GNSS-R soil humidity real-time inversion method based on FPGA + DSP architecture

Abstract

The invention provides a GNSS-R soil humidity real-time inversion method based on an FPGA + DSP architecture, and belongs to the field of earth surface typical parameter inversion. The method comprises the following steps: receiving GNSS direct and reflected signals, filtering and amplifying the signals, and performing frequency mixing down-conversion on the signals and local oscillation signals; filtering high-frequency components of the down-converted signal through a low-pass filter, and carrying out ADC (Analog to Digital Converter) sampling; performing multi-phase rotation down-conversion and correlation processing on the sampled signal to obtain five paths of correlation outputs; the real-time result of the soil humidity is obtained by combining the power ratio obtained by using the related value with a Topp model, and the real-time result of the soil humidity is compared with a true value acquired by the soil hygrometer on site so as to test the reliability of the data result. By utilizing the technology, the soil humidity can be monitored in real time, the method is low in power consumption and high in operation efficiency, and a real-time feasible monitoring means is provided for agricultural development and grain production.

Description

Technical Field

[0001] The present invention relates to the technical field of typical surface parameter inversion, and in particular to a GNSS-R soil moisture real-time inversion method based on FPGA+DSP architecture. Background Art

[0002] The Global Navigation Satellite System (GNSS) was originally used only for positioning, timing and navigation. With the continuous advancement of technology in various countries, today's GNSS signals have the characteristics of all-day and all-weather, multi-frequency and multi-mode, high temporal and spatial resolution and high coverage. These characteristics have led to the emergence of Global Navigation Satellite System Reflection Remote Sensing (GNSS-R) technology, which is a new type of remote sensing method that has developed rapidly in recent years by using the direct and reflected signals of GNSS to invert and detect typical environmental parameters on the surface. Among them, soil moisture monitoring is one of the important application areas of GNSS-R technology. Soil moisture is closely related to major issues related to food production, soil drought and human life.

[0003] There are obvious bottlenecks in the current GNSS-R soil moisture inversion. The traditional method mainly relies on software receivers and tests the baseband signal processing algorithm. However, this traditional method has a serious problem of poor real-time performance. In agricultural production, real-time acquisition of accurate soil moisture data is crucial for crop planting, irrigation and other links. However, due to the lack of real-time performance of traditional methods, it is impossible to provide effective data support for agricultural production in a timely manner, and it is difficult to meet the urgent needs of agricultural production for the real-time and accuracy of soil moisture data. This greatly limits the widespread application of GNSS-R technology in food production, agricultural monitoring and other aspects. New technological breakthroughs are urgently needed to solve these problems and promote the further development of this technology in related fields. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a GNSS-R soil moisture real-time inversion method based on FPGA+DSP architecture, with the aim of providing a real-time soil moisture inversion technical means through the high-speed computing and multi-unit parallel processing capabilities of FPGA.

[0005] The present invention adopts the following scheme: A GNSS-R soil moisture real-time inversion method based on FPGA+DSP architecture includes the following steps: S1. Build a navigation satellite receiver baseband processing platform based on FPGA+DSP architecture: The platform includes: The right-handed GNSS antenna is tilted upward to receive the direct signal from the navigation satellite; the left-handed GNSS antenna is tilted downward to receive the reflected signal from the navigation satellite; The GNSS antenna is connected to the FPGA+DSP real-time processing device via a radio frequency line for processing; The soil moisture meter collects soil moisture values ​​at multiple locations on site and takes the average value as the true value for comparison; S2, fast capture of navigation satellite signals based on FPGA+DSP; The fast capture module searches for signals based on the segmented correlation accumulation in the domain when the master clock is running and the DFT transform in the frequency domain. The module is divided into 22 channels for parallel code phase search, and each channel shares the real-time cached data. S3, channel baseband signal processing; By performing complex phase rotation down-conversion and narrow correlation operation on the intermediate frequency signal, five-way correlation value output is obtained, and then the real-time soil moisture value is obtained through the Topp model. It is compared with the true value collected on-site by the soil moisture meter to verify the reliability of the data results.

