CN110208752A - A kind of radar MTI/MTD implementation method based on GPU - Google Patents
A kind of radar MTI/MTD implementation method based on GPU Download PDFInfo
- Publication number
- CN110208752A CN110208752A CN201910566395.4A CN201910566395A CN110208752A CN 110208752 A CN110208752 A CN 110208752A CN 201910566395 A CN201910566395 A CN 201910566395A CN 110208752 A CN110208752 A CN 110208752A
- Authority
- CN
- China
- Prior art keywords
- data
- gpu
- dimension
- mti
- mtd
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/28—Details of pulse systems
- G01S7/285—Receivers
- G01S7/295—Means for transforming co-ordinates or for evaluating data, e.g. using computers
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention belongs to Radar Signal Processing Technology fields, are related to a kind of radar MTI/MTD implementation method based on GPU.The present invention handles the Clutter inhibited at zero Doppler frequency by MTI;It cannot be exactly matched by the notch width and clutter spectral width of mti filter clutter reduction, making output end still has more residual spur, and MTD can further suppress clutter;The acceleration that algorithm is realized using GPU is carried out three pulses in MTI and offseted, matrix transposition, all uses parallel algorithm, function admirable in FFT operation.
Description
Technical field
The invention belongs to Radar Signal Processing Technology fields, are related to a kind of radar MTI/MTD implementation method based on GPU.
Background technique
The rise of high speed development and GPU parallel computation recently as GPU technology, GPU can only be applied to from initially
Accelerate operation to develop to research application in the parallel algorithm of every field in image procossing, and is suitably opened as GPU has
Interface and exploitation environment are sent out, it also can be in the parallel processing technique and floating-point operation energy that general-purpose computations field makes full use of its powerful
Power reaches good acceleration effect.The characteristics of CUDA is programmed is the heterogeneous system using CPU+GPU, and CPU is allowed to become in model
Host (Host), GPU are then used as coprocessor or equipment (Device) to cooperate jointly with CPU and complete calculating task.GPU and CPU
When collaborative work, using CPU and the respective structural advantage of GPU, CPU is allowed to execute the serial code in program, complete logicality compared with
Strong task, such as control data transmit operation related with controlling equipment completion etc., and GPU then completes the number of wherein high degree of parallelism
According to parallel processing task, acceleration of the GPU in operation highly dense is given full play to.
MTI processing is to improve signal to noise ratio to inhibit the Clutter in radar illumination region, reduces false alarm rate, and
Suitable algorithm can be designed according to the actual working environment and index request of the array radar when realizing to be handled.Clutter suppression
Output after system also will do it MTD filtering processing, i.e. correlative accumulation, further inhibits residual spur and improves signal-to-noise ratio, mentions
High target detection probability.
Summary of the invention
In order to meet the requirement of real-time of Radar Signal Processing, MTI/MTD must have very high real-time performance, the present invention
The high-speed computational capability of GPU is utilized, devises a kind of MTI/MTD implementation method accelerated based on GPU, it can be with faster speed
Degree is completed, and the requirement of system signal processing real-time is met.
The radar MTI/MTD implementation method based on GPU that the invention proposes a kind of, comprising:
S1, the fast time dimension echo data of radar is sequentially stored in video memory with each channel of each pulse.
S2, MTI is carried out based on GPU, i.e., three pulses is successively carried out to all pulses in every frame and is offseted, GPU thread knot is set
Structure is three-dimensional grid, and all thread blocks of each direction the x dimension of three-dimensional grid is enabled to complete the calculating in a corresponding channel, the grid side y
To dimension be port number, grid z-dimension is that need to handle umber of pulse;
S3, that each pulse data treatment process in processing is offseted due to three pulses is independent, so by this step of GPU Parallel Implementation
Suddenly.Thread one group of three pulse data of completion of each direction the x dimension of GPU thread grid offsets processing.At N number of pulse data
N-2 group result is obtained after reason.
