Frequency estimation device

Abstract

The present invention relates to a frequency estimation device, and a specific embodiment of the device includes: a frequency comb generating unit, a pre-sampling unit and a signal acquisition and processing unit; the frequency comb generating unit is used to generate a pre-sampling frequency comb; the pre-sampling unit uses the pre-sampling frequency comb to pre-sample the signal to be measured in the analog domain to obtain a pre-sampling signal; the pre-sampling signal includes: multiple signal copies located in each Nyquist zone of the pre-sampling frequency comb; the signal acquisition and processing unit performs analog-to-digital conversion on the pre-sampling signal, performs fast Fourier transform on the pre-sampling signal in the digital domain formed by the analog-to-digital conversion, obtains the frequency measurement value of the signal copy and the frequency measurement value of the signal copy on the signal to be measured; and determines the average value of the frequency measurement value of the signal copy on the signal to be measured as the frequency estimation value of the signal to be measured. This embodiment can significantly improve the signal frequency estimation accuracy through pre-sampling and multi-measurement value averaging at the front end of the FFT processing link.

Description

Technical Field

[0001] The present invention relates to the technical field of microwave frequency measurement, and in particular to a frequency estimation device. Background Art

[0002] Accurate frequency estimation is an important issue in signal processing in the fields of communication, radar, sonar, and electronic countermeasures. The Fast Fourier Transform (FFT) algorithm is widely used to obtain the frequency information of a signal. Usually, the frequency value of the signal can be obtained by directly reading the frequency grid point where the peak power of the signal is located from the spectrum information calculated by FFT. However, the frequency estimation accuracy of this method depends on the length of the time window for calculating the FFT. Due to the application requirements of the actual application scenario and the signal processing cost of the detection end, the window length cannot be very long, resulting in the FFT frequency grid point accuracy being relatively rough and difficult to meet practical needs. In order to improve the frequency estimation accuracy based on FFT, people have proposed a variety of high-precision frequency estimation methods, which can further improve the frequency estimation accuracy on the FFT spectrum results, such as polynomial fitting, zero padding, interpolated fast Fourier transform (IFFT), weighted phase averaging (WPA), and zoomFFT. Although these methods are effective, they are all based on digital signal processing, that is, for signals that have been digitally quantized, after the FFT processing link, the estimation accuracy of the signal frequency can be further improved through supplementary algorithms. The complexity of these algorithms may occupy a lot of computing resources and limit the timeliness of frequency estimation. Therefore, it is necessary to propose a method to improve the accuracy of signal frequency estimation by signal preprocessing at the front end of FFT processing. Summary of the invention

[0003] In view of the defects in the prior art, the present invention provides a frequency estimation device that can significantly improve the signal frequency estimation accuracy through pre-sampling and multi-measurement value averaging at the front end of the FFT processing link. At the same time, this device is compatible with the existing FFT-based post-processing high-precision frequency estimation method.

[0004] The frequency estimation device of the present invention comprises: a frequency comb generating unit, a pre-sampling unit and a signal acquisition and processing unit; wherein the frequency comb generating unit is used to generate a pre-sampling frequency comb; the pre-sampling unit uses the pre-sampling frequency comb to pre-sample the signal to be measured in the analog domain to obtain a pre-sampling signal; the pre-sampling signal comprises: a plurality of signal copies of the signal to be measured located in each Nyquist zone of the pre-sampling frequency comb; the signal acquisition and processing unit performs analog-to-digital conversion on the pre-sampling signal, performs fast Fourier transform on the pre-sampling signal in the digital domain formed by the analog-to-digital conversion, and obtains frequency measurement values ​​of the plurality of signal copies; the signal acquisition and processing unit selects a preset number of Nyquist zones from the Nyquist zone, calculates the frequency measurement value of the signal copy to the signal to be measured by using the frequency measurement value of the signal copy selected in each Nyquist zone, and determines the average value of the frequency measurement value of the signal copy selected in each Nyquist zone to the signal to be measured as the frequency estimation value of the signal to be measured.

[0005] The signal acquisition and processing unit calculates the frequency measurement value of the signal to be measured by the signal replica of any selected Nyquist zone through the following formula:

[0006] At the frequency f of the signal to be measured in When in the 2n+1th Nyquist zone:

[0007] When i is an odd number,

[0008] When i is an even number,

[0009] In f in When in the 2n+2nd Nyquist zone:

[0010] When i is an odd number,

[0011] When i is an even number,

[0012] Where n is a non-negative integer, i is the serial number of any selected Nyquist zone, and f inM，i is the frequency measurement value of the signal to be measured by the signal replica of any selected Nyquist zone, f M，i is the frequency measurement value of the signal replica in any selected Nyquist zone, f c is the repetition frequency of the pre-sampling frequency comb.

[0013] The preset number is equal to the number of signal copies used to calculate the frequency estimate of the signal to be measured; the frequency of each signal copy can be expressed as a linear combination expression of the frequency of the signal to be measured and the repetition frequency; and the frequencies of any two signal copies in the Nyquist zone selected by the signal acquisition and processing unit satisfy: the difference between the coefficients of the repetition frequency in the linear combination expression is not equal to an integer multiple of the preset number.

