KR102415969B1 - Denoising method and apparatus for fmcw radar signal based on deep-learning - Google Patents

Abstract

A method of removing noise from a received signal of a frequency modulated continuous wave (FMCW) radar apparatus according to an embodiment of the present disclosure includes generating a frequency domain radar signal by Fourier transforming a radar signal received through an antenna, frequency domain radar Detecting a target region in which a target exists from a signal, converting a signal of a frequency domain other than the target region in a frequency domain radar signal to a zero value, and a frequency domain in which at least a partial frequency region is converted to a zero value The method may include generating a time domain radar signal by inverse Fourier transforming the radar signal.

Description

DENOISING METHOD AND APPARATUS FOR FMCW RADAR SIGNAL BASED ON DEEP-LEARNING

The present disclosure relates to a method and apparatus for removing noise from a Frequency Modulated Continuous Wave (FMCW) radar signal.

Radar sensors are being used not only in security and national defense, but also in transportation and space. In contrast to light sensors such as camera sensors that do not operate stably in a specific environment, radar sensors use radio waves. Due to this, it has the advantage of overcoming the disadvantages of the light-based sensor.

The maximum detectable distance of a radar sensor is determined by various parameters such as the period between transmit pulses and transmit power. Signal to noise ratio (SNR) can be increased as a method to extend the detectable distance. Signal to noise ratio increases the intensity of transmit power, reduces the intensity of noise components added to the received signal, or processes There are ways to increase the gain, etc.

However, increasing the intensity of transmission power has a problem in that it is difficult to introduce it to all equipment due to changes in hardware characteristics and limitations in radio wave regulations.

Prior art 1: Korean Patent Publication No. 10-2132296 (Notice on Aug. 8, 2020) Prior art 2: Korean Patent Publication No. 10-2016-0083947 (published on July 12, 2016)

An embodiment of the present disclosure provides a method and apparatus for reducing noise of an FMCW radar reception signal.

Another embodiment of the present disclosure provides a method and apparatus for reducing noise of an FMCW radar reception signal based on an auto-encoder.

An embodiment of the present disclosure provides an apparatus and method for removing noise of an FMCW radar signal reflected from a target.

A method of removing noise from a received signal of an FMCW radar device according to an embodiment of the present disclosure includes generating a frequency domain radar signal by Fourier transforming a radar signal received through an antenna, wherein a target exists in the frequency domain radar signal Detecting a target region, converting a signal of a frequency domain other than the target region in a frequency domain radar signal to a zero value, and converting the frequency domain radar signal obtained by converting at least some frequency domain into a zero value inverse Fourier converting to generate a time domain radar signal.

The method of removing noise from a received signal of an FMCW radar apparatus according to an embodiment of the present disclosure further includes generating a final radar signal by inputting a time domain radar signal to an autoencoder, wherein the autoencoder is It may be trained to reduce the difference between the output third radar signal and the first radar signal by inputting a second radar signal in which noise is added to the first radar signal in the region.

The detecting of the target area in the method for removing noise of the received signal of the FMCW radar apparatus may include setting a frequency domain radar signal, where signals greater than or equal to a first threshold value exist, as the target area.

The step of detecting the target area in the method for removing noise of the received signal of the FMCW radar device includes: moving a window having a preset frequency interval in the frequency domain radar signal according to the frequency; The method may include calculating a dispersion value of the magnitude of the radar signal and setting the target area based on a change in the dispersion value of the magnitude of the frequency domain radar signal.

at least one antenna for transmitting a first radar signal and receiving a second radar signal on which the first radar signal is reflected;

The FMCW radar signal processing apparatus according to an embodiment of the present disclosure includes at least one processor, a memory electrically connected to the processor and storing at least one code executed by the processor, and the memory is provided through the processor. When executed, the processor Fourier transforms the second radar signal received through the antenna to generate a frequency domain radar signal, detects a target area in which a target exists in the frequency domain radar signal, and other than the target area in the frequency domain radar signal It is possible to store a code causing the frequency domain signal to be converted to a zero value and the frequency domain radar signal obtained by converting at least a partial frequency domain to a zero value to be inverse Fourier transformed to generate a time domain radar signal.