S4, matrix transposition is carried out, the two-dimensional data matrix of m- slow time, data when data are by being one fast after MTI
Matrix is with fast time dimension to go, that is, it is continuous for storing the data of fast time dimension, and the data of slow time dimension are discrete;
Matrix transposition is carried out, slow time dimension data is changed into row, fast time dimension changes into column, finally according to the slow time data of continuous data
It is allocated memory, obtains the slow time data of contiguous memory;
S5, the slow time dimension progress FFT in slow time data matrix m- when having obtained fast is converted by FFT operation
To distance-Doppler data matrix;
S6, restoring data carry out transposition, as MTD result to FFT result again.
The principle that MTI is used is Doppler effect, tune of the reflected train of pulse of moving target due to Doppler frequency shift
Production can be changed with the pulse envelope of different pulse repetition periods, and the envelope of echo impulse string is caused to occur difference each other,
And the envelope of Clutter is fixed, when being offseted with the echo impulse of different pulse repetition periods, the inside includes static
Clutter will be cancelled, and echo-signal has residue, can be detected, to achieve the effect that inhibit Clutter.
The principle of MTD processing be using one group of adjacent and partly overlapping narrowband Doppler filter group, frequency band subdivisions,
The output clutter of each filter is allowed to only have in respective passband, rather than the clutter in entire frequency band.
Beneficial effects of the present invention are that MTI processing inhibits the Clutter at zero Doppler frequency;By mti filter
The notch width of clutter reduction cannot be exactly matched with clutter spectral width, and making output end still has more residual spur, and MTD can be with
Further suppress clutter;The acceleration that algorithm is realized using GPU is carried out three pulses in MTI and offseted, matrix transposition, in FFT operation
All use parallel algorithm, function admirable.
Detailed description of the invention
Fig. 1 is three pulse canceller structural schematic diagram;
Fig. 2 is three pulse canceller MTI parallel computation schematic diagram;
M- slow time data matrix schematic diagram when Fig. 3 is fast;
M- slow time data matrix transposition Memory Allocation schematic diagram when Fig. 4 is fast;
Fig. 5 is MTD concurrent operation design implementation process schematic diagram;
Fig. 6 is the MTI functional verification of GPU;
Fig. 7 is that GPU completes MTD result.
Specific embodiment
Further technical solution of the present invention is described with reference to the accompanying drawing.
Step 1: the fast time dimension echo data of radar is sequentially stored in video memory with each channel of each pulse.
Step 2: setting GPU thread structure is three-dimensional grid, all threads of each direction the x dimension of three-dimensional grid is enabled to complete
The calculating in a corresponding channel, wherein grid x dimension number of threads is each channel data number, the direction grid y dimension is logical
Road number, grid z-dimension are that need to handle umber of pulse.
Step 3: since to offset each pulse data treatment process in processing independent for three pulses, thus by GPU Parallel Implementation this
Step.As shown in Fig. 2, thread one group of three pulse data of completion of each direction the x dimension of grid offsets processing.To N number of pulse
N-2 group result is obtained after data processing.As shown in Figure 1.Wherein, delay T indicates pulse recurrence interval (1/PRF), and be easy to get input
Output relation is y (t)=x (t) * h (t)=x (t) -2x (t-T)+x (t-2T)
According to relational expression, corresponding three input pulse signal datas are carried out addition and subtraction processing and produced by each x dimension sets of threads
Raw signal output, i.e., offset the output signal vector of generation, each of vector element is all by three by three pulses
Different moments, corresponding element was added and subtracted, so corresponding different vector element, which carries out processing, complies fully with CUDA's
SIMT model coincide, and different threads executes identical instruction, but is different thread and is responsible for handling different data, thus
Just obtain different output elements.
Step 4: carrying out matrix transposition using GPU.The two-dimemsional number of m- slow time when data are by being one fast after MTI
According to matrix, data matrix is that store the data of fast time dimension for row with fast time dimension be continuous, and the data of slow time dimension
It is discrete.The spectrum analysis of MTD is to be handled with slow time dimension data, therefore carry out transposition as shown in figure 4, the slow time
Dimension data changes into row, and fast time dimension changes into column, is finally allocated memory according to the slow time data of continuous data, obtains continuous
The slow time data of memory carry out spectrum analysis afterwards for it and make place mat.