[0014] The decimal part of the ratio of the repetition frequency to the frequency resolution of the fast Fourier transform satisfies: being equal to the reciprocal of the preset number.

[0015] The signal acquisition and processing unit includes: an analog-to-digital converter and a signal processing module; wherein the analog-to-digital converter is used to perform the analog-to-digital conversion; the signal processing module is used to determine the Nyquist zone sequence number where the frequency of the signal to be measured is located, perform the fast Fourier transform, and calculate the frequency measurement value of the signal to be measured and the frequency estimation value of the signal to be measured for the signal replica of each selected Nyquist zone.

[0016] The frequency comb generating unit has a first structure, a second structure or a third structure; in the first structure, the frequency comb generating unit is an optical frequency comb generating device, including a mode-locked laser or an electro-optical modulated optical frequency comb generating device, and the pre-sampling frequency comb is an optical frequency comb; in the second structure, the frequency comb generating unit is an electrical frequency comb generating device, including a nonlinear transmission line or a step recovery diode, and the pre-sampling frequency comb is an electrical frequency comb; in the third structure, the frequency comb generating unit includes the optical frequency comb generating device and a first photodetector, and the pre-sampling frequency comb is an electrical frequency comb.

[0017] When the frequency comb generating unit is of the first structure, the pre-sampling unit includes: an electro-optical modulator, an optical amplifier, an optical filter and a second photodetector; wherein the electro-optical modulator pre-samples the signal to be measured, and the optical amplifier, the optical filter and the second photodetector are used to process the pre-sampled signal; when the frequency comb generating unit is of the second structure or the third structure, the pre-sampling unit includes an electrical mixer for pre-sampling the signal to be measured.

[0018] The device further includes a first signal conditioning unit connected to the pre-sampling unit, and a second signal conditioning unit respectively connected to the pre-sampling unit and the signal acquisition and processing unit; wherein the first signal conditioning unit includes a first electrical filter, a first electrical amplifier and a first electrical attenuator, which are used to process the signal to be measured and transmit the processed signal to be measured to the pre-sampling unit; the second signal conditioning unit includes a second electrical filter, a second electrical amplifier and a second electrical attenuator, which are used to process the pre-sampling signal and transmit the processed pre-sampling signal to the signal acquisition and processing unit.

[0019] The signal acquisition and processing unit is further used to: after calculating the frequency estimation value of the signal to be measured, use a preset frequency estimation algorithm to perform further calculation based on the calculated frequency estimation value of the signal to be measured; wherein the frequency estimation algorithm includes at least one of the following: polynomial fitting, zero padding, interpolated fast Fourier transform, weighted phase averaging, and band-selected Fourier transform.

[0020] The device further includes: a control unit respectively connected to the frequency comb generating unit, the pre-sampling unit, the signal acquisition and processing unit, the first signal conditioning unit and the second signal conditioning unit; wherein the control unit is used to: set the repetition frequency and power of the pre-sampling frequency comb, set the parameters of the first electrical filter, the first electrical amplifier, the first electrical attenuator, the second electrical filter, the second electrical amplifier and the second electrical attenuator, set the parameters of each device in the pre-sampling unit, set the sampling rate of the analog-to-digital conversion, the number of calculation points of the fast Fourier transform and the preset number, and provide a reference clock frequency for the frequency comb generating unit and the signal acquisition and processing unit.

[0021] The frequency estimation device of the present invention has the following beneficial effects: the signal is pre-sampled in the analog domain by using a frequency comb with a specific repetition frequency, and the signal to be measured is expanded into multiple pre-sampled Nyquist zones. Then, in the digital domain, the signal replica frequency values ​​of multiple Nyquist zones in the FFT spectrum of the pre-sampled signal are extracted, and a simple averaging algorithm is used to significantly improve the signal frequency estimation accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is a schematic structural diagram of a frequency estimation device according to an embodiment of the present invention;

[0023] Figure 2 is a schematic diagram of pre-sampling in an embodiment of the present invention;

[0024] Figure 3 is a schematic diagram of measurement error in an embodiment of the present invention;

[0025] Figure 4is another measurement error schematic diagram of an embodiment of the present invention;

[0026] Figure 5 is a structural schematic diagram of a first signal conditioning unit according to an embodiment of the present invention;

[0027] Figure 6 is a structural schematic diagram of a frequency comb generating unit according to an embodiment of the present invention;

[0028] Figure 7 is a structural schematic diagram of a pre-sampling unit in an embodiment of the present invention;

[0029] Figure 8 is another structural schematic diagram of a pre-sampling unit according to an embodiment of the present invention;

[0030] Fig. 9 It is a structural schematic diagram of a signal acquisition and processing unit according to an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] The present invention proposes a high-precision frequency estimation device based on pre-sampling and multi-measurement value averaging, that is, using a frequency comb with a specific repetition frequency to pre-sample the signal in the analog domain, and expand the signal to be measured into multiple pre-sampling Nyquist zones to form multiple signal copies. Then, in the digital domain, the pre-sampled signal is transformed by FFT, and the frequency measurement values ​​of the signal copies in multiple Nyquist zones are extracted in the FFT spectrum. Through a simple frequency averaging algorithm, the estimation accuracy of the frequency of the signal to be measured can be significantly improved. The pre-sampling process and processing algorithm of the device are very simple and reliable, and there is no need to introduce other complex algorithms. In addition, as a representative signal processing method, the device of the present invention is also compatible with other existing high-precision frequency estimation algorithms based on FFT. The specific structure of the present invention can be found in Figure 1 , the technical concept and principle of the present invention are first explained below.