The noise removal method and apparatus of the FMCW radar apparatus according to an embodiment of the present disclosure may reduce the total noise power of the FMCW reception radar signal by zero-converting signals in a region other than the target region sensed in the frequency domain.

Another embodiment of the present disclosure may further reduce the noise power of the FMCW received radar signal by using auto-encoder-based deep learning.

1 is a diagram illustrating an environment for performing a noise removal method of an FMCW radar apparatus or driving an FMCW radar signal processing apparatus according to an embodiment of the present disclosure.
2 is a block diagram illustrating a configuration of an FMCW radar signal processing apparatus according to an embodiment of the present disclosure.
3 is a flowchart illustrating a method for removing noise of an FMCW radar apparatus according to an embodiment of the present disclosure.
4 is a flowchart illustrating a method for removing noise of an FMCW radar apparatus according to an embodiment of the present disclosure.
5 is a view for explaining a noise removal method of the FMCW radar apparatus according to an embodiment of the present disclosure.
6 is a diagram for explaining a method of removing noise of a window-based FMCW radar apparatus according to an embodiment of the present disclosure.
7 is a view for explaining a training method of an auto-encoder of a noise removal method of an FMCW radar apparatus according to an embodiment of the present disclosure.

Hereinafter, the embodiments disclosed in the present specification will be described in detail with reference to the accompanying drawings, but the same or similar components are assigned the same reference numbers regardless of reference numerals, and redundant description thereof will be omitted. The suffixes "module" and "part" for components used in the following description are given or mixed in consideration of only the ease of writing the specification, and do not have distinct meanings or roles by themselves. In addition, in describing the embodiments disclosed in the present specification, if it is determined that detailed descriptions of related known technologies may obscure the gist of the embodiments disclosed in this specification, the detailed description thereof will be omitted. In addition, the accompanying drawings are only for easy understanding of the embodiments disclosed in the present specification, and the technical idea disclosed herein is not limited by the accompanying drawings, and all changes included in the spirit and scope of the present invention , should be understood to include equivalents or substitutes.

Terms including an ordinal number such as 1st, 2nd, etc. may be used to describe various elements, but the elements are not limited by the terms. The above terms are used only for the purpose of distinguishing one component from another.

When an element is referred to as being “connected” or “connected” to another element, it is understood that it may be directly connected or connected to the other element, but other elements may exist in between. it should be On the other hand, when it is said that a certain element is "directly connected" or "directly connected" to another element, it should be understood that the other element does not exist in the middle.

An environment for performing the noise removal method of the FMCW radar apparatus or driving the FMCW radar signal processing apparatus according to an embodiment of the present disclosure will be described with reference to FIG. 1 .

1 (a), the environment for performing the noise removal method of the FMCW radar apparatus or driving the FMCW radar signal processing apparatus according to an embodiment of the present disclosure includes a transmission 143 and a reception antenna 144 It may include a radar signal processing apparatus 100 equipped with an antenna.

The radar signal processing apparatus 100 may transmit (Tx) a radar signal to an area where the target 200 is estimated to exist, and may receive (Rx) a radar signal reflected from the target 200 .

The radar signal processing apparatus 100 may reduce the power of noise of the received radar signal, and for this purpose, pre-processing of the received radar signal, such as ADC, and signal processing such as FFT (Fast Fourier Transform) conversion may be performed. .

Preprocessing and signal processing may be configured to process all of them in at least one processor, or may be configured to be processed in each separate processor.

In another embodiment, the method of removing noise from the radar signal may be performed in an apparatus separate from the apparatus for receiving the radar signal reflected from the target. In this case, the received radar signal may be transmitted through a network to a separate device after pre-processing such as ADC, and the device may be referred to as the radar signal processing device 100 .

