RU2519443C1 - Method of detecting reflected signal during radar location - Google Patents

Abstract

FIELD: radio engineering, communication.

SUBSTANCE: method of detecting a reflected signal at one or more points in space involves periodically emitting ultra-wideband pulses using at least one transmitter on a signal from at least one clock generator connected to the corresponding transmitter; using receivers, at more than one point in space, to pick up electrical signals, recording and storing said signals in memory; processing the picked up signals using a processing unit; using at least one receiver to pick up instantaneous values of signals at instances determined by another corresponding clock generator connected to said receiver, with a pick up period different from the period of emitting pulses.

EFFECT: eliminating electrical clock circuits between spatially separated transmitters and receivers, reduced requirements for accuracy and stability of induced time delay, which simplifies the device for realising the method or reduces errors in stored data caused by imperfect clocking.

12 cl, 5 dwg

Description

Technical field

The invention relates to ultra-wideband (UWB) radar and can be used to solve problems requiring the determination of the three-dimensional shape of objects or determining the position of objects. An example of such tasks may be: screening, medical diagnostics, detection of intrusion into a protected area. The advantage of this approach is: the ability of radio waves to penetrate clothing, non-metallic walls, the human body, good resolution (up to a centimeter), the ability to work on reflection when all the components of the device are on one side of the object under study (vision through walls), harmless waves person.

In the task of conducting an inspection, it becomes possible in a few seconds to detect hidden objects under a person’s clothing or in bags, while the person does not need any special actions, and the harmlessness of radiation allows you to use them at objects for daily inspection.

In medical diagnostics, a tomograph based on the principles of ultra-wideband radio engineering can be used as an alternative or addition to ultrasound, X-ray or NMR imaging. The advantages of UWB tomography are harmlessness, low cost and other contrast compared to NMR tomography, which allows it to be used to detect cancerous tumors.

Detection of intrusion into a protected area and tracking movement in it is another task for which the proposed invention can be applied. This task consists not so much in determining the exact shape as in the two previous ones, but in detecting movement and approximating the shape and size of an object (to distinguish between a person, a dog, a rat).

A feature of ultra-wideband (UWB) radar is the use of short pulse signals without filling the carrier with a characteristic pulse duration of 1 ns or less. An example of a pulse (Gaussian monocycle) is shown in Fig. 1 (pos. 1).

Such signals have a very wide spectrum, comparable in order of magnitude with the operating frequencies. For example, regulatory agencies in the United States and Russia allow civilian applications with a radiation spectrum of 3 GHz-10 GHz.

The main advantages of ultra-wideband location technology are associated with a short pulse length. When propagating in space, a pulse with a duration of 100 ps has a length of 3 cm, which ensures high spatial resolution. In the case of radar, it becomes possible to distinguish between signals coming from objects, if the distance difference to them is a few centimeters; Due to the short pulse length, there is practically no “dead zone" near the device.

One of the most promising areas of application of this technology is "radio vision" - the determination of the three-dimensional shape of an object by sensing it with ultra-wideband pulses. In a simplified version, the surface shape of the object is determined. This may be required, for example, to detect weapons under clothing. In a more advanced version, the internal structure of the object is studied, for example, a human tomogram is made to diagnose diseases.

This problem breaks down into two significant parts - electronic and algorithmic. This application discloses a method for registering a reflected signal during radar, that is, it solves the electronic part. A useful result is the recording of the reflected signal in memory, which can be used to determine the three-dimensional shape of objects by digitally processing this data, that is, to solve the algorithmic part of the problem. Methods of processing data and constructing three-dimensional images are disclosed in patents US 5,668,555, US 8089396 and others. In the field of radio electronics, it is required to design a device that contains a certain number of transmitters and receivers, and the number of transmitters may differ from the number of receivers. The exact number of transmitters and receivers is determined by the specialist in algorithmic processing, based on the requirements for image quality and the requirements for the shape and size of the workspace, within which the three-dimensional shape of the object is determined.

State of the art

A known method of radio vision on the principles of ultra-wideband radio technology US 5,668,555, in which to record the reflected signal uses the method disclosed in US 5,361,070.