[0033] For a frequency measurement system based on FFT (Fast Fourier Transform), the sampling rate is F s , the number of calculation points is X, then the frequency resolution is f res =F s / X, the maximum frequency estimation error is fres / 2. Assume that the center frequency of the signal to be measured is f in ,like Figure 2 As shown, it can be expressed as

[0034] f in =(m+δ)f res (1)

[0035] Where m is a positive integer and δ is a decimal in the interval [0, 1). In the FFT spectrum, the signal frequency measurement is expressed as

[0036] f inM =m×f res +[δ]×f res (2)

[0037] Where [·] means rounding to the nearest integer. From formulas (1) and (2), we can get the measurement error of the signal frequency:

[0038] Δ=f inM -f in ={[δ]-δ}f res (3)

[0039] Since [δ] = int(δ) + [rmod(δ)], where int(x) and rmod(x) represent the integer part and the decimal part of x, respectively, formula (3) is simplified to

[0040] Δ={[rmod(δ)]-rmod(δ)}f res (4)

[0041] Obviously, since 0≤δ<1, so -0.5<Δ≤0.5, so the maximum estimation error of the original signal is f res / 2, which is caused by the discreteness of spectrum distribution in FFT frequency detection system.

[0042] If the repetition frequency is f c =(α+ε)f res Frequency comb (such as Figure 2 As shown in Figure 2, the signal to be measured is pre-sampled, and the signal spectrum after sampling is as follows: Figure 2 As shown, the signal is copied to each pre-sampled Nyquist zone of the pre-sampled frequency comb, that is, each Nyquist zone of the pre-sampled frequency comb will contain a copy of the signal to be measured. If the signal to be measured is located between the nth (n is a non-negative integer) comb tooth and the n+1th comb tooth, that is, the 2n+1th or 2n+2th Nyquist zone, then after pre-sampling, it is located in the i-th pre-sampled Nyquist zone The frequency f of the inner signal replica i It can be expressed as:

[0043] (1) In the signal f in When in the 2n+1th Nyquist zone:

[0044]

[0045] or

[0046]

[0047] (2) In the signal f in When in the 2n+2nd Nyquist zone:

[0048]

[0049] or

[0050]

[0051] Here f c =(α+ε)f res is the repetition frequency of the pre-sampled frequency comb, nf c represents the frequency of the nth comb tooth, α is a positive integer, and ε is a decimal in the interval [0, 1). In the FFT spectrum, the measured value f of the frequency value (5-1) to (5-4) M，i It is expressed as:

[0052] (1) In the signal f in When in the 2n+1th Nyquist zone:

[0053]

[0054] or

[0055]

[0056] (2) In the signal f in When in the 2n+2nd Nyquist zone:

[0057]

[0058] or

[0059]

[0060]

[0061] Obviously, if the frequency f of the signal to be measured is known in The Nyquist zone where it is located, that is, the comb order n or n+1 of the adjacent pre-sampling frequency is known, and the frequency measurement value f of the signal replica located in the i-th pre-sampling Nyquist zone is obtained by using the signal acquisition and processing unit. M，i, then it is easy to calculate the frequency measurement value f of the signal to be measured from the frequency measurement value through formulas (6-1) to (6-4) inM，i , which is the frequency measurement value of the signal replica of the signal to be measured.

[0062] (1) In the signal f in When in the 2n+1th Nyquist zone:

[0063]

[0064] or

[0065]

[0066] (2) In the signal f in When in the 2n+2nd Nyquist zone:

[0067]

[0068] or

[0069]

[0070] Then, the frequency measurement error of the measured signal can be deduced as:

[0071] (1) In the signal f in When in the 2n+1th Nyquist zone:

[0072]

[0073] or

[0074]

[0075]

[0076] (2) In the signal f in When in the 2n+2nd Nyquist zone:

[0077]

[0078] or

[0079]

[0080] Obviously, Δ i The expression of is similar to that of Δ, and its value depends on δ and ε, and its range is (-0.5, 0.5]. Without loss of generality. For a frequency-stable signal to be measured, it has a certain δ, and the repetition frequency of the pre-sampling frequency comb is variable, its ε is also settable, and the fractional part of the comb teeth of different orders has the property of rmod(nε). Δ i The relationship between the value of and ε is as follows Figure 3 As shown in Figure 3, δ is taken as 0.2f res , au represents dimensionless, Δ max is Δ i The maximum value of .

