CN102353939B - Improved constant false alarm method - Google Patents

Abstract

The invention discloses an improved constant false alarm method in the technical field of signal processing. According to the method, data serially output by a fast Fourier transform (FFT) module are subjected to constant false alarm detection one by one; if a useful signal is discovered, the skip is carried out; the useful signal is not introduced into mean value calculation; the two-side constant false alarm detection is adopted for a signal; and the signal which is detected by the constant false alarm detection and the two-side constant false alarm detection to be the useful signal is a true useful signal. The improved constant false alarm method has high recognition precision for the signal; the problem of multi-target detection is solved; furthermore, an algorithm has quite high stability in various environments; further, the hardware cost is low; and the improved constant false alarm method has extremely high actual application value.

Description

A kind of improved CFAR Methods

Technical field

The invention belongs to the signal processing technology field, relate in particular to a kind of improved CFAR Methods.

Background technology

At present, the major function of car radar is to detect exactly all objects in the effective range of radar.Usually, if the environment that detects is spacious, and only having an object to be placed in the dead ahead of radar, is easily to the detection of target so.In this case, the reflection wave signal of target object can compare with the inner fixing threshold value of radar system, if reflection wave signal intensity greater than threshold value, just is identified as useful signal.Yet in the radar application environment of reality, target object always is in the environment that various somes faces reflection clutters are arranged, and the position of these background clutters on frequency spectrum in time and spatial position change.Be directed to a kind of like this phenomenon, a kind of technology that can adjust to the white adaptation of the distribution of the Strength Changes on frequency spectrum according to clutter compare threshold just seem extremely important.Such technology can guarantee that signal processing circuit can detect useful signal in the environment of any clutter background.A kind ofly like this can guarantee that in any clutter background the technology of constant false drop rate is called CFAR CFAR.

The main thought of CFAR algorithm is according to all multifrequency point of certain frequency D periphery { x1, x2 ... the signal intensity that xn} is corresponding judges that signal corresponding to this frequency D is useful signal or noise signal.As seen, at the diverse location of frequency spectrum, this threshold value relatively is all different, can guarantee that false drop rate is constant in any clutter background.CFAR is widely used in the digital radar baseband system, usually coordinates fast fourier transform FFT and peak detection block COG jointly to use.The basic procedure that signal is processed is as follows:

The signal of radio-frequency front-end after analog to digital conversion serial input in the RAM of fast fourier transform FFT module, through computing, real part and the imaginary part of fast fourier transform FFT module serial output frequency-region signal, ask again mould to process, call afterwards CFAR CFAR module and filter out the frequency range of useful signal and determine the peak value number.Data detect at peak value the peak value that detects each useful signal in the COG module, and the frequency that peak value is corresponding deposits in RAM, the digital base band processor after being convenient to.

At present more common CFAR CFAR technology has two kinds in the world, and a kind of is average CFAR method CA-CFAR, is exactly with the average of the signal amplitude of certain frequency range threshold value as a comparison.A kind of is sequence CFAR method OS-CFAR, gets exactly the little number of k (k should less than the contained frequency number of this frequency range) threshold value as a comparison after the signal amplitude sequence with certain frequency range.

All there is certain defect in these two kinds of algorithms:

1. the major defect of average CFAR method CA-CFAR:

During adjacent multi-target detection, useful signal may be brought in the arithmetic mean computing, has a strong impact on result.

2. the major defect of sequence CFAR method OS-CFAR:

The hardware spending that algorithm is realized is larger, needs a large amount of comparers.When in the face of serial input data, data of every input will be to the rearrangement of the data in all registers, and operation time is longer.

Summary of the invention

For mentioning the deficiencies such as existing method is not high to the accuracy of signal judgement, hardware spending is large in the above-mentioned background technology, the present invention proposes a kind of improved CFAR Methods.

