A multi-channel ultra-wideband under-sampling instantaneous frequency measurement method

Abstract

The application relates to a multi-channel ultra-wideband undersampling instantaneous frequency measurement method, belonging to the technical field of wideband digital radars, and solving the problem of spectrum overlap of multiple signals; for a scheme of multiple channels with different time delays, phase frequency measurement is combined with the remainder theorem to demodulate the frequency of an input signal; for signals near f s / 2, phase frequency measurement is carried out through the time delay difference between different channels to solve the frequency band in a targeted manner; for signals near f s / 2, multi-rate sampling is realized through a time-sharing sampling method, threshold discrimination and parameter matching are carried out on the FFT frequency measurement result through the remainder theorem, and finally, the PDW signal is demodulated, the method does not need to provide a high-speed pseudo-random sequence, has hardware realizability, has the characteristics of small size, no need of channelization, low algorithm implementation complexity, strong robustness and strong demodulation capability compared with the prior art, and can realize high-precision undersampling frequency estimation of radar signals under the index of 0 to 18G instantaneous bandwidth.

Description

Technical Field

[0001] This invention belongs to the field of broadband digital radar technology and relates to a multi-channel ultra-wideband undersampling instantaneous frequency measurement method. Background Technology

[0002] Electronic reconnaissance requires acquiring various useful information from complex electromagnetic environments, and is currently developing towards higher performance targets such as ultra-wideband, complex multi-signal processing, high sensitivity, large dynamic range, and miniaturization. With the improvement of the performance of digital chips such as ADCs, FPGAs, and DSPs, and the rapid development of signal processing technology, using digital devices to achieve broadband signal sampling and processing offers better stability, higher frequency resolution, and smaller equipment size compared to analog receivers. For ultra-wideband digital receivers, direct RF sampling can effectively solve the complex system structure problems caused by the need for a large number of RF components such as frequency converters, amplifiers, and filters, as well as local oscillator clocks in traditional receivers. However, direct RF sampling not only poses challenges to the sampling rate and analog input bandwidth requirements of the ADC chip, but also generates a huge amount of data from direct Nyquist sampling, posing difficulties for the back-end data processing module and hardware implementation. Therefore, how to acquire broadband signals using low-speed ADCs has become a research hotspot.

[0003] Undersampling real-time frequency measurement algorithms can achieve frequency measurement of broadband signals under conditions of under-Nyquist sampling rates, thus overcoming the sampling rate bottleneck problem. The purpose of undersampling algorithms is to de-ambiguate the folded spectrum and obtain information from the original spectrum. Many algorithms have been proposed, such as undersampling algorithms that use the phase difference between the peak values ​​of the spectrum before and after the delay to calculate the signal frequency. A single undersampling channel can de-ambiguate, but single-channel delay has poor noise immunity. s / 2 spectral overlap can lead to blind zones and inaccurate measurements. The MUSIC algorithm can resolve ambiguity through multi-channel delay and address the problem of multiple signal overlap, but it requires a greater number of delay channels than the number of signals and necessitates search operations, making real-time processing difficult. Compressed sampling-based algorithms include structures such as random sampling (PRS), random demodulator (RD), and modulation-bandwidth converter (MWC), requiring a pseudo-random sequence at the Nyquist rate for modulation, placing high demands on hardware. For bandwidths of 0-18 GHz, existing technologies struggle to achieve high-speed pseudo-random sequences that meet the requirements. Using the remainder theorem to solve frequency ambiguity in undersampled signals at different sampling rates requires coprime sampling rates. This method offers strong real-time performance, but parameter pairing errors can occur when the amplitudes and phases of two signals are very close in the measurement of multiple signals. Therefore, current conventional algorithms all have certain limitations for the 0-18 GHz range. Summary of the Invention

[0004] The technical problem to be solved by this invention is how to perform direct radio frequency sampling and deambiguation of ultra-wideband (0 to 18G) under low sampling rate (2G to 3G) hardware conditions, thereby solving the problem of multi-signal spectrum overlap.

[0005] The present invention solves the above-mentioned technical problems through the following technical solutions:

[0006] A multi-channel ultra-wideband undersampling instantaneous frequency measurement method includes the following steps:

[0007] S1. Select the sampling channel and set the delay parameters according to the required bandwidth. Select the ADC chip and signal processing chip that meet the sampling rate requirements. Design the board and build the hardware platform according to the design requirements of each chip.