[0081] Obviously, depending on the value of ε, Δ i The distribution is in the whole (-0.5, 0.5] interval, while the value of rmod(nε) can cover the whole [0, 1) interval. Therefore, we can consider res , set a specific ε value so that ε×N=1, N is a preset number, which is a positive integer, representing the number of signal copies selected to calculate the frequency estimate of the signal to be measured (that is, the frequency value of the signal to be measured obtained by averaging multiple measurement values). By collecting signal copies of the signal to be measured located in N different pre-sampled Nyquist zones, their frequency measurement values ​​f are obtained. M，i (i.e., the frequency measurement value of the signal to be measured by each signal replica), and further calculating the average value of these frequency measurement values, which may help reduce the frequency measurement error. The specific analysis process is as follows.

[0082] For any integers k and l, since [rmod(δ-(n±k±lN)ε)]- rmod(δ-(n±k±lN)ε)=[rmod(δ-(n±k)ε)]-rmod(δ- (n±k)ε), the following value conditions should be met when selecting signal replicas: In expressions (6-1)~(6-4) and (8-4)~(8-4) of any two signal replicas, the difference in the coefficient of ε cannot be an integer multiple of N, otherwise the measured values ​​of the two signal replicas are equivalent, which will affect the measurement accuracy. Since the coefficient of ε mentioned above is f c That is, for the frequencies of any two selected signal copies in the Nyquist zone, the difference between the coefficients of the repetition frequency in the linear combination expression is not equal to an integer multiple of the above-mentioned preset number N.

[0083] Given that the FFT spectrum range of the pre-sampled signal is usually limited, generally from DC to a limited cutoff frequency, it mainly includes the low-order Nyquist zone of the pre-sampled signal. Therefore, N signal copies that meet the above-mentioned value conditions should be selected from these Nyquist zones for average value calculation. A typical average value calculation method is to select the 1st to Nth consecutive pre-sampled Nyquist zone signal copies in the FFT spectrum of the pre-sampled signal, and use the frequency measurement values ​​of the signal to be measured obtained from these copies to further average the frequency of the signal to be measured. For simplicity, assume that the signal to be measured f in Located in the 2n+1th pre-sampled Nyquist zone [nf c , (n+1 / 2)fc ], N is an even number, and the pre-sampled Nyquist zone ordinal number i∈[1,N], then the average frequency measurement value based on the 1st to Nth pre-sampled Nyquist zone signal replicas (i.e., the frequency estimation value of the signal to be measured) is

[0084]

[0085] Formula (9) can be simplified to

[0086]

[0087] Here Δ a is the error in the measured mean, expressed as

[0088]

[0089] Δ and Δ a The relationship with different δ values ​​is as follows Figure 4 As shown, where ε=0.1, ε×N=1, is Δ a The maximum value of Δ a The value range is reduced from (-0.5, 0.5] to (-0.05, 0.05], which means that the frequency estimation accuracy is improved by 10 times. Based on the above pre-sampling and average value calculation method, other types of high-precision frequency estimation algorithms based on FFT can also be applied to further reduce the measurement error (i.e. Figure 4 measurementderivation in ).

[0090] In addition to the above conditions set for simplicity, for the measured signal f in Located in any pre-sampled Nyquist zone, and N is any natural number, as long as the condition ε×N=1 is satisfied, and the difference between the coefficients of ε of any two signal copies selected for calculating the average frequency of the signal to be measured is not an integer multiple of N, the effect of improving the frequency estimation accuracy of the signal to be measured as in formula (11) can be achieved. For example, the average frequency of the signal to be measured is calculated using N consecutive odd-numbered pre-sampled Nyquist zone signal copies i.

[0091] According to the above technical concept, it can be designed Figure 1 The frequency estimation device shown in FIG. Figure 1 As shown, the frequency estimation device of the embodiment of the present invention may include: a frequency comb generating unit, a pre-sampling unit and a signal acquisition and processing unit.

[0092] Among them, the frequency comb generating unit is connected to the pre-sampling unit, and the pre-sampling unit is connected to the signal acquisition and processing unit. The frequency comb generating unit generates a pre-sampling frequency comb for pre-sampling. In practical applications, the above pre-sampling frequency comb can be an optical frequency comb or an electrical frequency comb. The frequency comb generating unit can adopt any generation method that can meet the requirements of the pre-sampling frequency comb. Typical methods include optical frequency comb generating devices (such as mode-locked lasers, electro-optical modulation optical frequency comb generating devices, etc., which can generate optical frequency comb signals), electrical frequency comb generating devices (such as nonlinear transmission lines, step recovery diodes, etc., which can generate electrical frequency comb signals). It is also possible to first generate an optical frequency comb through an optical frequency comb generating device, and then input the optical frequency comb into a photodetector to form an electrical frequency comb. The basic requirement that the pre-sampling frequency comb needs to meet is that the repetition frequency f can be specially set. c , the comb teeth interval is strictly equal to the repetition frequency, and the comb teeth have a relatively flat power distribution on the spectrum. The pre-sampling frequency comb generated by the frequency comb generation unit is input into the pre-sampling unit to pre-sample the signal to be measured.