The method for recording the reflected signal during radar, disclosed in this patent (US 5,361,070), is implemented using a master reference generator (PM generator 20), which determines the period of emission of ultra-wideband pulses. A special unit contains controlled delay lines in front of each receiver and transmitter (24 and 28) through which the signal from the reference generator passes. Thus, the delay in front of the receiver (28) sets the pulse emission time relative to the front of the reference generator. The delay in front of the receiver (26) sets the moment of capture of the reflected signal in the sampling-storage device (SEC) (40). For one emitted pulse, one capture is made. By controlling the delay in front of the receiver, a range scan is performed.

The main disadvantage of this approach as applied to radio vision is the difficulty of using it in the presence of several receiving channels.

For the problem of radio vision, danger is also posed by random fluctuations in the signal transmission: the so-called phase jitter (jitter) and slow delay drift associated with a change in temperature. In the case of applications for an inspection cabin or other applications with a large antenna array, it is necessary to distribute the signal to many consumers over cables of a long order of a meter. This requires the use of output and input buffer amplifiers, each of which increases phase jitter and introduces a drift in the delay of the signal passage.

In addition, the stated patents mean the capture of only one point of the reflected signal, which reduces the pace of the device. Modern ADCs make it possible to capture hundreds of millions of samples per second, while the radiation frequency of probing pulses is usually made substantially lower than 100 MHz so that reflection of one pulse does not overlap another.

Disclosure of invention

The objective of the invention is to register on the principles of radar by several spaced transmitters and receivers and store in memory a set of digital data by which computer processing can restore a three-dimensional picture of the location of objects in a certain area of space.

To obtain an image, it is fundamentally necessary either to record a signal at a plurality of points, or radiation from a plurality of points, or a combination of both. There is a fundamental possibility of obtaining an image with a device with one transmitter and receiver, provided that it is mechanically moved through many points. But this option in many cases is unacceptably cumbersome, and most importantly, moving and scanning takes a long time. You can reduce the time by simultaneously recording a signal at several points by several receivers. For example, in the case of an inspection cabin, it is advisable to place 32 receivers vertically and organize horizontal mechanical movement of the receivers around the person with recording the reflected signal through every degree. As a result of using 32 receivers, the recording time is reduced by 32 times, it becomes possible to limit itself to mechanical movement in one dimension instead of two, which simplifies the design.

To take advantage of the simultaneous recording of several receivers, it is necessary to synchronize the receivers with the transmitters in the conditions of their spatial diversity. Considering that in order to obtain a practically demanded resolution of the order of a centimeter, a pulse duration of the order of 100 ps is required, the synchronization accuracy should be of the order of 10 ps or better.

The problem is solved in that in the method of recording the reflected signal during radar, which consists in the fact that

- at one or more points in space, ultra-wideband pulses are periodically emitted using at least one transmitter by a signal from at least one clock connected to the corresponding transmitter,

- at more than one point in space using receivers capture electrical signals, record and store them in memory,

- process the captured signals using the processing unit,

according to the invention, with the help of at least one receiver, the instantaneous values of the signals are captured, at the moments specified by the other corresponding clock generator connected to it, with a capture period different from the pulse emission period.

The technical result that allows us to solve the problem is to eliminate electrical synchronization circuits between spatially separated transmitters and receivers or to reduce the requirements for them in terms of accuracy and stability of the introduced time delay, which allows us to simplify the device for implementing the method or to reduce errors in stored data caused by not ideal synchronization.

The achievement of the technical result is facilitated by the following particular cases of the invention.

Preferably, the signal capture period Tr and the pulse emission period Tt are selected such that Tr / Tt is equal to the irreducible fraction M / N, where M and N are natural numbers, and a signal of duration Tt is recorded with a sampling step Tt / N by rearranging any N consecutive values of the captured signals stored in memory at which the value of the signal with serial number n in the original sequence of values of the captured signals stored in memory is put in place with the number n · M% N in the resulting sequence of these values , in-% where the - sign operation receiving remainder.

In addition, it is advisable to use the clock generators of the receivers, each of which is located in close proximity to the component of the corresponding receiver, which captures the reflected signal.

In addition, synchronize the operation of at least part of the transmitter clocks with the operation of at least part of the receiver clocks by synchronizing them with a common main generator.

In this case, clocks of transmitters and receivers are used, the frequency of each of which is a multiple of the frequency of the main generator.

In addition, in the captured signals, the pulse transmitted directly from one of the transmitters is identified, at least one of the values related to this pulse is analyzed, and a corrective effect on the receiver clock is generated to synchronize it with the clock of this transmitter.

