With the rapid development of radar industry technology, the corresponding signal processing technology becomes more and more complex. For the radar with short-range detection function, its corresponding signal mostly presents the characteristics of wide bandwidth and multiresolution. In the traditional data processing process, a large number of signals will interfere with the signal, which makes the final signal processing difficult or even impossible. Based on this problem, this paper proposes a principal component linear prediction processing algorithm based on clutter suppression processing on the basis of traditional signal processing algorithm. According to the curve characteristics of the data returned by the target detected by the signal, through certain image signal measurement and transformation, the clutter can be effectively suppressed and the typical characteristics of the corresponding target curve can be enhanced. For the convergence problem of signal processing and the corresponding image chromatic aberration compensation problem, this paper will realize the chromatic aberration compensation of the corresponding target echo image based on the radial pointing transverse mode algorithm and enhance the convergence speed of the whole algorithm system. In the experimental part of this paper, the optimization algorithm proposed in this paper is compared with the traditional algorithm. The experimental results show that the algorithm proposed in this paper has obvious advantages in the convergence of signal processing and antijamming performance and has the promotion value.

With the rapid development of information technology and electronic technology, radar technology has been further developed. The range resolution of common low altitude detection radar mainly depends on the bandwidth of the transmitted signal, the corresponding broadband data signal processing technology and antijamming technology [

The traditional stepped frequency multiresolution digital signal processing mainly focuses on the system hardware design optimization and echo signal optimization processing. In the corresponding hardware level, it is not the key problem to be considered in this paper. For the digital signal processing level, it mainly deals with the close range digital clutter. The corresponding traditional processing methods include average method digital signal processing algorithm, digital signal principal component analysis algorithm, and corresponding digital signal linear prediction algorithm [

The structure of this paper is as follows: in the second section, the current research status of stepped frequency multiresolution digital signal processing algorithm is analyzed and discussed. In the third section of this paper, we will focus on the analysis of stepped frequency multiresolution digital signal processing algorithm, mainly on the analysis of principal component linear prediction processing algorithm based on clutter suppression processing and radius pointing transverse mode algorithm. In the fourth section of this paper, we will make a comparative experiment on the algorithm. Finally, this paper is summarized.

The main technical difficulty in the field of stepped frequency low altitude radar detection is digital signal processing algorithm. Stepped frequency radar mainly transmits a group of wideband pulse signals by carrier frequency hopping. The corresponding echo signal processing technology becomes the difficulty of digital signal processing [

This section mainly analyzes and studies the stepped frequency multiresolution signal processing algorithm. The main core algorithms are the principal component linear prediction processing algorithm and the radius pointing transverse mode algorithm. The corresponding algorithm architecture is shown in Figure

Architecture of stepped frequency multiresolution signal processing algorithm.

In the corresponding experimental level, this paper first carries out the accurate modeling of the echo data signal and then carries out the experiment based on the correlation modeling. In the experimental part, this paper not only carries out the comparative experiments of four different algorithms for the processing of stepped frequency multiresolution digital signal, but also simulates the convergence and convergence speed of the four different algorithms based on MATLAB and the corresponding modeling. The simulation and experimental frameworks are shown in the figure.

At the clutter suppression level, this section is mainly based on the linear prediction of principal components for clutter suppression. At the level of principal component processing, it mainly analyzes and processes the collected echo signal from the perspective of two-dimensional determinant. At the level of principal component, the corresponding core processing ideas are as follows: based on the perspective of two-dimensional determinant analysis, it constructs and calculates the corresponding unrelated secondary fields, so as to analyze the relationship between the weights of the secondary fields. Taking the echo data as an example, the principal component analysis is carried out. Assuming that the corresponding echo data is

Based on formula (

Based on this, the corresponding principal component processing algorithm mode is determined as shown in Figure

Step 1: take the corresponding echo data as the corresponding processing object, and the corresponding processing matrix calculation formula is shown in formula (

Step 2: use the traditional mean algorithm to process the echo data, and preprocess the corresponding echo data.

Step 3: based on the corresponding two-dimensional processing matrix, the corresponding echo data is processed by covariance matrix.

Step 4: analyze the corresponding eigenvalues and eigenvector values of the matrix after processing the corresponding echo data.