[0093] The pre-sampling unit uses the pre-sampling frequency comb generated by the frequency comb generating unit to pre-sample the signal to be tested in the analog domain to obtain a pre-sampled signal. The pre-sampled signal may include: multiple signal copies of the signal to be tested located in each Nyquist zone of the pre-sampling frequency comb. Figure 2 is a schematic diagram of pre-sampling in an embodiment of the present invention, such as Figure 2 As shown, A shows the input signal f in That is, the signal to be measured. B shows the spectrum of the pre-sampled frequency comb with a repetition frequency of f C Integer multiple frequencies such as 0, f C , 2f C 、3f C ...are the teeth of the frequency comb. Starting from the tooth with a frequency of zero, there are two Nyquist zones on the left and right between adjacent teeth. The serial numbers of each Nyquist zone increase from 1. The serial numbers of the left Nyquist zone are odd numbers, and the serial numbers of the right Nyquist zone are even numbers. Figure 2 C in the figure shows the pre-sampling result, i.e., the spectrum of the pre-sampling signal. The principle of pre-sampling is to multiply the signal to be measured by the pre-sampling frequency comb, so that the signal to be measured is copied to each pre-sampling Nyquist zone of the pre-sampling frequency comb, i.e., there is a signal copy of the signal to be measured in each Nyquist zone of the pre-sampling frequency comb. The pre-sampling unit can be composed of any device that meets the requirements of the pre-sampling process. Typical devices include but are not limited to various types of electro-optical modulators (corresponding to optical frequency combs), electrical mixers (corresponding to electrical frequency combs), etc.

[0094] The signal acquisition and processing unit performs analog-to-digital conversion on the pre-sampled signal output by the pre-sampling unit, and performs fast Fourier transform on the pre-sampled signal in the digital domain formed by the analog-to-digital conversion, thereby obtaining frequency measurement values ​​of multiple signal replicas. Thereafter, the signal acquisition and processing unit selects a preset number of Nyquist zones from the Nyquist zones of the pre-sampling frequency comb, calculates the frequency measurement value of the signal replica for the signal to be measured using the frequency measurement value of the signal replica for each selected Nyquist zone, and finally determines the average value of the frequency measurement value of the signal replica for the signal to be measured in each selected Nyquist zone as the frequency estimation value of the signal to be measured, thereby achieving accurate estimation of the frequency of the signal to be measured.

[0095] In practical applications, after obtaining the signal replicas in each Nyquist zone through FFT, the signal acquisition and processing unit first needs to select N (i.e., the preset number) of signal replicas for subsequent calculations. It can be seen from the above formulas 5-1 to 5-4 that the frequency of each signal replica can be expressed as a linear combination expression of the frequency of the signal to be measured and the repetition frequency. In the above selection process, the frequencies of any two selected Nyquist zone signal replicas need to satisfy the following: the difference between the coefficients of the repetition frequency in the linear combination expression is not equal to an integer multiple of the above preset number N. Only in this way can the highest measurement accuracy be obtained, otherwise the best effect cannot be achieved. In addition, as mentioned above, the repetition frequency f c The frequency resolution f of the fast Fourier transform res The decimal part ε of the ratio satisfies: equal to the reciprocal of the preset number N.

[0096] Thus, after selecting N signal replicas, the above formulas 7-1 to 7-4 can be used to measure the value f according to the frequency of the signal replicas. M，i Calculate the frequency measurement value f of the signal replica inM，i ,Right now:

[0097] At the frequency f of the signal to be measured in When in the 2n+1th Nyquist zone:

[0098] When i is an odd number,

[0099] When i is an even number,

[0100] In f in When in the 2n+2nd Nyquist zone:

[0101] When i is an odd number,

[0102] When i is an even number,

[0103] Finally, the frequency estimation value of the signal to be measured can be calculated according to the above formula 9, that is, the average value of the frequency measurement value of the signal replica of each selected Nyquist zone of the signal to be measured is calculated and determined as the frequency estimation value of the signal to be measured.

[0104] Optionally, in practical applications, the frequency estimation device of the embodiment of the present invention may further include a first signal conditioning unit and a second signal conditioning unit. The first signal conditioning unit is connected between the input port of the signal to be measured and the pre-sampling unit, and can perform signal processing such as filtering, amplification, and attenuation on the signal to be measured, so as to facilitate the signal processing requirements of subsequent units of the system. The second signal conditioning unit is connected between the pre-sampling unit and the signal acquisition and processing unit, and its function is similar to that of the first signal conditioning unit, and is used to perform signal processing such as filtering, amplification, and attenuation on the pre-sampling signal output by the pre-sampling unit, so as to facilitate the acquisition and processing requirements of the signal acquisition and processing unit.