[0008] S2. After the signal enters multiple sampling channels, it passes through the sample-and-hold and ADC chip, and then the signal processing chip obtains the undersampled and aliased spectrum. First, the spectrum peak is found, and then the parameter matching and phase calculation are performed in combination with the remainder theorem, so as to de-ambiguate and obtain the frequency information.

[0009] Spectrum peak finding: The Candan estimation algorithm is used to perform frequency and phase correction on the peak values ​​of each FFT spectrum, and the remainder information and phase information are extracted for remainder filtering;

[0010] Defuzzing includes parameter pairing and phase calculation: the remainder information f obtained after peak retrieval corresponds to the actual signal frequency nf. s +f, combining the remainder theorem and phase difference methods to find the number of folds n;

[0011] The phase calculation is based on the delay of the sampling channels in the hardware system. In a multi-channel system, the delay of each channel is different. For example, the delay difference between channel 1 and channel 2 is Δt1, the delay difference between channel 2 and channel 3 is Δt2, and the delay difference between channel 1 and channel 3 is Δt1-Δt2 (Δt1>Δt2). Channels with different delay differences are used to solve for different frequency band signals.

[0012] The parameter pairing method is as follows: input p signals, each ADC outputs 2p remainders, so n ADCs will have n(2p) remainders. 2 The results of group matching are as follows: only p groups are correct, and the rest are false frequencies of incorrect pairings. The correct matching results are solved by the remainder theorem.

[0013] The phase calculation method is as follows: the phase difference is determined by both frequency and delay, and its calculation formula is θ = 2πfΔt, where θ is the phase difference and Δt is the delay; therefore, as long as the error of the measured phase difference does not exceed 45°, the actual phase difference value can be determined, thereby obtaining the frequency value;

[0014] S3. Pack the frequency information obtained by defuzzification processing in the signal processing chip and output it to the host computer or signal processing module.

[0015] The method of this invention can simultaneously solve parameter matching and nf s The problem of the / 2 spectrum blind zone is solved, and high-speed pseudo-random sequences are not required, making it hardware-feasible; for schemes with multiple channels having different delays, phase frequency measurement combined with the remainder theorem is used to de-ambiguate the input signal frequency; for non-f s For signals near / 2, phase frequency measurement can be performed using the delay difference between different channels, allowing for targeted solutions across specific frequency bands. For f... s Signals near / 2 are sampled at multiple rates using a time-division sampling method. The remainder theorem is used to perform over-threshold discrimination and parameter matching on the FFT frequency measurement results, ultimately deblurring the signal and emitting the PDW signal. The deblurring process in this invention involves applying two algorithms to signals in different frequency bands. s For frequencies near the / 2 spectrum, the remainder theorem is used as the primary reference for solving the problem. For other frequency bands and multiple signals, the delay phase calculation is used as the primary reference. The different delays for each channel are calculated for signals in different frequency bands, which is a frequency band adaptive algorithm. The delay phase defuzzification algorithm requires at least three delay channels to meet the bandwidth of 0-18G, ensuring that the delays between each pair of channels are different. This reduces the defuzzification errors caused by the influence of temperature on the delay in actual use. For high-frequency signals, channels with small delay differences are selected for solving the problem, while channels with large delay differences are selected for solving the problem for low-frequency signals. Compared with the prior art, this invention has the characteristics of small size, no need for channelization, low algorithm implementation complexity, strong robustness, and strong defuzzification capability. It performs segmented undersampling frequency estimation from 0 to 18G with higher accuracy.

[0016] Further, the hardware platform construction method described in step S1 is as follows: After the 0 to 18G radio frequency signal is subjected to simulated quadrature mixing, the power is divided into four signals, each pair being 90° out of phase. In order to ensure a fixed delay, the delay line part is completed in the simulated mixing part. The two channels of each power divider have inter-channel delays, and the delays are different. After simulated mixing, an ultra-wideband sample-and-hold chip is used to sample the radio frequency signal. The sampled signal is then sampled by four synchronous ADCs and the data enters the signal processing chip for frequency measurement processing. Finally, the PDW signal is sent to the host computer.

[0017] Furthermore, the ADC chip and signal processing chip mentioned in step S1 are Zynq UltraScale+RFSoC single-chip ICs from Xilinx products.

[0018] Furthermore, the calculation formula for the Candan estimation algorithm described in step S2 is as follows:

[0019]

[0020] Where N is the number of sampling points, Km is the peak position of the amplitude spectrum, X[Km] is the peak frequency, X[Km-1] and X[Km+1] are the frequencies corresponding to the points before and after the peak position, respectively, and Re{} is the operation of taking the real part.