[0006] Furthermore, in step S1, the FPGA+DSP real-time processing method is: The navigation satellite antenna receives the GNSS antenna signal, which first passes through a bandpass filter to filter out signals in other frequency bands except the navigation signal band to eliminate signal interference; The filtered signal is amplified by a low noise amplifier for further processing; The frequency synthesizer generates a local oscillation signal based on the local oscillator and amplifier of the device, and mixes and down-converts the signal with the GNSS signal through the mixer; The down-converted signal is passed through an amplifier and a low-pass filter to remove the high-frequency components in the signal, and is sampled by an automatic gain control and an analog-to-digital converter and input into the FPGA+DSP for processing.

[0007] Furthermore, the specific method for fast capture of navigation satellite signals based on FPGA+DSP in step S2 is as follows: (1) Call the code NCO generation module to generate a pseudo code base frequency rate clock enable signal, that is, a 2.046 MHz clock enable signal; (2) The input 60M data rate zero intermediate frequency baseband signal is accumulated and down-converted through the enable signal to generate a down-converted baseband signal with a data rate of 2.046M; (3) Store the decelerated baseband signal in real time in a dual-port RAM with a storage depth of 6138 points, or 3 ms; (4) Generate 22-channel local pseudo-code sequence and output it at the master clock rate; (5) Read 4096 points of data from the dual-port RAM and align them with the local pseudo-code sequence and perform correlation accumulation. One point is accumulated every 64 points, and a total of 64 points of correlation accumulation values ​​are obtained; (6) Driven by the master clock, a 64-point DFT operation is performed on each accumulated value, and the result is stored in a 64-point dual-port RAM. When the second accumulated value is obtained, a second 64-point DFT operation is performed on it, and the result is added to the first result. After the 64th DFT operation is completed, a complete 64-point DFT result is obtained. (7) Calculate DFT peak energy and average energy in real time and buffer them into registers; (8) Re-read the dual-port RAM data and align it with the local pseudo code sequence with a delay of one data point and perform correlation accumulation to obtain a 64-point DFT result. Calculate the DFT peak energy and average energy. Repeat 2046 times and record the one with the largest peak energy among the 64-point DFTs. You can then obtain the pseudo code initial phase, the frequency point with the maximum DFT peak energy, and the peak energy and average energy information. (9) Generate a capture completion flag. When the DSP finds that this flag is valid, it reads the capture result and makes judgments and processes it.

[0008] Furthermore, the steps of channel baseband signal processing are as follows: (1) Driven by the 60MHz system master clock input, the carrier NCO is called to generate a local 1561.098MHz oscillation signal, which is mixed with the input orthogonal signal to complete digital down-conversion; (2) When performing narrow correlation, a 1 / 4 chip time interval is selected for correlation operation, that is, five pseudo-code sequences of advance, advance instant, instant, instant lag, and lag are generated locally, and five outputs are output after correlation operation; (3) When processing the reflected signal of the navigation satellite, the two channels with the highest signal-to-noise ratio (SNR) in the direct channel are selected for open-loop tracking processing. In order to prevent frequent satellite changes, it is stipulated that the satellite will be changed only when the third satellite is 3 dB greater than any one of the first two satellites. The Doppler frequency shift is 15 channels, which are non-intervally distributed, namely 0 Hz, ±25 Hz, ±50 Hz, ±100 Hz, ±200 Hz, ±300 Hz, ±400 Hz and ±500 Hz. The 60 MHz master clock is driven to reduce the speed to obtain a 4 MHz rate clock enable signal, and then the selector selects 15 data streams in turn to pass through the accumulator operation and cache the accumulated results accordingly. After the next polling clock arrives, the previously cached results are called out and participate in the accumulation operation of the corresponding channel; (4) The Topp model is used to invert the soil moisture of the relevant results. Based on the reflectivity obtained from the relevant output, the soil dielectric constant can be obtained as: ; in, is the soil dielectric constant, is the satellite altitude angle, is the power ratio, and the real-time soil moisture is further obtained through the Topp model ( )data: .