[0105] In specific applications, the frequency estimation device of the embodiment of the present invention may further include a control unit for overall control of each unit of the system, mainly including: setting and controlling the frequency comb repetition frequency of the frequency comb generating unit, the filtering, amplification, and attenuation parameters of the first signal conditioning unit and the second signal conditioning unit, the parameters of each device of the pre-sampling unit, the sampling rate of the signal acquisition and processing unit, the number of FFT calculation points, and the number of copies required for high-precision frequency estimation (i.e., the above-mentioned preset number) and other related parameters. Figure 1 In the figure, the solid line represents the analog link and the dotted line represents the control link. It can be seen that the control unit is connected to the frequency comb generating unit, the pre-sampling unit, the signal acquisition and processing unit, the first signal conditioning unit and the second signal conditioning unit through the control link.

[0106] Figure 5 is a schematic diagram of the structure of the first signal conditioning unit according to an embodiment of the present invention, such as Figure 5 As shown, the first signal conditioning unit may include a first electrical filter, a first electrical amplifier and a first electrical attenuator, which are used to filter, amplify and attenuate the signal to be measured, respectively. The first signal conditioning unit also includes a signal input port and a signal output port. The processed signal to be measured is transmitted from the signal output port to the pre-sampling unit. The second signal conditioning unit has a similar structure, which includes a second electrical filter, a second electrical amplifier and a second electrical attenuator, which are used to process the pre-sampled signal and transmit the processed pre-sampled signal to the signal acquisition and processing unit. The specific composition, parameter selection and connection relationship of the first signal conditioning unit and the second signal conditioning unit can be adjusted according to actual needs.

[0107] In an embodiment of the present invention, the above frequency comb generation unit can use any frequency comb generation method that can generate a repetition frequency and meet the pre-sampling requirements. Exemplarily, the frequency comb has a first structure, a second structure, or a third structure. In the first structure, the frequency comb generation unit is an optical frequency comb generation device, including a mode-locked laser or an electro-optical modulated optical frequency comb generation device, and the generated pre-sampling frequency comb is an optical frequency comb.

[0108] In the second structure, the frequency comb generating unit is an electrical frequency comb generating device, and the pre-sampling frequency comb generated is an electrical frequency comb. The electrical frequency comb generating device includes a nonlinear transmission line or a step recovery diode. These devices can generate a pre-sampling electrical frequency comb with the required repetition frequency under the stimulation of a high-power point frequency signal with the required frequency comb repetition frequency.

[0109] In the third structure, the frequency comb generating unit includes the above optical frequency comb generating device and the first photodetector, and its structure is as follows: Figure 6 As shown, the generated frequency comb is an electrical frequency comb signal. The optical frequency comb generating device is used to generate an optical frequency comb, and the first photodetector is used to convert the optical frequency comb into an electrical frequency comb. The above optical frequency comb generating device can use a mode-locked laser or an electro-optical modulated optical frequency comb generating device, and the first photodetector can use any photodetector that can convert an optical frequency comb signal into an electrical frequency comb signal.

[0110] Preferably, when the frequency comb generating unit is of the first structure, the pre-sampling unit may include: an electro-optic modulator, an optical amplifier, an optical filter and a second photodetector, and may also include an optical frequency comb input port and a signal output port, such as Figure 7 As shown. Among them, the electro-optic modulator is a required device, which is used to pre-sample the input signal to be tested, that is, to modulate the signal to be tested to an optical frequency comb. The electro-optic modulator can adopt any type of electro-optic modulator that can up-convert the signal to be tested to a pre-sampled optical frequency comb, including but not limited to intensity modulators, polarization modulators, phase modulators, etc. The optical amplifier and optical filter are optional devices, which are used for signal amplification and filtering respectively. The connection relationship between the two devices can be adjusted according to actual needs. Whether to use, which type of device to use and which connection relationship to use depends on the quality requirements of the pre-sampled signal output by the pre-sampling unit during the system design process. The second photodetector is required, and any photodetector that can meet the requirements of down-converting the pre-sampled signal from the optical domain to the electrical domain can be adopted.

[0111] When the frequency comb generating unit is of the second structure or the third structure, the pre-sampling unit includes an electrical mixer for pre-sampling the signal to be measured. Figure 8As shown, the pre-sampling unit includes an electric mixer, a signal input port, an electric frequency comb input port and a signal output port. The function of the signal input port is to receive the signal to be measured, the function of the electric frequency comb input port is to receive the pre-sampled electric frequency comb, the function of the signal output port is to output the pre-sampled signal, and the function of the electric mixer is to modulate the signal to be measured onto the electric frequency comb. The electric mixer can be any electronic device that meets the above modulation requirements. If necessary, an electric amplifier and an electric filter can be added between the electric frequency comb input port and the electric mixer, in order to achieve the regulation of the power of the electric frequency comb and the number of comb teeth, so as to facilitate the pre-sampling of the signal to be measured.

[0112] like Fig. 9 As shown, the signal acquisition and processing unit of the embodiment of the present invention may include: an analog-to-digital converter and a signal processing module, and also includes a signal input port and a data output port. Among them, the signal input port is used to input a pre-sampled signal, the data output port is used to output the calculated frequency estimation value of the signal to be measured, the analog-to-digital converter is used to perform the aforementioned analog-to-digital conversion of the pre-sampled signal, and the signal processing module is used to determine the Nyquist zone number where the frequency of the signal to be measured is located, perform FFT, and calculate the frequency estimation value of the signal to be measured. The signal processing module has the following four functions.