Technical scheme of the present invention is that a kind of improved CFAR Methods is characterized in that the method comprises the following steps:

Step 1: the average of a front setting amplitude that frequency n is corresponding of obtaining the Fast Fourier Transform (FFT) FFT output of given frequency N is Z;

The product of step 2: computation of mean values Z and control signal T is designated as threshold value S;

Step 3: the next frequency of a described front setting frequency n of temporary Fast Fourier Transform (FFT) FFT output, and to be designated as amplitude corresponding to this frequency be serial input signals D; With described serial input signals D and threshold value S relatively, if described serial input signals D greater than threshold value S, execution in step 4; Otherwise, execution in step 5;

Step 4: the described serial input signals D of mark is useful signal, and makes n=n+1, returns to step 3;

Step 5: the described serial input signals D of mark is noise signal;

Step 6: compare the amplitude maximal value from continuous useful signal, the frequency that this amplitude maximal value is corresponding is designated as frepeak;

Step 7: if frepeak less than (N-1)/2, makes n=n+1, return to step 3; If the input the frequency number greater than (N-1)/2, execution in step 8;

Step 8: compare frepeak and N-1-frepeak, obtain frequency corresponding to required frequency peak.

Described step 8 is specially:

Frepeak is deposited in RAM, and be labeled as RAMpeak;

If RAMpeak＜N-1-frepeak reads the address constant, write address is constant, does not write;

If RAMpeak=N-1-frepeak reads address decrement, write address subtracts 1, does not write;

If RAMpeak＞N-1-frepeak, RAM writes 0, reads address and write address and subtracts simultaneously 1, gets RAMpeak and N-1-frepeak relatively, until front two kinds of situations occur;

Finish when the read/write address of RAM is 0, the non-zero value that deposit in RAM this moment is exactly frequencies corresponding to all detected frequency spectrum upward peak.

The inventive method has the following advantages:

1. has the multi-target detection ability;

2. hardware spending is low;

3. arithmetic speed is high;

4. can be used for parallel data and process, can be used for again serial data and process;

5. can coordinate FFT and COG module, be applied in the digital radar baseband system.

Description of drawings

Fig. 1 is structure principle chart of the present invention.

Embodiment

Below in conjunction with accompanying drawing, preferred embodiment is elaborated.Should be emphasized that, following explanation is only exemplary, rather than in order to limit the scope of the invention and to use.

In order to solve in the lump Main Problems in average CFAR method CA-CFAR and sequence CFAR method OS-CFAR, the present invention proposes a kind of improved CFAR Methods, solved all problems that average CFAR method CA-CFAR and sequence CFAR method OS-CFAR run into fully.

The core concept of algorithm:

Data to fast fourier transform FFT module serial output are done CFAR detection one by one, if the discovery useful signal is just skipped this signal, do not bring in mean value computation, can get rid of so the high amplitude of useful signal to the impact of threshold value, still can keep CFAR in target-rich environment.

Structure principle chart of the present invention is as shown in Figure 1:

For the data mag (the intensity amplitude that certain frequency fre is corresponding) from Fast Fourier Transform (FFT) FFT serial output, front n data to Fast Fourier Transform (FFT) FFT output deposit in First Input First Output fifo register group successively.Since n+1 data, the data of Fast Fourier Transform (FFT) FFT output will be temporary in (namely amplitude corresponding to these data is serial input signals D) in register D.Simultaneously n register value in the fifo register group do sums on average (available totalizer realization) obtain Z, multiplying each other with control signal T obtains threshold value S, i.e. S=Z * T.Serial input signals D and S are compared:

1. if serial input signals D is greater than threshold value S, this serial input signals D is useful signal, and the output port Alb of comparer sets to 0, and serial input signals D does not write in First Input First Output fifo register group, and First Input First Output fifo register group is not shifted; Serial input signals D delivers to peak value and detects the COG module; When the output port Alb of comparer was negative edge, peak value detected the COG module and starts working, and when the output port Alb of comparer was rising edge, peak value detected COG module end-of-job, obtained peak value and the frequency frepeak corresponding with it of signal amplitude;

2. if serial input signals D is less than threshold value S, this serial input signals D is noise signal, and the output port Alb of comparer puts 1, and serial input signals D writes in First Input First Output fifo register group, the displacement of First Input First Output fifo register group; Serial input signals D is not sent to peak value and detects the COG module, and this moment, relatively detection peak detected data in the COG module, and frequency corresponding to the maximal value that obtains is frepeak.

The data of FFT serial output are symmetrical about center frequency point on frequency domain.Adopting like this present invention to sweep from the 0th frequency to N-1 frequency is equivalent to for one time first sweep to (N-1)/2 a frequency another mistake to get back to the 0th frequency from (N-1)/2 frequency point scanning from the 0th frequency.