The radar signal processing apparatus 100 may apply a deep learning-based auto-encoder to a radar signal in the time domain for noise reduction, and is performed by a processor performing signal processing or a separate machine learning transmission processor can do. Hereinafter, a training method of the auto encoder for noise reduction of the FMCW radar signal will be described in detail with reference to FIG. 7 .

A configuration of the radar signal processing apparatus 100 according to an embodiment of the present disclosure will be described with reference to FIG. 2 .

In an embodiment, the radar signal processing apparatus 100 may include the communication unit 110 for receiving the digitally converted FMCW radar signal from the radar signal receiving apparatus when the radar signal receiving apparatus is separately implemented.

The communication unit 110 may include a communication module 111 for transmitting and receiving data with an external device, and the communication module 111 includes at least one of a mobile communication module, a wireless Internet module, a short-range communication module, and a location information module. can do.

The mobile communication module includes technical standards or communication methods for mobile communication (eg, GSM (Global System for Mobile communication), CDMA (Code Division Multi Access), CDMA2000 (Code Division Multi Access 2000)), EV-DO ( Enhanced Voice-Data Optimized or Enhanced Voice-Data Only), Wideband CDMA (WCDMA), High Speed Downlink Packet Access (HSDPA), High Speed Uplink Packet Access (HSUPA), Long Term Evolution (LTE), Long Term (LTE-A) Evolution-Advanced), etc.), transmits and receives a radio signal with at least one of a base station, an external terminal, and a server on a mobile communication network.

The wireless Internet module refers to a module for wireless Internet access, and may be built-in or external to the radar signal processing apparatus 100 . The wireless Internet module is configured to transmit and receive wireless signals in a communication network according to wireless Internet technologies.

As wireless Internet technology, for example, WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Wi-Fi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), WiBro (Wireless Broadband), WiMAX (World Interoperability for Microwave Access), High Speed Downlink Packet Access (HSDPA), High Speed Uplink Packet Access (HSUPA), Long Term Evolution (LTE), Long Term Evolution-Advanced (LTE-A), and the like.

The short-range communication module is for short-range communication, and includes Bluetooth™, Radio Frequency Identification (RFID), Infrared Data Association (IrDA), Ultra Wideband (UWB), ZigBee, and Near Field (NFC). Communication), Wi-Fi (Wireless-Fidelity), Wi-Fi Direct, and Wireless USB (Wireless Universal Serial Bus) technology may be used to support short-distance communication.

The radar signal processing device 100 may be implemented in the form of a server device or may be implemented in a form including a processor and memory such as a laptop computer, a personal computer, a tablet computer, etc., may be implemented as one module such as a vehicle, and drive the processor Therefore, if it is a computing device capable of signal processing, the type is not particularly limited.

The radar signal processing device 100 displays or transmits a result of recognizing a data processing process or operation to a user or a control device (eg, a central processing unit of a vehicle), or receives an input from a user or an interface that the user can control The unit 120 may be included. The interface unit 120 may include a touch type or mechanical button 121 , a display 122 , an LED capable of outputting light, or a speaker 123 capable of outputting a voice.

As described with reference to FIG. 1 , the radar signal processing apparatus 100 may be implemented to include an antenna 142 for transmitting or receiving a radar signal, and in this case, to transmit (Tx) or receive (Rx) a radar signal It may include at least one or more antennas 143 and 144 for may include Since the configuration of hardware or software for transmitting or receiving a radar signal is known to those skilled in the art, a detailed description thereof will be omitted herein.

The radar signal processing apparatus 100 stores intermediate or final data, such as a received radar signal and a radar signal converted to a frequency domain, or a machine learning-based trained learning model capable of removing noise by applying it to a radar signal in a time domain. It may include a memory 130 for storing.

Although not shown in FIG. 2 , the radar signal processing apparatus 100 is a separate interface that serves as a passage with various types of external devices, and includes a wired/wireless data port, a memory card port, and an identification module. At least one of a port for connecting a device equipped with this device, an audio I/O (Input/Output) port, a video I/O (Input/Output) port, and an earphone port can do.