Each element is independent during matrix transposition, and it is mobile to be responsible for an element transposition for per thread in GPU.
Step 5: using the FFT operation based on GPU by the slow time dimension in slow time data matrix m- when having obtained fast
It carries out FFT and is converted to " distance-Doppler " data matrix.
Step 6: restoring data, carries out transposition, as MTD result to FFT result again.
As shown in fig. 6, offseting the inhibition carried out to clutter, exquisite target using three pulses, upper figure is MTI two dimension output knot
Fruit figure, includes 32 group pulses, the fast time dimension index dimension target range index where bright rays.
As shown in fig. 7, maximum gain is obtained in speck position, to obtain Doppler's letter of target.
The following table 1 is that 32 pulses progress MTI/MTD processing are compared in CPU and GPU time-consuming
It can be obtained by table 1, under the situation of 32 pulses, be significantly lower than using the time required for GPU completion MTI/MTD
CPU, performance improve significant.
Claims (1)
1. a kind of radar MTI/MTD implementation method based on GPU characterized by comprising
S1, the fast time dimension echo data of radar is sequentially stored in video memory with each channel of each pulse;
S2, setting GPU thread structure are three-dimensional grid, and all thread blocks of each direction the x dimension of three-dimensional grid is enabled to complete corresponding one
The calculating in a channel, the direction grid y dimension are port number, grid z-dimension is that need to handle umber of pulse;
S3, MTI is carried out based on GPU, i.e., three pulses is successively carried out to all pulses in every frame and offseted, since three pulses offsets processing
In each pulse data treatment process it is independent, so by this step of GPU Parallel Implementation;The line of each direction the x dimension of GPU thread grid
Journey one group of three pulse data of completion offsets processing, obtains N-2 group result after handling N number of pulse data;
S4, matrix transposition is carried out, the two-dimensional data matrix of m- slow time, data matrix when data are by being one fast after MTI
It is with fast time dimension to go, that is, it is continuous for storing the data of fast time dimension, and the data of slow time dimension are discrete;It carries out
Slow time dimension data is changed into row by matrix transposition, and fast time dimension changes into column, is finally carried out according to the slow time data of continuous data
Storage allocation obtains the slow time data of contiguous memory;
S5, by FFT operation by slow time data matrix m- when having obtained fast slow time dimension carry out FFT be converted to away from
From-doppler data matrix;
S6, restoring data carry out transposition, as MTD result to FFT result again.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910566395.4A CN110208752A (en) | 2019-06-27 | 2019-06-27 | A kind of radar MTI/MTD implementation method based on GPU |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910566395.4A CN110208752A (en) | 2019-06-27 | 2019-06-27 | A kind of radar MTI/MTD implementation method based on GPU |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110208752A true CN110208752A (en) | 2019-09-06 |
Family
ID=67795026
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910566395.4A Pending CN110208752A (en) | 2019-06-27 | 2019-06-27 | A kind of radar MTI/MTD implementation method based on GPU |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110208752A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112986944A (en) * | 2021-03-04 | 2021-06-18 | 西安电子科技大学 | CUDA heterogeneous parallel acceleration-based radar MTI and MTD implementation method |
CN113093180A (en) * | 2021-03-20 | 2021-07-09 | 中国人民解放军空军工程大学 | Low-speed weak target detection method based on cluster extrapolation |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104849698A (en) * | 2015-05-21 | 2015-08-19 | 中国人民解放军海军工程大学 | Radar signal parallel processing method and system based on heterogeneous multinucleated system |
CN107153547A (en) * | 2017-05-15 | 2017-09-12 | 电子科技大学 | A kind of software implementation radar storehouse mode signal handles middleware |
CN107576950A (en) * | 2017-09-28 | 2018-01-12 | 西安电子科技大学 | A kind of optimized treatment method of pulse compression radar echo-signal |
CN108919227A (en) * | 2018-08-17 | 2018-11-30 | 电子科技大学 | A kind of multichannel FBLMS implementation method accelerated based on GPU |