[0021] Furthermore, the correct matching result described in step S2, obtained using the remainder theorem, is based on a time-division sampling method. While the sampling rate of multiple channels in the hardware system is consistent at each moment, the consistent sampling rate of the ADC varies over time, thus achieving multi-rate sampling. The sampling rate for the 0-T1 time period is f. s1 The sampling rate for the T1-T2 time period is f s2 The sampling rate for the T2-T3 time period is f s3 Similarly, parameter matching is performed based on the remainder theorem to calculate the number of frequency folds at different sampling rates, thereby deblurring the data.

[0022] Furthermore, the frequency point information obtained by the defuzzification process in step S3 is packaged and output to the host computer or signal processing module using a standard serial port, wireless network, high-speed serial interface or USB interface.

[0023] The advantages of this invention are:

[0024] The method of this invention can simultaneously solve the problems of parameter matching and undersampling nf. s The problem of the spectral dead zone near the / 2 frequency is solved, and it does not require a high-speed pseudo-random sequence, making it hardware-feasible; for schemes with multiple channels having different delays, phase frequency measurement combined with the remainder theorem is used to de-ambiguate the input signal frequency; for non-f s For signals near / 2, phase frequency measurement can be performed using the delay difference between different channels, allowing for targeted solutions across specific frequency bands. For f... s Signals near / 2 are sampled at multiple rates using a time-division sampling method. The remainder theorem is used to perform over-threshold discrimination and parameter matching on the FFT frequency measurement results, ultimately de-ambiguousing the signal before PDW signal transmission. This scheme offers high frequency measurement accuracy and low computational cost. The multi-channel delay also reduces the impact of temperature on delay accuracy, thus mitigating the de-ambiguity issue. Compared to existing technologies, this invention features small size, no need for channelization, low algorithm complexity, strong robustness, and strong de-ambiguity capability. It achieves higher accuracy for segmented undersampling frequency estimation from 0 to 18 GHz; and for signal frequencies close to f... s The phase ambiguity present at integer multiples of 2 also has a good deblinding effect, realizing high-precision frequency estimation of broadband signals. Attached Figure Description

[0025] Figure 1 This is a block diagram of the 0 to 18G bandwidth-based four-channel multi-rate undersampling system design of the present invention;

[0026] Figure 2 This is a block diagram of the 0 to 18G bandwidth-based four-channel delay undersampling system design of the present invention;

[0027] Figure 3 It is a frequency close to nf s / 2 is a blind zone formed due to spectral symmetry. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments:

[0030] Example 1

[0031] like Figure 1 The diagram shown is a block diagram of the 0-18G bandwidth four-channel delay undersampling system design of the present invention. The specific design is as follows:

[0032] Multi-channel delay settings: Delay parameters T1 and T2 are set separately, with a difference between the two delay times. Channels 1 and 2 have a delay of T1, channels 3 and 4 have a delay of T2, channels 2 and 4 have a delay of T2-T1, and channels 1 and 3 have a delay of 0. Initial simulations verified that smaller delays are more suitable for solving high-frequency signals, while larger delays are more suitable for solving low-frequency signals. Therefore, based on the delay difference between channels, signals of different frequency bands can be solved accordingly. For example, if T1 is set to 100ps, its unambiguous bandwidth is 1 / T1 = 10G; if T2 is set to 150ps, its unambiguous bandwidth is 1 / T2 = 6.7G; the unambiguous bandwidth corresponding to T2-T1 is 20G. This satisfies the unambiguity requirements of ultra-wideband signals. Furthermore, multiple channels are simultaneously verified for solving mid- and low-frequency signals, resulting in more reliable results. For channels 1 and 3, the signals are synchronized and 90° out of phase, allowing for the solution of IQ signals and expanding the range of bandwidth available for the solved signals.

[0033] Four-channel ADC time-division sampling: To address the "blind zone" problem, sampling is performed near the NF in each channel. s There will be a blind spot at each of the two locations, such as Figure 3As shown, the blind zone causes the phase spectrum peak to change due to spectral overlap, so signal deblurring in the blind zone will be incorrect. In this method, the sampling rate of the ADC is divided into two groups. At a certain moment, the sampling rate is f1, and in the next moment, the sampling rate is f2, achieving time-division multiplexing of 4 channels of ADCs to supplement the measurement of the frequency points in the "blind zone". Since the sampling rates of the two are different, when the input signal frequency is close to nf s / 2 of one channel, it will surely be far from nf s / 2 of the other channel. Therefore, in the outputs of the two channels, there is always one frequency reading that is correct, thus solving the problem of signal deblurring in the blind zone.

