal is 100.

[0060] Any segmented noise signal is used as the input of the FFT algorithm (Fast Fourier Transform). In the FFT algorithm, the frequency range of the necessary parameters is selected as 0HZ to 450kHZ, and the sampling frequency is selected as 1kHZ. Thus, for each segmented noise signal, a vector of length M is obtained and recorded as the noise spectrum vector. In this embodiment, M is 450, that is, there are 450 elements, and the value of the mth element is the energy amplitude at a frequency of 1000m, that is, each element corresponds to a frequency position. The FFT algorithm is a well-known technology and will not be described in detail here; the noise spectrum vector is as follows Figure 5 shown.

[0061] For stationary noise, it is usually relatively stable at one frequency position. Therefore, for K segmented noise signals, the standard deviation of the energy amplitude of the K elements with the same frequency is calculated. The smaller the standard deviation, the more stable the frequency position corresponding to the mth element of the noise spectrum vector is, and the more obvious the stationary noise exists at the corresponding frequency position.

[0062] The noise stationary weight at each frequency position is obtained based on the standard deviation corresponding to each element. The expression is:

[0063] , where represents the standard deviation corresponding to the mth element, represents the linear normalization function, Represents the noise stationary weight of the frequency position corresponding to the mth element. The larger the standard deviation of each element, the smaller the noise stationary weight. The smaller the noise stationary weight, the less stable the energy amplitude of the frequency position corresponding to the mth element, and the less obvious stationary noise exists at the frequency position corresponding to the mth element.

[0064] In the above method, the noise signal is segmented, and each segmented noise signal is converted into a spectrum using the FFT algorithm. Each segmented noise signal has an energy amplitude in the corresponding frequency position. The stability of this frequency position is obtained by the amplitude change of all segmented noise signals in the same frequency position, and the noise stability weight of each frequency position is obtained.

[0065] At this point, the noise stationary weight corresponding to each frequency position is obtained.

[0066] Step S003, obtaining the energy attenuation multiple of the harmonic according to the preset harmonic attenuation amplitude; obtaining the harmonic based on the frequency position and the order of the harmonic, and obtaining the frequency difference of each order of harmonic according to the difference between the harmonic and the standard carrier frequency; obtaining the frequency weight of each order of harmonic at each frequency position according to the positive or negative value of the frequency difference; obtaining the noise interference weight at each frequency position according to the frequency weight and energy attenuation multiple of all orders of harmonics at the same frequency position, combined with the noise stabilization weight at each frequency position.

[0067] Noise is usually distributed over the entire frequency spectrum. For noise with lower frequencies and frequencies far from the standard carrier frequency, the carrier offset caused by it itself is relatively minor, and is usually caused by the harmonics it induces in the circuit.

[0068] The frequency position of the harmonic is usually a multiple of the noise frequency. Usually, the larger the multiple is, the higher the harmonic order is. The energy amplitude of the harmonic usually decays sharply with the order of the harmonic. Therefore, the higher the order of the harmonic, the smaller the impact on the carrier frequency deviation is. Usually, the closer the multiple position of the noise is to the standard carrier frequency, the greater the impact of the harmonic of the noise on the carrier. Therefore, this application determines the impact of the noise at each frequency position on the carrier frequency deviation by calculating the harmonic frequency characteristics of the noise.

[0069] As the harmonic order of noise increases, the harmonic energy value usually decreases sharply. The reduction of the harmonic energy value is usually related to the material properties of the current and the grid structure. In this embodiment, the harmonic attenuation amplitude of the grid is taken as an empirical value of -10 dB, which means that the energy amplitude of the hth harmonic of a signal is 1 / 10 of the fundamental wave. times, that is, the harmonic energy attenuation times is At the same time, this embodiment considers that the harmonic energy amplitude exceeding the H order is extremely small and can be ignored. In this embodiment, the maximum harmonic order H=7.

[0070] For each frequency position, the influence of its H-order harmonic on the carrier frequency deviation is calculated to obtain the noise interference weight.

[0071] The frequency difference of each harmonic at each frequency position is calculated based on the difference between the stationary noise at each frequency position at different harmonics and the standard carrier frequency. The expression is:

[0072] ,in, represents the frequency corresponding to the mth frequency position, h represents the order of harmonics, HF represents the standard carrier frequency, Represents the frequency difference corresponding to the h-order harmonic at the m-th frequency position, Indicates the magnitude of the harmonic. The larger the frequency difference, the smaller the frequency deviation effect of the corresponding harmonic on the carrier.