The signal processing unit 141 or the processor 180 of the radar signal processing apparatus 100 may perform a Fourier transform (FFT) or an inverse Fourier transform (IFFT) on the received radar signal.

A method of removing noise of a radar signal processing apparatus according to an embodiment of the present disclosure will be described with reference to FIG. 3 .

The radar signal processing apparatus may receive or receive a radar signal reflected from a plurality of regions in which the target is estimated to exist at different times.

The radar signal processing apparatus may generate a frequency domain radar signal by Fourier transforming the received radar signal ( S110 ).

Thereafter, the radar signal processing apparatus may detect a target region in which a target exists in the frequency domain radar signal (S120), and convert a signal in a frequency domain other than the target region in the frequency domain radar signal into a zero value (S130). ).

It will be described in detail with reference to FIGS. 4 to 5 .

Referring to FIG. 4 , according to an embodiment, the radar signal processing apparatus sets an area in which signals greater than or equal to a first threshold value exist in a frequency domain radar signal as a target area, and converts a signal in a frequency domain other than the target area into a zero value. can

The first threshold value may be preset by an experiment or may be set based on a change in the magnitude of a signal of a frequency domain radar signal. For example, a peak having a high signal is detected, and the highest peak value among peaks except for peaks in which the difference between the averages of the signal values in a frequency region of a certain magnitude around each peak is greater than or equal to a preset range is set as the first threshold. can

Referring to FIG. 5 , in an embodiment, the radar signal processing apparatus moves a window S121 having a preset frequency interval in a frequency domain radar signal according to a frequency ( S123 ), and a frequency domain radar signal within the window moved according to the frequency It is possible to calculate a dispersion value of the magnitude of ( S125 ) and set it as a target region based on a change in the dispersion value of the magnitude of the frequency domain radar signal ( S127 ).

For example, referring to FIG. 6 , while moving windows 610 , 620 , and 630 having the same frequency domain width according to the frequency ( 640 ), the dispersion value of the frequency domain radar signal 650 within the window at the location of the window can be calculated.

Thereafter, the radar signal processing apparatus sets a frequency domain in which the dispersion value of the magnitude of the frequency domain radar signal within the window is equal to or greater than a preset reference dispersion value as the target domain, or the dispersion value of the magnitude of the frequency domain radar signal within the window and the frequency within the previous window. When the change in the variance value of the size of the area radar signal is equal to or greater than a preset reference, the frequency area within the corresponding window may be set as the target area.

The radar signal processing apparatus may generate a time-domain radar signal by converting a signal in a frequency domain other than the target domain into a zero value in the frequency domain radar signal, and then inverse Fourier transforming the signal in the frequency domain again ( S140 ).

A training method of an auto-encoder for reducing noise of an FMCW radar signal according to an embodiment of the present disclosure will be described with reference to FIG. 7 .

The radar signal processing apparatus may generate a final radar signal from which noise is additionally removed by reducing noise in the frequency domain radar signal and then inputting the time domain converted radar signal to the auto encoder.

The auto-encoder may be an auto-encoder trained to reduce a difference between the output third radar signal and the first radar signal by inputting a second radar signal in which noise is added to the first radar signal in the time domain.

An autoencoder is a neural network that aims to reproduce an input itself as an output. The autoencoder includes an input layer, at least one hidden layer, and an output layer.

In this case, since the number of nodes in the hidden layer is smaller than the number of nodes in the input layer, the dimension of data is reduced, and thus compression or encoding is performed.

Also, the data output from the hidden layer is input to the output layer. In this case, since the number of nodes of the output layer is greater than the number of nodes of the hidden layer, the dimension of data is increased, and decompression or decoding is performed accordingly.

Meanwhile, the auto-encoder controls the neuron's connection strength through learning, so that the input data is expressed as latent feature data. The hidden layer expresses information with fewer neurons than the input layer, and being able to reproduce the input data as an output may mean that the hidden layer found and expressed hidden patterns from the input data.

