Marine target detection in cluttered environments

Abstract

Method of slowly moving target detection with application for coastal surveillance radars. This method improves the well know other methods and efficiently detects targets with a high accuracy. The proposed method consists of three steps that are: step of generation and processing of signals with complex modulation; step of target clustering and step of detection of slowly moving targets in clutter environments.

Claims

1. A method of detecting a slowly moving radar target in clutter environments comprising the steps of: Step 1: generate and process signal with complex modulation, wherein in this step, a signal with complex modulation, high range resolution and high compression ratio is generated and processed; Step 2: target clustering, wherein in this step, all targets are clustered by their radial velocities to 2 groups comprising slowly moving targets and fast moving targets; and Step 3: slowly moving target detection by using together three detectors that are a 3D Ordered Statistical Constant False Alarm Rate (OS-CFAR) detector, a detector based on statistical analysis of clutter distributions and a detector based on a clutter map.

2. The method of detecting a slowly moving radar target in clutter environments according to claim 1, in which Step 1: generate and process signal with complex modulation consists of the following: a block of signal generator with complex modulation and high resolution; a block of coherent signal generator with complex generator; and a block of pulse compression with weighted window.

3. The method of detecting a slowly moving radar target in clutter environments according to claim 1, in which Step 2: target clustering comprises using the following blocks: a block of coherent intergration; and a block of doppler bank filters.

4. The method of detecting a slowly moving radar target in clutter environments according to claim 1, in which Step 3: Detect slowly moving targets in clutter environments comprising using together three independent detectors, First detector: 3D OS-CFAR (3 Dimensional Ordered Statistical Constant False Alarm Rate), where the powers of echo signals are arranged into a cube in three dimensional space representing range, azimuth, doppler), for each cell under test (CUT) in this cube, all powers in the reference cells are arranged as an increased sequence, and then the optimal threshold is estimated, This method has a better performance in comparison with 1D CFAR; Second detector based on the clutter distributions, by applying statistical techniques the distributions of clutters is obtained, for example, sea clutter has a lognormal distribution; weather clutter is followed by Rayleigh distribution while ground clutter has a Weibull distribution, from the probability of false alarm (Pfa) a threshold for target detection is obtained; and Third detection is based on clutter map, the clutter map is performed automatically using digital filtering, the target's echo signals are compared with this map to classify exactly these echo signals as being from clutters or from real targets.

5. The method of detecting a slowly moving radar target in clutter environments according to claim 1, in which The use of semiconductor amplifier technology solid state (amplifier by using of semiconductor transistors) to generate the explorer signal with complex modulation, high range resolution of below 3 meters after processing, in the result, reflected signals are easily classified from true targets or from clutters.

6. The method of detecting a slowly moving radar target in clutter environments according to claim 1, in which semiconductor amplifier technology is used to allow coherent processing, in the result target doppler shifts and target radial velocities can be obtained, hence, all targets can be combined into two groups that are the group of fast moving targets and the group of slowly moving targets in clutter environments.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1: is a block diagram that illustrates steps of processing;

(2) FIG. 2: is a block diagram that illustrates method of detection of slowly moving target;

(3) FIG. 3: is a diagram that illustrates result of compression pulse;

(4) FIG. 4: is a diagram that illustrates method of cluster target according to velocity;

(5) FIG. 5: is a diagram that illustrates results of noise distribution statistics;

(6) FIG. 6: is a diagram that illustrates method target detection by 3D-CFAR Algorithm; and

(7) FIG. 7: is a diagram that illustrates result target detection by clutter map.

DETAILED DESCRIPTION OF THE INVENTION

(8) Method of detection of slowly moving targets contain 3 processing steps:

(9) Step 1: Generate and Process Wideband Modulation Signal

(10) We make a wideband modulate of the signal such that the range resolution after pulse compression is 2.5 m and the compression ratio Kn=35 dB.

(11) One of the wideband modulation that is usually used in the radar system is the linear frequency modulation (LFM):

(12) $\begin{array}{cc}S\left(t\right)=e^{j2\left(f_{0}t+\frac{}{2}{t}^2\right)}& \left(1.1\right)\end{array}$

(13) where: t is the time; f.sub.0 is the frequency of carrier wave; is the LFM coefficient.

(14) The pulse compression output at receiver is:
S(t,f.sub.d)=.sub..sup.+s(t)S*(tt)e.sup.j2f.sup.d.sup.tdt(1.2)

(15) where: t is the time delay; f.sub.d is the doppler frequency; t is the time variable of echo signal.

(16) Since the modulated signal has a low peak side-lobe ratio, we proposed a compression method by using Nuttal weighted window. This method improved the peak side-lobe ratio and its result is expressed by:
S(t,f.sub.d)=W(t)*.sub..sup.+s(t)S*(tt)e.sup.j2f.sup.d.sup.tdt(1.3)

(17) where: W (t) is the Nuttal weighted window.

(18) The advantage of using wideband signal will be detailing reflex signal between small target (fishing boat having width about 2-5 m) and sea clutter. Disadvantage of using wideband modulation signal is large computing. Therefore we suggest using structure of high performance computing containing: The structure by combining FPGA and DDS synthesizer at the signal generator. The GPU with high performance computing (approximately 5 TFLOPS) for pulse compression and digital processing.

(19) Moreover, to guarantee that the system is fully synchronized and phase coherent, we used a system clock of 10 Mhz which has a very small error (0.01 ppm).

(20) To do this step, blocks used are a block of signal generator with wideband modulation and high resolution; a block of coherent signal generator with complex generator; a block of pulse compression with weighted window.

(21) Step 2: Target Clustering

(22) Doppler effect happens when a target is moving with reference to the radar station and the Doppler shift in frequency is directly proportional to radial velocity of targets.

(23) We use the moving target detector (MTD) to cluster all targets into 2 groups by their velocities. The first group consists of all slowly moving targets and the other group consists of fast moving targets. Since the clutters (weather, sea, land, . . . ) are included in the group of slowly moving targets, so in the next step we need to recognize slow moving targets and clutters in the first group.

(24) To do this step, we propose using the following blocks: a block of coherent intergration; a block of select doppler filter banks.

(25) Step 3: Detect Slowly Moving Targets in Clutter Environments.

(26) To detect targets in clutter environments we will use together three independent detectors. The outputs of these detectors are then combined to perform the final result. First detector: 3D OS-CFAR (3 Dimensional Ordered Statistical Constant False Alarm Rate). The powers of echo signals are arranged into a cube in three dimensional space (range, azimuth, doppler). For each cell under test (CUT) in this cube, all powers in the reference cells are arranged as an increased sequence, and then the optimal threshold is estimated. This method has a better performance in comparison with 1D CFAR. Second detector based on the clutter distributions. All power values of the radar echo signals in the considered area are fitted by a distribution. The test of goodness-of-fit is done by using chi-square test. For example, sea clutter has a lognormal distribution; weather clutter is followed by Rayleigh distribution while ground clutter has a Weibull distribution. From the clutter distribution and probability of false alarm (Pfa) we obtain the threshold for target detection by
threshold=F.sup.1(1Pfa),

(27) where F is the cumulative distribution function for clutters and F.sup.1 is the inverse function of F. Third detection is based on clutter map. The clutter map is performed automatically using digital filtering. The target's echo signals are compared with this map to classify exactly these echo signals (from clutters or from real targets).

(28) By the disclosed systems and methods improved radar detection capabilities for slowly moving targets in clutter is provided.