[0009] Compared with the prior art, the present invention has the following beneficial effects: (1) The soil moisture monitoring method designed by the present invention is based on the FPGA+DSP architecture and uses the GNSS satellite as the signal source, which can realize real-time soil moisture monitoring, and the obtained results have a good correlation with the real results; (2) The FPGA+DSP architecture receiver in the present invention only serves as a signal receiving source rather than a signal transmitting source, and has the advantages of low power consumption and low cost. Due to the programmability of FPGA, the present invention can achieve high computing speed while retaining the flexibility of the device, so that the present technology can be more widely used in various soil moisture monitoring scenarios. (3) The present invention makes up for the shortcoming of the time lag in the traditional use of ground-based GNSS-R to invert soil moisture, which can promote the application of GNSS-R technology and play an important role in agricultural monitoring, food production, soil drought monitoring and other fields. The present invention has high operating efficiency and low power consumption, and due to the flexibility of FPGA, it can be further modified according to different application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a real-time system based on FPGA+DSP architecture of the present invention; Figure 2 This is the baseband signal processing flow in the present invention; Figure 3 is a soil moisture result measured by using BeiDou (BDS) as a signal source in one embodiment of the present invention; Figure 4 It is the soil moisture result measured by using GPS as the signal source in one embodiment of the present invention. DETAILED DESCRIPTION

[0011] 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.

[0012] The present invention is further described in detail below in conjunction with the accompanying drawings: A GNSS-R soil moisture real-time inversion method based on FPGA+DSP architecture, the implementation time is November 15, 2024, the location is the experimental farm of Weihai Academy of Agricultural Sciences, Shandong Province, the signal source is BDS B1I signal and GPS L1C / A signal, according to the specific situation of the embodiment, the specific method of real-time inversion of soil moisture includes: S1. Construction of navigation satellite receiver baseband processing platform based on FPGA+DSP architecture: 1) Use a tripod to build an up-and-down antenna. Rotate the GNSS antenna right-handedly upward to receive direct signals from navigation satellites, and rotate the GNSS antenna left-handedly downward to receive reflected signals from navigation satellites. 2) The GNSS antenna is connected to the FPGA+DSP real-time processing device through a radio frequency line, and the device processes and saves the data results in real time; 3) Use a soil moisture meter to collect soil moisture values ​​at multiple locations on site and take the average as the true value for comparison.

[0013] S2. Real-time system composition and parameter settings based on FPGA+DSP architecture and specific processing flow: The system composition and parameter settings are as follows: 1) The FPGA used is Xilinx's XC6VLX240T, which has 241152 logic units and is a high-performance mainstream FPGA chip specifically used for signal processing development and programming; 2) DSP uses TI's TMS320C6748 floating-point processor with a main frequency of 456MHz; 3) The analog-to-digital converter ADC uses the AD9233 chip from Analog Devices, which is a single-chip, 12-bit, 125 MSPS analog-to-digital converter with a maximum sampling rate of 125MHz, 12-bit resolution, and an analog bandwidth of up to 650MHz, capable of RF direct bandpass sampling; 4) On-board reference clock: 10MHz, constant temperature crystal oscillator, up to 0.005ppm; 5) Doppler range is -16kHz ~ +16kHz; 6) Delay-Doppler image (DDM) mapping mode is 1ms coherent integration and 1000ms incoherent integration; 7) The intermediate frequency data sampling rate is 60MHz.

[0014] For specific processing procedures, refer to Figure 1 As shown: The navigation satellite antenna receives the GNSS antenna signal, which first passes through a bandpass filter to filter out signals in other frequency bands except the navigation signal band to eliminate signal interference; Since GNSS signals have the characteristic of low power, the filtered signals are amplified by a low noise amplifier to facilitate the next step of processing; The frequency synthesizer generates a local oscillation signal based on the local oscillator and amplifier of the device, and mixes and down-converts the signal with the GNSS signal through the mixer; The down-converted signal is passed through an amplifier and a low-pass filter to remove the high-frequency components in the signal, and is sampled by an automatic gain control (VGA) and an analog-to-digital converter (ADC) and input into the FPGA+DSP for processing.