[0113] Rough frequency estimation: Roughly judge the frequency of the signal to be measured to determine in which Nyquist zone of the pre-sampling frequency comb the signal to be measured is located, that is, to determine the specific value of n and f in the above formulas 7-1 to 7-4. in Is it located in the 2n+1th Nyquist zone or the 2n+2th Nyquist zone? In this case, the subsequent calculations can be performed using formulas 7-1 to 7-4. In practical applications, a FFT transformation with fewer points (e.g., 128 points) can be performed on the signal to be tested, or a spectrum analyzer or other instrument can be used to achieve a rough frequency estimate.

[0114] FFT: Computes the FFT power spectrum of the presampled signal.

[0115] Multi-Signal Frequency Estimation: Calculates frequency measurements of signal replicas in multiple Nyquist zones of the pre-sampled signal power spectrum and the frequency measurements of the signal replicas on the signal to be measured.

[0116] Frequency Averaging: Calculates the average of the frequency measurements of the signal replicas on the signal under test and determines it as the frequency estimate of the signal under test.

[0117] As a preferred solution, in addition to the above four functions, the signal processing module can also add other sub-functions, such as digital down-conversion, downsampling, resampling, etc. In particular, the signal processing module can be further used to: after calculating the frequency estimation value of the signal to be measured, use a preset frequency estimation algorithm to perform further calculations based on the calculated frequency estimation value of the signal to be measured to further improve the frequency estimation accuracy. The above frequency estimation algorithm includes at least one of the following: polynomial fitting, zero padding, interpolated fast Fourier transform, weighted phase averaging, and band-selected Fourier transform.

[0118] In an embodiment of the present invention, the frequency estimation device may further include a control unit, which is respectively connected to the control units of the frequency comb generation unit, the pre-sampling unit, the signal acquisition and processing unit, the first signal conditioning unit, and the second signal conditioning unit, and is used to control the system as a whole. Specifically, it can be used to set the repetition frequency and power of the pre-sampling frequency comb, set the parameters of the first electric filter, the first electric amplifier, the first electric attenuator, the second electric filter, the second electric amplifier, and the second electric attenuator, set the parameters of each device in the pre-sampling unit, set the sampling rate of the analog-to-digital conversion, the number of calculation points of the fast Fourier transform, and the above-mentioned preset number, and provide a reference clock frequency for the frequency comb generation unit and the signal acquisition and processing unit.

[0119] In summary, in the technical solution of the embodiment of the present invention, a high-precision frequency estimation device based on pre-sampling and multi-measurement value averaging is proposed, that is, a frequency comb with a specific repetition frequency is used to pre-sample the signal in the analog domain, and the signal to be measured is extended to multiple pre-sampled Nyquist zones. Then, in the digital domain, the frequency measurement values ​​of the signal replicas in multiple Nyquist zones in the FFT spectrum of the pre-sampled signal are extracted, and a simple averaging algorithm is used to achieve a significant improvement in the accuracy of signal frequency estimation. There are two key features:

[0120] First, a frequency comb with a specific repetition frequency is used to pre-sample the signal to be measured. The repetition frequency of the pre-sampling frequency comb preferably satisfies the following relationship, that is, according to the system FFT spectrum resolution f res , set a specific pre-sampling repetition frequency value so that the fractional part ε of the repetition frequency satisfies ε×N=1, where N is the number of signal copies in the FFT spectrum of the pre-sampling signal used to calculate the frequency of the signal to be measured (i.e. the above preset number).

[0121] Second, frequency measurement values ​​of signal replicas located in a plurality of pre-sampled Nyquist zones in the FFT spectrum of the pre-sampled signal are extracted, and the frequency estimation value of the signal to be measured is obtained by averaging.

[0122] By pre-sampling the signal to be measured in hardware and using a simple signal processing method based on FFT spectrum data, super-resolution measurement of the signal frequency is achieved, solving the main problems in improving the frequency estimation accuracy at the hardware and algorithm levels. The pre-sampling process and processing algorithm of the device of the present invention are very simple and reliable, without the need to introduce other complex algorithms. In addition, as a representative signal processing method, the device of the present invention is also compatible with other existing high-precision frequency estimation algorithms based on FFT.