[0073] Furthermore, the frequency weight is obtained according to all frequency differences of the H-order harmonics at M frequency positions. The expression of the frequency weight is:

[0074] , Represents the frequency difference corresponding to the h-order harmonic at the m-th frequency position, Indicates the frequency weight corresponding to the h-order harmonic at the m-th frequency position, Represents a linear normalized function. The larger the absolute value of the frequency weight, the greater the impact of the carrier frequency deviation caused by the corresponding harmonic. When the frequency difference is negative, it means that the harmonic tends to reduce the carrier frequency after being confused with the carrier, and vice versa, the harmonic tends to increase the carrier frequency.

[0075] The noise interference weight corresponding to each frequency position is obtained according to the frequency weights of harmonics of different orders at the frequency position, the harmonic energy attenuation multiples of harmonics of different orders and the noise stationary weight corresponding to the frequency position.

[0076] It should be noted that positive correlation means that when one variable increases, the other variable also increases, and the two variables change in the same direction. When one variable changes from large to small or from small to large, the other variable also changes from large to small or from small to large. The specific relationship is determined by actual application and this application does not impose any special restrictions.

[0077] Preferably, in this embodiment, the expression of the noise interference weight is:

[0078] , represents the noise stationary weight at the mth frequency position, represents the frequency weight of the h-th order harmonic at the m-th frequency position, represents the energy attenuation multiple of the h-th order harmonic, represents the maximum harmonic order, Represents the noise interference weight at the mth frequency position.

[0079] Among them, the smaller the frequency weight of each harmonic, or the smaller the energy attenuation multiple, the smaller the influence of the h-th order harmonic on the frequency deviation of the carrier; the positive or negative frequency weight of the h-th order harmonic represents whether the h-th order harmonic tends to increase or decrease the carrier frequency; the sum of the frequency weight of the H-th order harmonic at the m-th frequency position and the energy attenuation multiple represents the overall influence weight of the harmonic at the m-th frequency position on the carrier; the noise stability weight represents the noise stability degree at the m-th frequency position. The more stable the noise at the m-th frequency position, the more the noise itself and its corresponding high-order harmonics can interfere with the carrier frequency.

[0080] The larger the absolute value of the noise interference weight is, the stronger the noise interference capability at the mth frequency position is, and the more serious the frequency deviation is. The positive or negative value of the noise interference weight corresponds to whether the noise increases or decreases the carrier frequency.

[0081] At this point, the noise interference weight at each frequency position is obtained.

[0082] Step S004, the noise signal and the standard modulation signal at each moment are converted into the frequency domain to obtain a frequency domain noise vector and a frequency domain signal vector. The energy amplitude of the standard carrier frequency position of the frequency domain signal vector is the energy amplitude of the carrier. The noise interference intensity at each moment is obtained according to the ratio of the energy amplitude at different frequency positions in the frequency domain noise vector to the carrier energy amplitude combined with the noise interference weight at each frequency position.

[0083] According to the above steps, the noise interference weight of each frequency position is obtained; the noise interference weight can be used to distinguish the influence of noise at different frequency domain positions on the carrier frequency deviation in the past 1 millisecond. Therefore, it is necessary to further combine the communication modulation signal to obtain the degree of influence of noise on the carrier, judge the influence of noise on the carrier in the past 1 millisecond, and then estimate the carrier frequency deviation of the next communication modulation signal.

[0084] First, the noise signal and the standard modulation signal are converted into frequency spectrum through the FFT algorithm. The input of the algorithm is the noise signal and the standard modulation signal. The necessary parameters are the frequency range and the sampling frequency. The frequency range is selected from 0 Hz to 450k Hz, and the sampling frequency is selected from 1k Hz. The output is two vectors with a length of M=450, which are recorded as the frequency domain noise vector and the frequency domain signal vector respectively.

[0085] The frequency domain noise vector and the frequency domain signal vector have an element value of each element that is the energy amplitude corresponding to the corresponding frequency position.

[0086] In the frequency domain signal vector, the energy amplitude corresponding to the standard carrier frequency position is the energy amplitude of the carrier.

[0087] Furthermore, the noise interference intensity is obtained according to the ratio of the energy amplitude at each frequency position in the frequency domain noise vector to the energy amplitude of the carrier, combined with the noise interference weight as a weight.

[0088] Preferably, in this embodiment, the expression of noise interference intensity is:

[0089] , represents the noise interference weight at the mth frequency position, Represents the energy amplitude of the mth frequency position of the frequency domain noise vector, represents the energy amplitude of the carrier, Indicates the noise interference intensity.

[0090] The noise interference intensity represents the influence of noise on the carrier frequency of the communication modulation signal in the past 1 millisecond. The positive or negative value of the noise interference intensity corresponds to whether the noise increases or decreases the carrier frequency as a whole. The absolute value of the noise interference intensity represents the interference strength of the noise on the carrier frequency.

[0091] At this point, the noise interference intensity at each moment is obtained.