)(710)에 노이즈를 추가(

)하여 입력층에 입력(720)할 수 있고, 입력층과 은닉층 사이에서 노이즈가 추가된 제2 레이더 신호가 인코딩 되고(

), 다시 출력층(740)으로 입력되면서 제3 레이더 신호로 디코딩(

) 된다. 이후, 미리 결정된 손실 함수에 기반하여 제1 레이더 신호와 제3 레이더 신호의 차이를 감소시키도록 오토 인코더를 훈련시킬 수 있다. Referring to FIG. 7 , the first radar signal in the time domain (

) add noise to (710)

) and input 720 to the input layer, the second radar signal with noise added between the input layer and the hidden layer is encoded (

), while being input to the output layer 740 again, decoded into a third radar signal (

) do. Thereafter, the auto-encoder may be trained to reduce the difference between the first radar signal and the third radar signal based on the predetermined loss function.

A plurality of hidden layers may exist, and the loss function may mainly use a mean squared error (MSE) or a cross entropy error (CEE), but the present invention is not limited thereto. A learning optimization algorithm can be used to minimize the loss function, and the learning optimization algorithms include Gradient Descent (GD), Stochastic Gradient Descent (SGD), Momentum, and Nesterov Accelerate (NAG). Gradient), Adagrad, AdaDelta, RMSProp, Adam, Nadam, etc. can be used.

The present disclosure described above can be implemented as computer-readable code on a medium in which a program is recorded. The computer-readable medium includes all types of recording devices in which data readable by a computer system is stored. Examples of computer-readable media include Hard Disk Drive (HDD), Solid State Disk (SSD), Silicon Disk Drive (SDD), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc. There is this. In addition, the computer may include the processor 180 of the radar signal processing apparatus.

Meanwhile, the program may be specially designed and configured for the present disclosure, or may be known and used by those skilled in the art of computer software. Examples of the program may include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like.

In the specification of the present disclosure (especially in the claims), the use of the term “above” and similar referential terms may be used in both the singular and the plural. In addition, when a range is described in the present disclosure, each individual value constituting the range is described in the detailed description of the invention as including the invention to which individual values within the range are applied (unless there is a description to the contrary). same as

The steps constituting the method according to the present disclosure may be performed in an appropriate order unless the order is explicitly stated or there is no description to the contrary. The present disclosure is not necessarily limited to the order in which the steps are described. The use of all examples or exemplary terminology (eg, etc.) in the present disclosure is merely for the purpose of describing the present disclosure in detail, and the scope of the present disclosure is limited by the examples or exemplary terms unless defined by the claims. it is not going to be In addition, those skilled in the art will appreciate that various modifications, combinations and changes can be made according to design conditions and factors within the scope of the appended claims or their equivalents.

Accordingly, the spirit of the present disclosure should not be limited to the above-described embodiments, and the scope of the spirit of the present disclosure as well as the claims to be described later are equivalent to or equivalently changed therefrom. will be said to belong to

100: radar signal processing unit
200: target
610, 620, 630: Windows
640: shift according to frequency
650: frequency domain radar signal

Claims (10)