[0015] S3, Rapid capture of navigation satellite signals based on FPGA+DSP: The fast capture module searches for signals based on the segmented correlation accumulation in the domain when the master clock is running and the DFT transform in the frequency domain. The module is divided into 22 channels for parallel code phase search, and each channel shares the real-time cached data, as follows: 1) Call the code NCO generation module to generate a pseudo code base frequency rate clock enable signal, that is, a 2.046MHz clock enable signal; 2) The input 60M data rate zero intermediate frequency baseband signal is accumulated and down-converted through the enable signal to generate a down-converted baseband signal with a data rate of 2.046M; 3) Store the decelerated baseband signal in real time in the dual-port RAM, with a storage depth of 6138 points, i.e. 3ms (2ms for data length and 1ms for searching through 2046 chips); 4) Generate 22-channel local pseudo code sequence and output it at the master clock rate; 5) Read 4096 points of data in the dual-port RAM and align them with the local pseudo code sequence and perform correlation accumulation. One point is accumulated every 64 points, and a total of 64 points of correlation accumulation values ​​are obtained; 6) Driven by the main clock, a 64-point DFT operation is performed on each accumulated value, and the result is stored in the 64-point dual-port RAM. When the second accumulated value is obtained, a second 64-point DFT operation is performed on it, and the result is accumulated to the first result. When the 64th DFT operation is completed, a complete 64-point DFT result is obtained; 7) Calculate DFT peak energy and average energy in real time and buffer them into registers; 8) Re-read the dual-port RAM data and align it with the local pseudo code sequence with a delayed data point and perform correlation accumulation to obtain the 64-point DFT result, calculate the DFT peak energy and average energy, repeat this 2046 times, record the one with the largest peak energy in the 64-point DFT, and you can get the pseudo code initial phase, the frequency point with the maximum DFT peak energy, and the peak energy and average energy information; 9) Generate a capture completion flag. When the DSP finds that this flag is valid, it reads the capture result and makes judgments and processes it.

[0016] S4, channel baseband signal processing flow: By performing complex phase rotation down-conversion and narrow correlation operation on the intermediate frequency signal, five-way correlation value output can be obtained, and then the real-time soil moisture value can be obtained through the Topp model, as follows: 1) Driven by the 60MHz system master clock input, the carrier NCO is called to generate a local 1561.098MHz oscillation signal, which is mixed with the input orthogonal signal to complete digital down-conversion; The in-phase orthogonal signal components obtained by the phase rotation method digital down-conversion in the present invention do not have other frequency components. Compared with the traditional orthogonal down-conversion method, the low-pass filter can be omitted, and the delay and attenuation of the signal can be reduced. The in-phase orthogonal signals are: ; ; It can be expressed in complex form as: ; The output of the carrier NCO is expressed as a complex number: ; so: ; Then the I and Q signals after phase rotation are: ; ; In the above method, d(t) is the data modulation signal, p(t) is the pseudo code signal, ω c Indicates the carrier angular frequency I in , Q in They are the in-phase and quadrature components before orthogonal down-conversion, which are the decomposition of the original signal based on trigonometric functions; S(t) is the conversion of I in and Q in The signal is represented in complex form. At the output of the carrier NCO, ω n is the oscillation angular frequency generated by the NCO, and its output signal is used to mix with the input signal S(t) to achieve digital down-conversion. During mixing, the signal frequency becomes ω c +ω n , through trigonometric and complex operations, the result is replaced by a new in-phase component and the orthogonal components Indicates. out and Q out They are the in-phase and quadrature signals after digital down-conversion using the phase rotation method.