[0034] Hardware system: For RF signals from 0 to 18G, after passing through the analog quadrature mixing module, the signal is power-divided into four signals, with a 90° phase difference between each pair. A ultra-wideband sampling and holding chip is used to capture the RF signal. To ensure a fixed delay, the delay line part is completed in the analog mixing module part. After the sampled and held signal is sampled by four-channel synchronous ADCs, the sampled data enters the FPGA for corresponding frequency measurement algorithm processing, and finally the PDW signal of the signal is sent out. This hardware system can simultaneously meet the hardware implementation of the delay undersampling system and the multi-rate undersampling system.

[0035] As Figure 2 shown, it is the design block diagram of the 0 to 18G bandwidth four-channel multi-rate undersampling system of the present invention, and the specific design is as follows:

[0036] Multi-rate AD sampling setting: Multi-rate sampling means selecting three ADCs with different sampling rates for sampling. The sampling frequency of each ADC is less than the Nyquist sampling frequency of the signal. Although the spectrum will be folded after sampling, the sampled signal and the actual spectrum only differ by an integer multiple of the sampling frequency, so it can be solved by the remainder theorem. The sampling frequencies of multiple channels of ADCs need to be relatively prime, and the signal frequency f < lcm(f1, f2, f3, f4). For example, sampling rates of 2.21G or 1.92G or – 2.09G or 2.03G can solve the spectrum up to 18G. Multiple channels can reduce the probability of deblurring matching errors compared to two channels.

[0037] Hardware system: The same hardware system as the four-channel delay undersampling system can be selected, which will not be elaborated here. The functions of the delay and analog mixing parts are not required in this hardware.

[0038] The method for multi-channel ultra-wideband undersampling instantaneous frequency measurement is as follows: Step (1) Set the delay parameters according to the required bandwidth, select an ADC chip and processing chip with a suitable sampling rate, and complete the hardware platform construction. Currently, Xilinx has a Zynq UltraScale+ RFSoC single-chip IC that realizes the sampling and detection of multi-channel broadband intermediate frequency signals and other signal processing. This series of devices has an integrated ADC (8 12-bit channels, sampling rate of 4.0GSPS), configurable logic elements, multi-processor embedded ARM Cortex-A53 application processing unit (APU) and ARM real-time processing unit (RPU), which can simultaneously meet the implementation of the multi-channel undersampling system in this scheme. At the same time, in order to meet the matching of the bandwidth of the ultra-wideband RF signal and the ADC bandwidth in the hardware, an ultra-wideband sample-and-hold chip is used to capture the RF signal. The sample-and-hold amplifier (THA) HMC1061 can be selected, which has an 18G sampling bandwidth, which can fundamentally expand the bandwidth, making it far exceed the ADC sampling bandwidth, and meet the requirements of demanding high-bandwidth applications. The signal after sampling and holding can be connected to the AD interface of the RFSOC. PCB design requires the use of high-frequency boards with more stable dielectric parameters throughout the entire signal bandwidth, to ensure impedance continuity of signal traces as much as possible, and to reduce noise between the power and ground reference planes.

[0039] Step (2): Obtain the aliased spectrum after sampling, perform spectrum peak finding, parameter pairing and phase calculation, and perform corresponding defuzzification processing;

[0040] The specific defuzzing algorithm is as follows: Input p signals, each AD converter outputs 2p remainders, so 4 AD converters will have 4(2p) remainders. 2 The results of group matching are as follows: only p groups are correct, while the rest are false frequencies due to incorrect remainder pairing. Correct matching results can be solved using the remainder theorem. However, when the number of p groups is large, solving the problem using only the remainder theorem for the frequency points will increase the error rate. Therefore, it is possible to filter out incorrect matching results by combining signal amplitude and phase, which effectively reduces the possibility of finding redundant signals.

[0041] Furthermore, since FFT yields discrete frequency values, the spectrum can only be observed at a finite number of N frequency points. The frequency resolution Δf limits the estimation accuracy of the frequency remainder. This invention employs the Candan estimation algorithm to perform frequency and phase correction on the peak values ​​of each FFT spectrum, and can provide accurate remainder and phase information for remainder selection. The calculation formula is as follows:

[0042]

[0043] Where N is the number of sampling points, Km is the peak position of the amplitude spectrum, X[Km] is the peak frequency, X[Km-1] and X[Km+1] are the frequencies corresponding to the points preceding and following the peak position, respectively, and Re{} is the operation of taking the real part. This method can effectively improve the frequency estimation accuracy.