-
2019
- 2019-06-27 CN CN201910566395.4A patent/CN110208752A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104849698A (en) * | 2015-05-21 | 2015-08-19 | 中国人民解放军海军工程大学 | Radar signal parallel processing method and system based on heterogeneous multinucleated system |
CN107153547A (en) * | 2017-05-15 | 2017-09-12 | 电子科技大学 | A kind of software implementation radar storehouse mode signal handles middleware |
CN107576950A (en) * | 2017-09-28 | 2018-01-12 | 西安电子科技大学 | A kind of optimized treatment method of pulse compression radar echo-signal |
CN108919227A (en) * | 2018-08-17 | 2018-11-30 | 电子科技大学 | A kind of multichannel FBLMS implementation method accelerated based on GPU |
Non-Patent Citations (2)
Title |
---|
史林: "《雷达***建模与仿真》", 31 December 2017, 北京:国防工业出版社 * |
朱晓芳: "基于GPU的阵列雷达信号处理技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112986944A (en) * | 2021-03-04 | 2021-06-18 | 西安电子科技大学 | CUDA heterogeneous parallel acceleration-based radar MTI and MTD implementation method |
CN112986944B (en) * | 2021-03-04 | 2023-09-08 | 西安电子科技大学 | Radar MTI and MTD implementation method based on CUDA isomerism parallel acceleration |
CN113093180A (en) * | 2021-03-20 | 2021-07-09 | 中国人民解放军空军工程大学 | Low-speed weak target detection method based on cluster extrapolation |
CN113093180B (en) * | 2021-03-20 | 2023-10-13 | 中国人民解放军空军工程大学 | Clustering extrapolation-based slow weak target detection method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104849698B (en) | A kind of radar signal method for parallel processing and system based on heterogeneous multi-core system | |
CN109143179A (en) | A kind of frequency-agile radar signal processing method and device based on change repetition technology | |
CN104237883B (en) | Airborne radar space time self-adaptation processing method with sparse representation | |
US4649394A (en) | Method of and a device for recognizing targets and suppressing spurious signals in radar equipment | |
CN110208752A (en) | A kind of radar MTI/MTD implementation method based on GPU | |
CN104237859B (en) | Method for achieving external illuminator radar multi-channel time domain clutter suppression by means of GPU | |
CN104977571B (en) | Range ambiguity clutter suppression method based on pitching frequency diversity STAP | |
CN108120976B (en) | Ground clutter spectrum leakage suppression method based on Doppler channel characteristics | |
CN104502898B (en) | The maneuvering target method for parameter estimation that modified R FT and amendment MDCFT are combined | |
CN106682820A (en) | Optimized radar task scheduling method of digital array based on pulse interlacing | |
CN104459661B (en) | Method for detecting rapid artillery type dim target | |
CN113504522B (en) | Space-time decoupling and super-resolution angle measurement method based on random switching of transmitting antennas | |
CN104239683B (en) | Resolving distance-velocity ambiguity method on basis of changing signal rate | |
CN105510895B (en) | A kind of signal processing method of random pulses intervening sequence | |
CN113221631B (en) | Sequence pulse anti-interference target detection method based on convolutional neural network | |
CN103777189A (en) | Radar weak target detecting method based on information geometry multiple autoregressive model | |
CN109375179A (en) | A kind of strong clutter suppression method of airborne radar short range based on rarefaction representation | |
CN106324596A (en) | Random pulse Doppler radar angle-Doppler imaging method based on compressed sensing | |
CN106772304A (en) | Doppler's adaptive processing method after airborne MIMO radar based on spatial domain multi-level decomposition | |
CN109709536A (en) | A kind of SAR moving target detection method based on convolutional neural networks | |
CN112255607B (en) | Sea clutter suppression method | |
CN107153178A (en) | External illuminators-based radar reference signal contains object detection method during multi-path jamming | |
CN104345299A (en) | Airborne MIMO (Multiple Input Multiple Output) radar space-time self-adaptive processing method based on simplified EC | |
CN107490788A (en) | A kind of space-time adaptive processing method suitable for MIMO airborne radar non homogeneous clutter suppressions | |
CN107390194A (en) | A kind of radar target detection method based on the graceful divergence of full Donald Bragg |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190906 |
|
WD01 | Invention patent application deemed withdrawn after publication |