[0017] The phase rotation method not only simplifies down-conversion into simple four arithmetic operations, but also does not introduce other frequency components. When , the Doppler information, the remaining pseudo code and the data modulation information can be completely eliminated; 2) When performing narrow correlation, a 1 / 4 chip time interval is selected for correlation operation, that is, five pseudo code sequences of advance, advance instant, instant, instant lag, and lag are generated locally, and five outputs are output after correlation operation; 3) When processing the reflected signal of the navigation satellite, the two channels with the highest signal-to-noise ratio (SNR) in the direct channel are selected for open-loop tracking processing. In order to prevent frequent satellite changes, it is stipulated that the satellite will be changed only when the third satellite is 3dB greater than any one of the first two satellites. The Doppler frequency shift is 15 channels, which are non-intervally distributed, namely 0Hz, ±25Hz, ±50Hz, ±100Hz, ±200Hz, ±300Hz, ±400Hz and ±500Hz. The 60MHz main clock is driven to reduce the speed to obtain a 4MHz rate clock enable signal, and then the selector selects 15 data streams in turn through the accumulator operation and caches the accumulated results accordingly. After the next polling clock arrives, the previously cached results are called out and participate in the accumulation operation of the corresponding channel; 4) Use the Topp model to invert the soil moisture of the relevant results. Based on the reflectivity obtained from the relevant output, the soil dielectric constant can be obtained as: ; in, is the soil dielectric constant, is the satellite altitude angle, is the power ratio, and the real-time soil moisture can be obtained through the Topp model data: .

[0018] S5. Implementation results: In this embodiment, the soil moisture results measured by the above steps are compared with the values ​​collected on-site using a soil moisture meter to verify the reliability of the data. After multi-point sampling by the soil moisture meter, the results of the measurement of bare soil just after the sweet potato harvest are as follows: Figure 3 , Figure 4 As shown, the true value is 0.15. The average soil moisture values ​​obtained with BDS and GPS as signal sources are 0.10 and 0.19, respectively. The root mean square errors (RMSE) are 0.06 and 0.06, respectively. The mean value differences are 0.05 and 0.04, respectively, which are significantly correlated with the true value obtained by the soil moisture meter.

[0019] In summary, the soil moisture monitoring method designed by the present invention is based on the FPGA+DSP architecture and uses GNSS satellites as signal sources, so as to realize real-time soil moisture monitoring, and the obtained results have a good correlation with the real results; the present invention makes up for the shortcoming of the time lag of the traditional use of ground-based GNSS-R to invert soil moisture, and can promote the application of GNSS-R technology, which plays an important role in agricultural monitoring, food production, soil drought monitoring and other fields; the present invention has high operating efficiency and low power consumption, and due to the flexibility of FPGA, it can also be further modified according to different application scenarios.

[0020] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A GNSS-R soil moisture real-time inversion method based on FPGA+DSP architecture, characterized in that: The following steps are involved: S1. Build a navigation satellite receiver baseband processing platform based on FPGA+DSP architecture: The platform includes: The right-handed GNSS antenna is tilted upward to receive the direct signal from the navigation satellite; the left-handed GNSS antenna is tilted downward to receive the reflected signal from the navigation satellite; The GNSS antenna is connected to the FPGA+DSP real-time processing device via a radio frequency line for processing; The soil moisture meter collects soil moisture values ​​at multiple locations on site and takes the average value as the true value for comparison; S2, fast capture of navigation satellite signals based on FPGA+DSP; The fast capture module searches for signals based on the segmented correlation accumulation in the domain when the master clock is running and the DFT transform in the frequency domain. The module is divided into 22 channels for parallel code phase search, and each channel shares the real-time cached data. S3, channel baseband signal processing; By performing complex phase rotation down-conversion and narrow correlation operation on the intermediate frequency signal, five-way correlation value output is obtained, and then the real-time soil moisture value is obtained through the Topp model. It is compared with the true value collected on-site by the soil moisture meter to verify the reliability of the data results.