Step 5: select the corresponding principal component for processing and analysis, and reconstruct the corresponding echo data.

Algorithm operation flowchart of principal component processing algorithm module.

Based on the corresponding eigenvalues and eigenvector values obtained by the above principal component algorithm as the reference data of the linear prediction algorithm, the corresponding echo data is further processed and analyzed, which mainly realizes the prediction of the current measurement data. Based on this, the corresponding core processing formula is shown in formula (

For the complex interference factors faced by close range detection, the mathematical model is defined based on the above formula and the corresponding close range environmental factors, and the corresponding mathematical model is shown in Figure

Mathematical model of linear prediction algorithm.

The echo data processed based on the principal component is taken as the corresponding reference data, and the corresponding bilateral data of the echo data matrix is selected as the linear prediction data of the current radar scan, and the number of corresponding reference data is set to W. When the number of reference data is small, the corresponding reference range will narrow, and the corresponding prediction effect will be greatly reduced. When the number of reference data is large, the detection difference of the corresponding radar in the corresponding area will be large. In this way, the corresponding data will lose its representativeness. Therefore, when dealing with the number of reference data, this section filters based on the data eigenvalues after the principal component preprocessing, so as to solve the problem of the number of reference data. In this section, the actual selection is generally between 10 and 20. Based on the above principle analysis, the corresponding linear prediction algorithm flow based on principal component preprocessing is shown in Figure

Step 1: recombine the collected data after preprocessing the corresponding principal components, rename the corresponding data matrix, and reanalyze all the new matrices.

Step 2: select a reasonable subspace from the above analysis results and recombine the corresponding information data.

Step 3: conduct linear bilateral processing analysis on the data in step 2 above, and the number of corresponding linear bilateral processing data is set to 6 (12 in total).

Step 4: use the least square method to predict the above data, so as to finally obtain the measured value and the corresponding estimated value, and based on the measured value and the estimated value, reduce the difference.

Algorithm operation flowchart of linear prediction processing algorithm module.

In order to solve the coupling interference between the equipment antennas corresponding to the short-range detection system, eliminate the external interference factors of nonalgorithm factors, so as to further optimize the above algorithm; this section designs the auxiliary design of antenna coupling interference elimination. The coupling between antennas mainly exists in the fixed and stable clutter between the transmitting antenna and the receiving antenna. Based on the characteristics of this clutter, this section uses the air acquisition as the actual reference and subtracts the actual collected data. The corresponding processing steps are as follows.

Step 1: set up the corresponding short-range detection equipment, debug the corresponding parameters, and keep the interval between the receiving and transmitting antennas constant.

Step 2: shoot the corresponding equipment antenna to a clean environment without clutter interference, and collect data for the corresponding clean environment.

Step 3: preprocess the collected data to remove the interference of the corresponding target source. The collected echo data is mainly the mutual coupling direct wave between the receiving and transmitting antennas.

Step 4: average and store the corresponding data. The current reference data should be subtracted from the subsequent collected echo data so that the cleaner data can be obtained in the actual data processing of this algorithm.

In order to improve the convergence speed of the whole linear prediction algorithm and solve the unipolarity problem of the corresponding image problem, this section proposes the radius pointing transverse mode algorithm based on independent cell analysis algorithm. When the corresponding echo data is preprocessed by principal component, part of the signal is preprocessed at the same time. The corresponding transverse mode algorithm is used to iterate the corresponding data. The corresponding iteration termination condition is set as the absolute value threshold. Based on this, the inverse matrix of the corresponding echo data can be obtained. Based on the inverse matrix obtained above, the radius pointing transverse mode algorithm is used for depolarization multiplexing, so that the processing of echo data can get fast convergence. Based on this, the corresponding algorithm flow chart is shown in Figure

Step 1: preprocess and analyze the collected echo data.

Step 2: based on constant modulus algorithm, the preprocessed data are processed iteratively.

Step 3: repeat step 2 until the iteration condition is terminated (a certain threshold is met) to obtain the corresponding inverse matrix of echo data.

Step 4: based on the inverse matrix of echo data, the radius pointing transverse mode algorithm is used for depolarization multiplexing to achieve fast convergence.

Flowchart of radius pointing transverse die algorithm.