[0044] In the delay scheme, the phase information of different channels at different frequency points can be reflected in the FFT output. The ratio of the phase difference to the signal's spectrum is 2πT, so the frequency can be solved by the phase difference. Delay undersampling can handle multiple signals simultaneously, but the measured frequency points have a certain error. Therefore, this invention uses phase difference combined with the FFT result for frequency measurement. The actual frequency is nf. s +f, and then combine with the phase difference to calculate the number of folds n. As long as the error of the measured phase difference does not exceed 45°, the actual phase difference value can be determined, thus obtaining the accurate frequency value.

[0045] Step (3): Pack the defuzzified frequency point information (PDW, including frequency, amplitude, pulse width, and arrival time) and output it to the host computer or subsequent processing module via RS422 serial communication, network communication, fiber optic communication or high-speed serial interface.

[0046] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-channel ultra-wideband undersampling instantaneous frequency measurement method, characterized in that, Includes the following steps: S1. Select the sampling channel and set the delay parameters according to the required bandwidth. Select the ADC chip and signal processing chip that meet the sampling rate requirements. Design the board and build the hardware platform according to the design requirements of each chip. The hardware platform is built as follows: 0 to 18 GHz radio frequency signals are divided into four signals by analog quadrature mixing, with each pair of signals differing by 90° in phase. To ensure a fixed delay, the delay line is completed in the analog mixing section. Each power divider has two channels with different inter-channel delays. After analog mixing, an ultra-wideband sample-and-hold chip is used to sample the radio frequency signals. The sampled data from the sampled signals is then processed by a four-channel synchronous ADC and sent to a signal processing chip for frequency measurement. Finally, the PDW signal is sent to the host computer. S2. After the signal enters multiple sampling channels, it passes through the sample-and-hold and ADC chip, and then the signal processing chip obtains the undersampled and aliased spectrum. First, the spectrum peak is found, and then the parameter matching and phase calculation are performed in combination with the remainder theorem, so as to de-ambiguate and obtain the frequency information. Spectrum peak finding: The Candan estimation algorithm is used to perform frequency and phase correction on the peak values ​​of each FFT spectrum, and the remainder information and phase information are extracted for remainder filtering; Defuzzing includes parameter pairing and phase calculation: the remainder information f obtained after peak finding in the spectrum corresponds to the actual signal frequency nf. s +f, combining the remainder theorem and phase difference methods to find the number of folds n; The parameter pairing method is as follows: input p signals, each ADC outputs 2p remainders, so n ADCs will have n(2p) remainders. 2 The results of group matching show that only p groups are correct, while the rest are spurious frequencies due to remainder errors. The correct matching results are obtained using the remainder theorem. This method of obtaining the correct matching results using the remainder theorem is based on a time-division sampling method. In the hardware system, the sampling rate of multiple channels is consistent at each moment, but the ADC's consistent sampling rate varies time-divisionally, thus achieving multi-rate sampling. The sampling rate for the 0-T1 time period is f. S1 The sampling rate for the T1-T2 time period is f S2 The sampling rate for the T2-T3 time period is f S3 Similarly, parameter matching is performed based on the remainder theorem to calculate the number of frequency folds at different sampling rates, thereby deblurring the image. The phase calculation method is as follows: the phase difference is determined by both frequency and delay, and its calculation formula is θ = 2πfΔt, where θ is the phase difference and Δt is the delay; therefore, as long as the error of the measured phase difference does not exceed 45°, the actual phase difference value can be determined, thereby obtaining the frequency value; S3. Pack the frequency information obtained by defuzzification processing in the signal processing chip and output it to the host computer or signal processing module.

2. The multi-channel ultra-wideband undersampling instantaneous frequency measurement method according to claim 1, characterized in that, The ADC chip and signal processing chip mentioned in step S1 are Zynq UltraScale+RFSoC single-chip ICs from Xilinx.

3. The multi-channel ultra-wideband undersampling instantaneous frequency measurement method according to claim 2, characterized in that, The calculation formula for the Candan estimation algorithm described in step S2 is as follows: Where N is the number of sampling points, Km is the peak position of the amplitude spectrum, X[Km] is the peak frequency, X[Km-1] and X[Km+1] are the frequencies corresponding to the points before and after the peak position, respectively, and Re{} is the operation of taking the real part.

4. The multi-channel ultra-wideband undersampling instantaneous frequency measurement method according to claim 3, characterized in that, The frequency point information obtained by the defuzzification process in step S3 is packaged and output to the host computer or signal processing module using a standard serial port, wireless network, high-speed serial interface or USB interface.

Images