[0092] Step S005, taking the noise interference intensity and signal intensity at the current moment as input, obtaining the carrier frequency offset estimation at the current moment through neural network training.

[0093] When the frequency offset of the carrier is estimated by the neural network model, a training data set needs to be obtained first. For any moment, the noise interference intensity at this moment can be obtained according to the time before 1 ms, and then the noise interference intensity at several moments is obtained by the sampling frequency. The sampling frequency is 900 kHz. In this embodiment, the number of data sets is selected as 5000, so the noise interference intensity at 5000 moments constitutes an interference intensity sequence;

[0094] For the communication modulation signal, the sampling frequency is 900 kHz, and the signal strength at each moment is obtained. In this embodiment, the data set is also selected as 5000, and the signal strength at 5000 moments is obtained to form a modulation signal sequence.

[0095] Furthermore, the communication modulation signal of the previous millisecond at each moment is obtained, and the communication modulation signal at this moment is used as input. The FFT algorithm is used for calculation, and the necessary parameters are the frequency range and sampling frequency. The frequency range is selected from 0 Hz to 450k Hz, and the sampling frequency is selected from 1 Hz. The output is a vector with a length of S=450000, which is recorded as the frequency domain communication vector. The frequency position with the highest energy amplitude in the frequency domain communication vector is obtained, and this frequency position is recorded as the frequency position where the carrier is located, and recorded as the carrier frequency label at each moment.

[0096] The carrier frequency label at each moment is obtained through the sampling frequency. In this embodiment, the carrier frequency labels at 5000 moments are obtained to form a sequence recorded as a carrier frequency sequence.

[0097] The interference intensity sequence and the modulation signal sequence are used as the input part of the training data set, and the carrier frequency sequence is used as the label part of the training data set. The training data set is used as the input to the LSTM neural network model, and the output of the neural network model is the frequency deviation carrier frequency. The activation function uses the ReLU function, and the optimizer uses the Adam optimizer; the loss function is obtained by taking the mean square error of the sequence value of the frequency deviation carrier frequency and the carrier frequency sequence; finally, the trained LSTM neural network model is obtained. The training process of the LSTM neural network model is a common technology in the field of communications.

[0098] When using the trained LSTM neural network model for carrier frequency offset estimation, the current noise interference intensity and signal strength are input into the trained LSTM model to obtain the current frequency offset carrier frequency. Subtract the standard carrier frequency from the frequency offset carrier frequency to obtain the current carrier frequency offset estimate.

[0099] In the frequency offset estimation module, the carrier frequency offset estimation is obtained through the above steps, and in the signal demodulation module, the communication modulated signal is demodulated using the carrier frequency offset estimation.

[0100] Based on the same inventive concept as the above method, an embodiment of the present invention also provides a carrier frequency offset estimation system based on a carrier chip, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above-mentioned carrier frequency offset estimation methods based on a carrier chip when executing the computer program.

[0101] It should be noted that the above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

[0102] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A carrier frequency offset estimation method based on a carrier chip, characterized in that: The method comprises the following steps: Preset a standard carrier frequency, obtain a communication modulation signal, take the communication modulation signal and the standard carrier frequency as input, and obtain the standard modulation signal through signal demodulation and signal modulation; The noise signal is obtained according to the difference between the communication modulation signal and the standard modulation signal, and the noise signal is evenly divided and then converted into the frequency domain to obtain the noise spectrum vector; the noise stability weight of each frequency position is obtained according to the fluctuation characteristics of the energy amplitude at the same frequency position of all noise spectrum vectors; The energy attenuation multiple of the harmonic is obtained according to the preset harmonic attenuation amplitude; the harmonic is obtained based on the frequency position and the order of the harmonic, and the frequency difference of each order of the harmonic is obtained according to the difference between the harmonic and the standard carrier frequency; the frequency weight of each order of the harmonic at each frequency position is obtained according to the positive and negative of the frequency difference; the noise interference weight at each frequency position is obtained according to the frequency weight and energy attenuation multiple of all orders of harmonics at the same frequency position, combined with the noise stabilization weight at each frequency position; The noise signal and the standard modulation signal at each moment are converted into the frequency domain to obtain the frequency domain noise vector and the frequency domain signal vector. The energy amplitude at the standard carrier frequency position of the frequency domain signal vector is the energy amplitude of the carrier. The noise interference intensity at each moment is obtained according to the ratio of the energy amplitude at different frequency positions in the frequency domain noise vector to the carrier energy amplitude combined with the noise interference weight at each frequency position; The current noise interference intensity and signal strength are used as input, and the carrier frequency offset estimation at the current moment is obtained through neural network training.