In this case, the formation of the corrective action is carried out using a proportional-integral controller, the input signal of which is the value related to the pulse transmitted directly, and the residual is the difference between the actual captured value and the value that should be captured with perfect synchronization of the clock generators.

It is advisable to select the periods Tt and Tr so that the maximum time interval without the formation of a corrective action is Tr · N / 2 or less.

It is also possible that the pulse transmitted directly from one of the transmitters is identified in the captured signals, the synchronization deviation parameters of the clock generators of the respective transmitter and receiver are calculated from this pulse, and these parameters are used to compensate for the synchronization deviation.

In addition, to improve the signal-to-noise ratio of the recorded and stored signals, the captured values are averaged, the time interval between the moments of capture of which is equal to or a multiple of Tr · N.

It is possible that, to improve the signal-to-noise ratio of the recorded and stored signals, the values of the captured signals are averaged, the time interval between the moments of capture of which is equal to or a multiple of Tr · N, and one or more values of pulses transmitted directly from one of the transmitters and used to correct the clock generator receiver, not subject to averaging.

You can also use only one transmitter to emit ultra-wideband pulses and synchronize the clocks of all receivers with the clock of this transmitter.

We indicate the main orders of magnitude for a typical problem. The main application of the present invention involves an inspection cabin and a medical tomograph. Based on this, we assume that it is necessary to restore the shape of the object with an accuracy of 1.5 cm (for a tomograph it is better, but for an inspection cabin it can be worse). To scan the object, we will use pulses in the form of a Gaussian monocycle (similar to one sinusoidal period) such that 100 ps passes between the maximum of the positive half-wave and the maximum (in amplitude) of the negative. This signal is similar to one 5 GHz sine wave, and the main signal power will lie in the range from 1.5 GHz to 9 GHz. For 100 ps, the radio signal passes 3 cm, which means that with radar due to doubled passage this will correspond to a difference of distances of 1.5 cm, which will provide the required resolution.

The reflected signal has the same characteristic frequencies as the radiated one, so it is necessary to convert it to digital form with a sampling frequency of at least 20 GHz. Such high sampling rate requirements that are not available with modern integrated ADCs create serious barriers to using this technology. One of the problems solved by this invention is to record the reflected signal with the required frequency of tens of gigahertz using the existing integrated ADC with a sampling frequency of hundreds of megahertz.

The scanning range in the case of an inspection cabin is 1.5 m, that is, hundreds of readings in range are required.

The required number of transmitters and receivers is highly dependent on the problem and on the approaches to solving the algorithmic part of the problem. The ideal option is an antenna array that is commensurate with the observed object and with the step of placing the transmitters and receivers in both coordinates in order of magnitude equal to the desired resolution. In many cases, this number of lattice elements is unacceptable for economic reasons. For example, in the case of an inspection cabin, 1,000 to 10,000 elements will be required. In this case, a moderate number of elements (10-100) are combined with their mechanical movement.

All values above are given to simplify the understanding of the invention and to demonstrate the useful result achieved with its help, and in no case limit the scope of its application. The stated principles are applicable to solving other problems of radar with other characteristic values.

List of drawings

The invention is illustrated by the following drawings.

Figure 1 (a, b) shows the emitted UWB pulse used in the proposed method: (a) is the ideal pulse, (b) is the real pulse corresponding to the recorded pulse transmitted directly from the transmitter to the receiver.

Figure 2 shows a block diagram of the best embodiment of the device for implementing the proposed method, which implements the correction of the clock of the receiver by the pulse transmitted directly from the receiver.

Figure 3 (a, b) is a table illustrating the permutation of values when writing to memory in order to obtain a sequential recording of the signal.

Figure 4 shows a variant of a block diagram of an alternative embodiment of a device for implementing the proposed method, using electrical signals to synchronize the clock of the receiver with the transmitter.

Figure 5 shows a drawing of an ideal emitted pulse, which indicates the interval suitable for generating corrective action on the generator.

Description of the invention

A feature of ultra-wideband radar is the use of short pulse signals without filling the carrier with a characteristic pulse duration of 1 ns or less. Examples of such a pulse are shown in figure 1. For calculations and simulations, the impulse specified by the Gaussian monocycle function, the first derivative of the Gaussian distribution, is often used. The form of this function is shown in figure 1 (a). The scale is not indicated because the pulses can be of different amplitudes and durations. The real impulse differs from the ideal one, an example of a real impulse is shown in Fig. 1 (b). The real pulse pattern is a fragment of the recorded data corresponding to the direct passage of the pulse from the transmitter to the receiver. A large horizontal division corresponds to 500 ps, i.e., the pulse duration is about 200 ps.