According to top analysis, although the present invention may be useful signal with the noise flase drop in " climbing " on frequency spectrum, the present invention does not have such problem in " descending ".So for all signals between the 0th frequency to the (N-1)/2 frequency, only have twice CFAR detection all to detect and be only real useful signal for useful signal.So just, got rid of the problem that has potential error signal, robustness is stronger in actual applications to make it.

As shown in Figure 1, the inventive method has a comparison control circuit.This comparison control circuit comprises a comparer and an Address generate module, and this module generates the read/write address of RAM, and contrl is the Enable Pin of Address generate module.The specific works flow process of this comparison control circuit is:

When inputting the frequency number less than (N-1)/2, contrl=1, Address generate forward generates the write address of RAM, and during the Alb negative edge, the address increases one.

When inputting the frequency number greater than (N-1)/2, contrl=0, the read/write address of the reverse generation of Address generate RAM.During the Alb rising edge, read RAMpeak (being frepeak) and compare with N-1-frepeak from RAM:

(1) if RAMpeak＜N-1-frepeak reads the address constant, write address is constant, does not write;

(2) if RAMpeak=N-1-frepeak reads address decrement, write address subtracts 1, does not write;

(3) if RAMpeak＞N-1-frepeak reads the address after RAM is write 0 and write address subtracts 1 simultaneously, then get RAMpeak and N-1-frepeak relatively, until front two kinds of situations occur.

(4) finish when the read/write address of RAM is 0, the non-zero value that deposit in RAM this moment is exactly frequencies corresponding to all detected frequency spectrum upward peak.

The present invention has successfully solved the multi-target detection problem, and the stability of this algorithm is all very high under various environment.The hardware cost of hardware cost of the present invention and average CFAR method CA-CFAR is substantially suitable, and many comparers and register control system, and this hardware spending can be ignored with respect to the fifo register group.So the present invention has the characteristics of the low hardware spending of the multi-target detection ability of sequence CFAR method OS-CFAR and average CFAR method CA-CFAR concurrently, has high actual application value.

The above; only for the better embodiment of the present invention, but protection scope of the present invention is not limited to this, anyly is familiar with those skilled in the art in the technical scope that the present invention discloses; the variation that can expect easily or replacement are within all should being encompassed in protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection domain of claim.

Claims (1)

1. improved CFAR Methods is characterized in that the method comprises the following steps:

Step 1: the average of a front setting amplitude that frequency n is corresponding of obtaining the Fast Fourier Transform (FFT) FFT output of given frequency N is Z;

The product of step 2: computation of mean values Z and control signal T is designated as threshold value S;

Step 3: the next frequency of a described front setting frequency n of temporary Fast Fourier Transform (FFT) FFT output, and to be designated as amplitude corresponding to this frequency be serial input signals D; With described serial input signals D and threshold value S relatively, if described serial input signals D greater than threshold value S, execution in step 4; Otherwise, execution in step 5;

Step 4: the described serial input signals D of mark is useful signal, and makes n=n+1, returns to step 3;

Step 5: the described serial input signals D of mark is noise signal;

Step 6: compare the amplitude maximal value from continuous useful signal, the frequency that this amplitude maximal value is corresponding is designated as frepeak;

Step 7: if frepeak less than (N-1)/2, makes n=n+1, return to step 3; If the input the frequency number greater than (N-1)/2, execution in step 8;

Step 8: compare frepeak and N-1-frepeak, obtain frequency corresponding to required frequency peak;

Be specially:

Frepeak is deposited in RAM, and be labeled as RAMpeak;

If RAMpeak＜N-1-frepeak reads the address constant, write address is constant, does not write;

If RAMpeak=is N-1-frepeak, read address decrement, write address subtracts 1, does not write;

If RAMpeak〉N-1-frepeak, RAM writes 0, reads address and write address and subtracts simultaneously 1, gets RAMpeak and N-1-frepeak relatively, until front two kinds of situations occur;

Finish when the read/write address of RAM is 0, the non-zero value that deposit in RAM this moment is exactly frequencies corresponding to all detected frequency spectrum upward peak.

Images