2. The carrier frequency offset estimation method based on a carrier chip according to claim 1, characterized in that: The method of taking the communication modulation signal and the standard carrier frequency as input and obtaining the standard modulation signal through signal demodulation and signal modulation is: The communication modulation signal and its actual carrier frequency are used as input, a signal demodulation algorithm is adopted, the modulation mode adopts the PSK mode, and the output is a string of digital bit streams and the carrier frequency of the communication modulation signal. The actual carrier frequency is the sum of the standard carrier frequency and the carrier frequency deviation value at the previous moment. If it is the first moment, the standard carrier frequency is the actual carrier frequency; The obtained digital bit stream and the carrier frequency of the communication modulation signal are used as input, the signal modulation algorithm is used for calculation, the modulation mode adopts the PSK mode, and the output is the standard modulation signal.

3. The carrier frequency offset estimation method based on a carrier chip as claimed in claim 1, characterized in that: The method of obtaining the noise signal according to the difference between the communication modulation signal and the standard modulation signal, and obtaining the noise spectrum vector by performing frequency domain conversion after averaging the noise signal is as follows: Subtracting the standard modulation signal from the communication modulation signal to obtain a noise signal; The noise signal is continuously divided into preset segments, each segment is recorded as a segmented noise signal, and the segmented noise signal is used as the input of the fast Fourier transform algorithm to obtain the noise spectrum vector.

4. The carrier frequency offset estimation method based on a carrier chip as claimed in claim 3, characterized in that: The method for obtaining the noise stationary weight at each frequency position according to the fluctuation characteristics of the energy amplitude at the same frequency position of all noise spectrum vectors is: The standard deviation of the energy amplitudes of all segmented noise signals at the same frequency position is calculated as the amplitude standard deviation of the frequency position. After normalizing the amplitude standard deviation at the frequency position, the number 1 is subtracted from the normalized amplitude standard deviation to obtain the noise stationary weight of the frequency position.

5. The carrier frequency offset estimation method based on a carrier chip according to claim 1, characterized in that: The method of obtaining harmonics based on frequency position and harmonic order, and obtaining the frequency difference of each order harmonic according to the difference between the harmonic and the standard carrier frequency is: ,in, represents the frequency corresponding to the mth frequency position, h represents the order of harmonics, HF represents the standard carrier frequency, Indicates the frequency difference corresponding to the h-th harmonic at the m-th frequency position.

6. The carrier frequency offset estimation method based on a carrier chip according to claim 1, characterized in that: The method for obtaining the frequency weight of each order harmonic at each frequency position according to the positive or negative frequency difference is: , Represents the frequency difference corresponding to the h-order harmonic at the m-th frequency position, Indicates the frequency weight corresponding to the h-order harmonic at the m-th frequency position, represents the linear normalization function.

7. The carrier frequency offset estimation method based on a carrier chip according to claim 1, characterized in that: The method for obtaining the noise interference weight at each frequency position according to the frequency weights and energy attenuation multiples of all harmonics at the same frequency position and the noise stationary weight at each frequency position is as follows: , represents the noise stationary weight at the mth frequency position, represents the frequency weight of the h-th order harmonic at the m-th frequency position, represents the energy attenuation multiple of the h-th order harmonic, represents the maximum harmonic order, Represents the noise interference weight at the mth frequency position.

8. The carrier frequency offset estimation method based on a carrier chip as claimed in claim 1, characterized in that: The method for obtaining the noise interference intensity at each moment according to the ratio of the energy amplitude at different frequency positions in the frequency domain noise vector to the carrier energy amplitude combined with the noise interference weight at each frequency position is: , represents the noise interference weight at the mth frequency position, Represents the energy amplitude of the mth frequency position of the frequency domain noise vector, represents the energy amplitude of the carrier, Indicates the noise interference intensity.

9. The carrier frequency offset estimation method based on a carrier chip according to claim 1, characterized in that: The method of obtaining the carrier frequency offset estimation at the current moment by using the noise interference intensity and signal intensity at the current moment as input through neural network training is: Obtain the noise interference intensity at several moments to form an interference intensity sequence; obtain the signal strength at several moments to form a modulation signal sequence; obtain the carrier frequency label by converting the frequency domain according to the communication modulation signal at the previous moment of each moment, and obtain the carrier frequency labels at several moments as the carrier frequency sequence; The interference intensity sequence and the modulation signal sequence are used as the input part of the training data set, and the carrier frequency sequence is used as the label part of the training data set to train the neural network to obtain the trained neural network; The noise interference intensity and signal intensity at the current moment are used as inputs to the trained neural network to obtain the frequency deviation carrier frequency, and the frequency deviation carrier frequency is subtracted from the standard carrier frequency to obtain the carrier frequency deviation estimate at the current moment.

10. A carrier frequency offset estimation system based on a carrier chip, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the carrier frequency offset estimation method based on the carrier chip as described in any one of claims 1-9 are implemented.

Images