The block diagram of the device for the best embodiment of the proposed method is shown in figure 2. Its distinguishing feature is the presence of its own generator for each transmitter and each receiver. The transmitter (9) is a radiator unit, which consists of a clock generator (1) of a digital periodic signal and a radio signal generator (2) connected to it. The periodic signal generator (1) generates a driving digital signal (101) with a frequency Ft, and a radio signal generator (2) generates a pulse along the active edge of this signal, and this pulse is emitted by the antenna (3) connected to it. Since the signal is emitted with a period Tt = 1 / Ft, the reflected signal also has the same period, if we neglect the movement of objects relative to the emitters and receivers. The recording of the period of the reflected signal is carried out over a time of the order of units and tens of microseconds, therefore, all objects can be considered stationary (and the antenna is also in the case of mechanical scanning).

In a preferred embodiment, there are several transmitters (9), but to simplify the processing algorithms, only one of them works at any time. That is, one of the transmitters (9) is activated, the period of the reflected signal is recorded, and then another transmitter (9) is activated. In a preferred embodiment, to reduce the number of emitters and receivers, they mechanically move around the object and scan the object from various positions. Therefore, in a preferred embodiment, the recording cycle of the reflected signal for each emitter is repeated at a new position of the transmitters and receivers.

Each receiver (29) consists of an antenna (25), series-connected devices for sampling and storage (UVX) (22), analog-to-digital converter (ADC) (23), a digital processing unit (24) and a clock generator (21) periodic signal, also connected to the IWC (22). The generator of the periodic signal (21) generates a driving digital signal (202) with a frequency of Fr, along the active edge of this signal, the I / O signal (22) remembers the instantaneous value of the radio signal (voltage) supplied to the input of the I / O signal (22) from antenna (25). This instantaneous value is stored in analog form in the I / O (22) until the next edge of the generator signal (21), during which time the analog value (203) from the output of the I / O (22) is converted by the ADC (23) into digital form and transmitted to the digital unit (24) ) processing with subsequent storage in memory.

In a preferred embodiment, generators (1) and (21) are used having a sufficiently low phase noise in order to obtain the required spatial resolution. To ensure a low temporal jitter of the transmitter (9) and receiver (29), the generator (1) or (21) and the device controlled by it are located as close as possible to each other, avoiding intermediate elements, while other techniques known in electronics are used to minimize jitter .

The digital processing unit (24) accumulates for some time the data coming from the ADC (23), and executing the algorithm described below, it receives a digital record of the period of the reflected signal with a duration of Tt = 1 / Ft. The resulting signal record is equivalent to period sampling with a frequency Fd equal to the least common multiple of Ft and Fr: Fd = HOK (Ft, Fr). Moreover, in order to accumulate the required number of ADC samples, recording the entire period requires a time equal to the reciprocal of the largest common divisor Ft and Fr.

In order to obtain a record of the signal period, no matching of the two generators (1) and (21) (transmitter and receiver) in phase is required. You can start the operation of recording the period at any time with an arbitrary shift of the active front of the generator (1) of the transmitter relative to the active front of the generator (21) of the receiver, however, the period will be accumulated over the specified time. However, randomness in the moment of recording start leads to uncertainty of the starting point of the recorded signal. The recorded data array can start from an arbitrary point in the period, or rather, it starts from the point that will be accidentally captured at the first step of the algorithm.

It is required to obtain a record of the signal in a timeline tied to the conditional moment of the beginning of the reflected signal. As such a moment, we choose the transition through zero of the signal transmitted directly from the transmitter (9) to the receiver (29), or more precisely, the transition through zero of its part between the positive and negative extremes (an ideal Gaussian monocycle has only one transition through zero, but the real signal may have several).

For this, in the obtained record of the signal period by digital processing methods, there are points corresponding to the direct passage of the emitted pulse from the transmitter (9) to the receiver (29), characterized by the maximum amplitude and known pulse shape. Further, in the interval from the positive maximum to the negative maximum of the signal, there is a transition through zero.