A method of removing noise from a received signal of a Frequency Modulated Continuous Wave (FMCW) radar device, the method comprising:
generating a frequency domain radar signal by Fourier transforming the radar signal received through the antenna;
detecting a target area in which a target exists from the frequency domain radar signal;
converting a signal of a frequency domain other than the target domain from the frequency domain radar signal into a zero value; and
generating a time domain radar signal by inverse Fourier transforming the frequency domain radar signal obtained by converting at least a partial frequency domain into a zero value;
The step of detecting the target area comprises:
moving a window having a preset frequency interval in the frequency domain radar signal along a frequency axis;
calculating a variance value of the magnitude of the frequency domain radar signal within the window moved along the frequency axis; and
and setting, as the target region, a frequency region in which a dispersion value of the magnitude of the frequency-domain radar signal within the window is equal to or greater than a preset reference dispersion value, wherein the target region is the target region that reflects the transmitted radar signal. area,
A method of denoising the FMCW radar device.
According to claim 1,
The method further comprising the step of inputting the time domain radar signal to an autoencoder to generate a final radar signal,
The auto-encoder is trained to reduce a difference between a third radar signal outputted by inputting a second radar signal in which noise is added to a first radar signal in the time domain and the first radar signal,
A method of denoising the FMCW radar device.
delete delete delete A method of removing noise from a received signal of a Frequency Modulated Continuous Wave (FMCW) radar device, the method comprising:
generating a frequency domain radar signal by Fourier transforming the radar signal received through the antenna;
detecting a target area in which a target exists from the frequency domain radar signal;
converting a signal of a frequency domain other than the target domain from the frequency domain radar signal into a zero value; and
generating a time domain radar signal by inverse Fourier transforming the frequency domain radar signal obtained by converting at least a partial frequency domain into a zero value;
The step of detecting the target area comprises:
moving a window having a preset frequency interval in the frequency domain radar signal along a frequency axis;
calculating a variance value of the magnitude of the frequency domain radar signal within the window moved along the frequency axis; and
and setting the target area as the target area based on a change in the variance value of the size of the frequency domain radar signal within the window and the variance value of the size of the frequency domain radar signal within the previous window, wherein the target area is the transmitted radar. The area where the target that reflects the signal is present,
A method of denoising the FMCW radar device.
at least one antenna for transmitting a first radar signal and receiving a second radar signal on which the first radar signal is reflected;
at least one processor;
It is electrically connected to the processor and includes a memory in which at least one code (code) executed by the processor is stored,
When the memory is executed through the processor, the processor performs a Fourier transform on the second radar signal received through the antenna to generate a frequency domain radar signal, and detects a target area in which a target exists in the frequency domain radar signal, , In the frequency domain radar signal, a signal in a frequency domain other than the target domain is converted to a zero value, and the frequency domain radar signal obtained by converting at least a partial frequency domain into a zero value is inversely Fourier transformed to perform a time domain radar store the code causing the signal to be generated;
The memory moves a window having a preset frequency interval in the frequency domain radar signal along a frequency axis before the processor sets the target area as the target area, and the frequency domain radar signal within the window moved along the frequency axis. a code for calculating a magnitude dispersion value, and causing a frequency domain in which a magnitude dispersion value of the frequency domain radar signal within the window to be equal to or greater than a preset reference dispersion value is set as the target zone, wherein the target zone transmits The area in which the target that reflects one radar signal is present,
FMCW radar signal processing unit.
8. The method of claim 7,
The memory causes the processor to input the time domain radar signal to an autoencoder to generate a final radar signal, and the autoencoder inputs a second radar signal in which noise is added to the first radar signal in the time domain Further storing a code trained to reduce the difference between the output of the third radar signal and the first radar signal as
FMCW radar signal processing unit.
delete at least one antenna for transmitting a first radar signal and receiving a second radar signal on which the first radar signal is reflected;
at least one processor;
It is electrically connected to the processor and includes a memory in which at least one code (code) executed by the processor is stored,
When the memory is executed through the processor, the processor performs a Fourier transform on the second radar signal received through the antenna to generate a frequency domain radar signal, and detects a target area in which a target exists in the frequency domain radar signal, , In the frequency domain radar signal, a signal in a frequency domain other than the target domain is converted to a zero value, and the frequency domain radar signal obtained by converting at least a partial frequency domain into a zero value is inversely Fourier transformed to perform a time domain radar store the code causing the signal to be generated;
The memory moves a window having a preset frequency interval in the frequency domain radar signal along a frequency axis before the processor sets the target area as the target area, and the frequency domain radar signal within the window moved along the frequency axis. A code for calculating a magnitude variance value, and setting it as the target area based on a change in a variance value of the magnitude of the frequency domain radar signal within the window and a variance value of the magnitude of the frequency domain radar signal within a previous window further storing, wherein the target area is an area in which the target that reflects the transmitted radar signal exists,
FMCW radar signal processing unit.

Images