The corresponding hardware system is designed based on the above algorithm, and the corresponding hardware system architecture is shown in Figure

Hardware system architecture of stepped frequency multiresolution digital signal processing algorithm.

In the corresponding data acquisition level, the relevant chips of ad company are mainly used, and the main models include ad9042, ad9631, ad8138, and ad4938. For the circuit composed of ad9631 and its related resistance and capacitance, its main function is DC coupling circuit, and its corresponding circuit needs to be equipped with emitter follower composed of operational amplifier. Its main function is to draw out the internal bias voltage corresponding to ad9042, and the adjustment of bias voltage is still mainly carried out through operational amplifier. In the ADC system, bias correction is needed. The corrected bias voltage and the input echo digital signal are synthesized by the broadband low noise operational amplifier and used as the analog input of ad9042. In the design of the above data acquisition circuit, the power supply part of the digital circuit should be separated from the analog part to prevent its impact on the ADC conversion speed.

The implementation of IFFT in the key FPGA circuit of the corresponding digital processing part mainly realizes FFT operation through the IP core designed by Altera company. In the aspect of selecting the corresponding DSP, this paper mainly selects the TMS320C3X series of Texas Instruments. Its corresponding structure is relatively simple, and the corresponding peripheral devices are relatively few. At the same time, it has the advantages of high-quality floating-point, low power consumption, rich registers, and so on. At the same time, the corresponding hardware resources of the processor are also very rich. It has two 2K∗32-bit RAM modules and two 16K∗32-bit RAM modules. Its independent program bus, data bus, and DMA bus enable the processor to realize parallel processing.

In the above hardware architecture system, the corresponding auxiliary circuit design mainly includes power management circuit module, clock circuit design part, and reset circuit design part. The design of the corresponding power management module mainly considers the power supply to the corresponding FPGA and the corresponding DSP processor. In the aspect of clock design, the corresponding DSP, FPGA, ADC, and all kinds of memory circuits in the hardware board are homologous with the external clock. Considering the self-test function of the whole system, crystal oscillator is set in the corresponding printed circuit board. The corresponding reset circuit design level mainly includes three types of reset design: power on reset design, manual reset design, and corresponding command reset design.

In order to better analyze the echo signal and prepare for the following experiments, this section will model and process the echo signal. The theoretical analysis shows that the echo signal is mainly composed of three parts: external interference clutter, external interference noise, and the corresponding short-range target signal. When the corresponding equipment collects the echo data signal, it can be directly reflected in the signal spectrum. According to practical experience, the corresponding clutter signal can be attributed to the clutter interference directly coupled from the near air or ground underground. To a certain extent, the modeling of echo signal can be understood as the modeling of clutter signal, which corresponds to the clutter in the near air, underground, and ground. It is mainly caused by the debris in the near environment. The characteristic of this kind of clutter signal is that its change in the corresponding amplitude point is not very large. The corresponding clutter hidden in the ground or in the air is closely related to the target signal source. The corresponding clutter is often caused by the serious distortion of the underground or near space environment or the uneven distribution of the media in the corresponding cavity. The corresponding clutter is a rapidly changing process. Therefore, the model corresponding to this kind of clutter is random. The signals generated by the corresponding target source are usually signals with small amplitude. The corresponding other types of clutter are usually random clutter, which are generally set as the standard Gaussian white noise mathematical model, and are basically processed and analyzed by this model in the actual model processing. The corresponding coupling signals between antennas are preprocessed by subtraction algorithm. Based on the above analysis, the corresponding mathematical model of echo signal is established as shown in the following formula:

The experimental environment is set as follows: the experimental data is collected by a research institute, the corresponding experimental environment is a close ground, and there are holes and small stones of different specifications on the corresponding ground. The experiment is based on the same hardware system, and the data is collected along the test site in the way of horizontal movement, and the corresponding moving speed is set to uniform. In the actual comparative experiment, the experimental condition variables of the two algorithms are kept unchanged. In experiment scenario 1, the signal-to-noise ratio (SNR) is introduced to distinguish the experimental results. As shown in Table

Calculation results of SNR of different algorithms in scenario 1.