[0123] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention 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. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A frequency estimation device, characterized in that: include: A frequency comb generating unit, a pre-sampling unit and a signal acquisition and processing unit; wherein, The frequency comb generating unit is used to generate a pre-sampling frequency comb; The pre-sampling unit pre-samples the signal to be tested in the analog domain using the pre-sampling frequency comb to obtain a pre-sampled signal; the pre-sampling signal includes: a plurality of signal copies of the signal to be tested located in each Nyquist zone of the pre-sampling frequency comb; The signal acquisition and processing unit performs analog-to-digital conversion on the pre-sampled signal, performs fast Fourier transform on the pre-sampled signal in the digital domain formed by the analog-to-digital conversion, and obtains frequency measurement values ​​of the multiple signal copies; The signal acquisition and processing unit selects a preset number of Nyquist zones from the Nyquist zone, calculates the frequency measurement value of the signal replica for the signal to be measured using the frequency measurement value of the signal replica in each selected Nyquist zone, and determines the average value of the frequency measurement value of the signal replica in each selected Nyquist zone for the signal to be measured as the frequency estimation value of the signal to be measured; The signal acquisition and processing unit calculates the frequency measurement value of the signal to be measured by the signal replica of any selected Nyquist zone through the following formula: At the frequency f of the signal to be measured in When in the 2n+1th Nyquist zone: When i is an odd number, When i is an even number, In f in When in the 2n+2nd Nyquist zone: When i is an odd number, When i is an even number, Where n is a non-negative integer, i is the serial number of any selected Nyquist zone, and f inM，i is the frequency measurement value of the signal to be measured by the signal replica of any selected Nyquist zone, f M，i is the frequency measurement value of the signal replica in any selected Nyquist zone, f c is the repetition frequency of the pre-sampling frequency comb.

2. The device according to claim 1, characterized in that The preset number is equal to the number of signal copies used to calculate the frequency estimation value of the signal to be measured; the frequency of each signal copy can be expressed as a linear combination expression of the frequency of the signal to be measured and the repetition frequency; and, The frequencies of any two signal replicas in the Nyquist zone selected by the signal acquisition and processing unit satisfy: the difference between the coefficients of the repetition frequency in the linear combination expression is not equal to an integer multiple of the preset number.

3. The device according to claim 1, characterized in that The decimal part of the ratio of the repetition frequency to the frequency resolution of the fast Fourier transform satisfies: being equal to the reciprocal of the preset number.

4. The device according to claim 1, characterized in that The signal acquisition and processing unit includes: an analog-to-digital converter and a signal processing module; wherein, The analog-to-digital converter is used to perform the analog-to-digital conversion; The signal processing module is used to determine the Nyquist zone number where the frequency of the signal to be measured is located, perform the fast Fourier transform, and calculate the frequency measurement value of the signal to be measured and the frequency estimation value of the signal to be measured for each selected signal replica of the Nyquist zone.

5. The device according to claim 1, characterized in that The frequency comb generating unit has a first structure, a second structure or a third structure; In the first structure, the frequency comb generating unit is an optical frequency comb generating device, including a mode-locked laser or an electro-optical modulated optical frequency comb generating device, and the pre-sampling frequency comb is an optical frequency comb; In the second structure, the frequency comb generating unit is an electrical frequency comb generating device, including a nonlinear transmission line or a step recovery diode, and the pre-sampling frequency comb is an electrical frequency comb; In a third structure, the frequency comb generating unit includes the optical frequency comb generating device and a first photodetector, and the pre-sampling frequency comb is an electrical frequency comb.

6. The device according to claim 5, characterized in that When the frequency comb generating unit is of the first structure, the pre-sampling unit comprises: an electro-optic modulator, an optical amplifier, an optical filter and a second photodetector; wherein the electro-optic modulator pre-samples the signal to be measured, and the optical amplifier, the optical filter and the second photodetector are used to process the pre-sampled signal; When the frequency comb generating unit is of the second structure or the third structure, the pre-sampling unit includes an electrical mixer for pre-sampling the signal to be measured.

7. The device according to claim 1, characterized in that The device further comprises a first signal conditioning unit connected to the pre-sampling unit, and a second signal conditioning unit connected to the pre-sampling unit and the signal acquisition and processing unit respectively; wherein, The first signal conditioning unit comprises a first electrical filter, a first electrical amplifier and a first electrical attenuator, and is used to process the signal to be measured, and transmit the processed signal to be measured to the pre-sampling unit; The second signal conditioning unit includes a second electrical filter, a second electrical amplifier and a second electrical attenuator, and is used to process the pre-sampled signal and transmit the processed pre-sampled signal to the signal acquisition and processing unit.

8. The device according to claim 1, characterized in that The signal acquisition and processing unit is further used to: after calculating the frequency estimation value of the signal to be measured, use a preset frequency estimation algorithm to perform further calculation based on the calculated frequency estimation value of the signal to be measured; wherein, The frequency estimation algorithm includes at least one of the following: polynomial fitting, zero padding, interpolation fast Fourier transform, weighted phase averaging, and band-selective Fourier transform.

9. The device according to claim 1, characterized in that The device further comprises: a control unit connected to the frequency comb generating unit, the pre-sampling unit, the signal acquisition and processing unit, the first signal conditioning unit and the second signal conditioning unit respectively; wherein, The control unit is used to: set the repetition frequency and power of the pre-sampling frequency comb, set the parameters of the first electrical filter, the first electrical amplifier, the first electrical attenuator, the second electrical filter, the second electrical amplifier and the second electrical attenuator, set the parameters of each component in the pre-sampling unit, set the sampling rate of the analog-to-digital conversion, the number of calculation points of the fast Fourier transform and the preset number, and provide a reference clock frequency for the frequency comb generating unit and the signal acquisition and processing unit.

Images