2. The GNSS-R soil moisture real-time inversion method based on FPGA+DSP architecture according to claim 1 is characterized in that: In step S1, the FPGA+DSP real-time processing method is: The navigation satellite antenna receives the GNSS antenna signal, which first passes through a bandpass filter to filter out signals in other frequency bands except the navigation signal band to eliminate signal interference; The filtered signal is amplified by a low noise amplifier for further processing; The frequency synthesizer generates a local oscillation signal based on the local oscillator and amplifier of the device, and mixes and down-converts the signal with the GNSS signal through the mixer; The down-converted signal is passed through an amplifier and a low-pass filter to remove the high-frequency components in the signal, and is sampled by an automatic gain control and an analog-to-digital converter and input into the FPGA+DSP for processing.

3. The GNSS-R soil moisture real-time inversion method based on FPGA+DSP architecture according to claim 1 is characterized in that: The specific method for fast capture of navigation satellite signals based on FPGA+DSP in step S2 is as follows: (1) Call the code NCO generation module to generate a pseudo code base frequency rate clock enable signal, that is, a 2.046 MHz clock enable signal; (2) The input 60M data rate zero intermediate frequency baseband signal is accumulated and down-converted through the enable signal to generate a down-converted baseband signal with a data rate of 2.046M; (3) Store the decelerated baseband signal in real time in a dual-port RAM with a storage depth of 6138 points, or 3 ms; (4) Generate 22-channel local pseudo-code sequence and output it at the master clock rate; (5) Read 4096 points of data from the dual-port RAM and align them with the local pseudo-code sequence and perform correlation accumulation. One point is accumulated every 64 points, and a total of 64 points of correlation accumulation values ​​are obtained; (6) Driven by the master clock, a 64-point DFT operation is performed on each accumulated value, and the result is stored in a 64-point dual-port RAM. When the second accumulated value is obtained, a second 64-point DFT operation is performed on it, and the result is added to the first result. After the 64th DFT operation is completed, a complete 64-point DFT result is obtained. (7) Calculate DFT peak energy and average energy in real time and buffer them into registers; (8) Re-read the dual-port RAM data and align it with the local pseudo code sequence with a delay of one data point and perform correlation accumulation to obtain a 64-point DFT result. Calculate the DFT peak energy and average energy. Repeat 2046 times and record the one with the largest peak energy among the 64-point DFTs. You can then obtain the pseudo code initial phase, the frequency point with the maximum DFT peak energy, and the peak energy and average energy information. (9) Generate a capture completion flag. When the DSP finds that this flag is valid, it reads the capture result and makes judgments and processes it.

4. The GNSS-R soil moisture real-time inversion method based on FPGA+DSP architecture according to claim 1 is characterized in that: The steps of channel baseband signal processing are as follows: (1) Driven by the 60MHz system master clock input, the carrier NCO is called to generate a local 1561.098MHz oscillation signal, which is mixed with the input orthogonal signal to complete digital down-conversion; (2) When performing narrow correlation, a 1 / 4 chip time interval is selected for correlation operation, that is, five pseudo-code sequences of advance, advance instant, instant, instant lag, and lag are generated locally, and five outputs are output after correlation operation; (3) When processing the reflected signal of the navigation satellite, the two channels with the highest signal-to-noise ratio (SNR) in the direct channel are selected for open-loop tracking processing. In order to prevent frequent satellite changes, it is stipulated that the satellite will be changed only when the third satellite is 3 dB greater than any one of the first two satellites. The Doppler frequency shift is 15 channels, which are non-intervally distributed, namely 0 Hz, ±25 Hz, ±50 Hz, ±100 Hz, ±200 Hz, ±300 Hz, ±400 Hz and ±500 Hz. The 60 MHz master clock is driven to reduce the speed to obtain a 4 MHz rate clock enable signal, and then the selector selects 15 data streams in turn to pass through the accumulator operation and cache the accumulated results accordingly. After the next polling clock arrives, the previously cached results are called out and participate in the accumulation operation of the corresponding channel; (4) The Topp model is used to invert the soil moisture of the relevant results. Based on the reflectivity obtained from the relevant output, the soil dielectric constant can be obtained as: ； in, is the soil dielectric constant, is the satellite altitude angle, The power ratio is further used to obtain the real-time soil moisture through the Topp model. data: 。

Images