Digital signal processing algorithm | Signal-to-noise ratio |
---|---|

Raw data 1 | −15.345 |

Mean method | 8.987 |

Principal component algorithm | 10.427 |

Linear prediction algorithm | 15.293 |

An improved algorithm is proposed in this paper | 30.114 |

Raw data 2 | −20.351 |

Mean method | 18.086 |

Principal component algorithm | 22.135 |

Linear prediction algorithm | 25.901 |

An improved algorithm is proposed in this paper | 41.231 |

Raw data 3 | −22.131 |

Mean method | 20.861 |

Principal component algorithm | 22.904 |

Linear prediction algorithm | 25.619 |

An improved algorithm is proposed in this paper | 33.891 |

In experiment scenario 2, metal interferences are added on the basis of experiment scenario 1, and the corresponding radar hardware equipment continuously detects the detected target for five times in the test area. The amplitude accumulation diagrams of corresponding different algorithms are shown in Figures

(a) Stacking chart of processing results under mean algorithm. (b) Stacking diagram of processing results under principal component algorithm. (c) Stacking chart of processing results under linear prediction algorithm. (d) The stack diagram of processing results under the algorithm proposed in this paper.

The SNR calculation table based on experimental scenario 2 is shown in Table

SNR calculation results of different algorithms in scenario 2.

Digital signal processing algorithm | Signal-to-noise ratio |
---|---|

Raw data 1 | −11.547 |

Mean method | 4.111 |

Principal component algorithm | 5.192 |

Linear prediction algorithm | 8.361 |

An improved algorithm is proposed in this paper | 20.324 |

Raw data 2 | −15.339 |

Mean method | 8.774 |

Principal component algorithm | 15.934 |

Linear prediction algorithm | 20.331 |

An improved algorithm is proposed in this paper | 25.143 |

Raw data 3 | −20.302 |

Mean method | 11.213 |

Principal component algorithm | 18.201 |

Linear prediction algorithm | 21.354 |

An improved algorithm is proposed in this paper | 30.119 |

In order to verify the convergence rate of the algorithm, the simulation is carried out based on the simulation framework shown in Figure

Simulation framework of algorithm convergence.

In the corresponding simulation architecture diagram, the simulation is mainly divided into four channels, corresponding to mean value method, principal component method, linear measurement method, and the algorithm proposed in this paper. In the actual simulation process, the principle of control variable method is followed, and the corresponding data source is guaranteed to be the same, and the corresponding data processing link is guaranteed to be the same. The only difference is the data processing module.

The convergence speed diagram of the four algorithms simulated based on the above simulation frame diagram is shown in Figure

Curve of convergence rate of the algorithm.

Based on the data results of scenario 1 and scenario 2, we can draw the following conclusions: the single mean algorithm, principal component analysis, and linear prediction algorithm can filter most of the clutter when removing the clutter of the echo signal, but it still has the situation that the clutter can not be filtered in the face of complex environment, and it is subject to the relatively large correlation level. Compared with the single algorithm mentioned above, the algorithm proposed in this paper has obvious clutter filtering advantages, and its corresponding convergence speed is also very fast. Therefore, based on the above experimental results, the proposed algorithm has obvious processing advantages and convergence advantages.

This paper mainly analyzes the current situation and disadvantages of stepped frequency multiresolution digital signal processing technology. Based on the traditional signal processing algorithm, this paper proposes a principal component linear prediction processing algorithm based on clutter suppression processing. According to the curve characteristics of the data returned by the target detected by the signal, through certain image signal measurement and transformation, the clutter can be effectively suppressed and the typical characteristics of the corresponding target curve can be enhanced. For the convergence problem of signal processing and the corresponding image chromatic aberration compensation problem, this paper realizes the chromatic aberration compensation of the corresponding target echo image based on the radial pointing transverse mode algorithm and enhances the convergence speed of the whole algorithm system. In the experimental part, the proposed optimization algorithm is compared with the traditional algorithm. The experimental results show that the proposed algorithm has obvious advantages in the convergence of signal processing and antijamming performance and has the promotion value. This paper will focus on the optimization of convergence and antijamming for large-scale echo data.

No data were used to support the findings of the study.

The author declares that there are no conflicts of interest.

The author acknowledges the scientific research projects of Xi’an Peihua University in 2020, Design of Wearable Motion Data Display System Based on Bluetooth (PHKT2029).