The position of this conditional beginning of the reflected signal in the memory is determined by the position of this transition through zero in the recorded signal image. This allows you to cyclically shift the recorded signal in the memory to obtain a record starting from the conditional moment of the beginning of the reflected signal.

The control unit (24) also performs the function of phase locked loop of the frequency of the generator (21). It determines the phase deviation of the generator (21) of the receiver by the direct signal from the transmitter (9) to the receiver (29) and makes a correction by controlling the generator (21) using the signal (201). This allows us to ensure strict correspondence of the given frequency ratio of the transmitter (9) and receiver (29) and minimize errors caused by phase noise of the generators.

Frequency selection in a simplified embodiment

The frequencies Ft and Fr are selected based on the requirements of a particular embodiment. In a simplified embodiment, the frequencies are selected as follows. The initial data are set: the maximum viewing range, the maximum frequency of operation of the UVX (22) and ADC (23), the required time step with which it is necessary to record the reflected signal.

First, the required time step is determined based on the requirements for the resolution of the device and the features of the algorithm for restoring the three-dimensional shape of the object.

The frequency Fr is chosen as large as possible, as far as the selected I / O (22), ADC (23) and digital block (24) allow. In this case, such a frequency is selected that its period Tr = 1 / Ft is a multiple of the required step Ts in time: Tr = n · Ts. Since Ts is usually set to less than 100 ps, and Tr is, at best, a few nanoseconds, the multiplicity requirement does not impose significant restrictions on Tr.

The frequency Ft is selected according to two criteria. First: the propagation time of the radio signal to the maximum range and vice versa should be less than the period of this frequency. Second: the reflected signal from the previous pulse should decay, that is, return with a negligible amplitude by the time the capture from the next pulse occurs. In most cases, the second criterion is more stringent. For example, in the case of an inspection cabin, the maximum range corresponds to the maximum range to the scanned object, and a sufficiently strong reflection can come from the opposite wall of the cabin, so the pulse repetition period is selected several times longer than the propagation time of the signal to the scanned object and vice versa. Let the frequency of the emitted pulses be selected, let us denote its period Tt = 1 / Ft. We slightly adjust Tt or Tr so that Tt = k · Tr-Ts, for a whole k. The meaning of k is the approximate number of UVC samples (22) for the period of pulse emission. Due to the term Ts in this formula, in the next period, all k samples are made on Ts earlier. Thus, the principle of a stroboscope is realized, when K samples from different ranges are made for one emitted pulse. In n cycles, the device captures the entire period Tt with step Ts.

The algorithm of the digital unit when recording the period of the reflected signal

Consider the algorithm of the digital unit (24) in more detail.

It is assumed that the transmitter (9) emits pulses with a period Tt, and in the receiver (29) the ADC (23) captures the reflected signal with a frequency Tr. Moreover, the ratio Tr / Tt can be represented as an irreducible fraction M / N, where M and N are natural numbers. It is required, receiving samples from the ADC for the time Tr · N, to obtain a record of the period Tt of the reflected signal containing N samples. This will be equivalent to analog-to-digital conversion in Tt / N time steps.

The algorithm consists of two parts: the data collection part and the data processing part.

The purpose of the data collection part is to obtain a digital recording of a signal of duration Tt with a sampling step of Tt / N, and in terms of programming to produce for subsequent processing an array of numbers R [0 ... N-1] containing a digital recording of the period of the reflected signal. In this case, it is allowed that the recording starts from an arbitrary place in the period. Since the recording time is equal to the period of the signal, the period is recorded in its entirety, regardless of the moment of start.

The algorithm of the first part is as follows. At some point in time, signal recording is initiated. The first value from the ADC (23) after this moment is written to cell 0. Starting from the next, the cell number is calculated from the previous number by the formula: the cell number for recording the value is the remainder of dividing by N the sum of M and the cell number into which The previous value was recorded. This procedure for calculating the next cell number from the previous one is repeated N-1 times. Or, in other words, the count number n after the start of recording is placed in the cell n · M% N, where% is the sign of the operation for obtaining the remainder of the division.

As a result, the array is completely filled with N ADC samples (23).

3a is a table illustrating a sequence of capturing a period of a signal at M = 5 and N = 11. The top line shows the cell number of the array R, the bottom shows the step number of the algorithm at which this cell is filled. This table is provided solely for the purpose of illustrating the operation of the algorithm. Typical embodiments use hundreds and thousands of gates in range, but a table with so many columns cannot be illustrated. Therefore, to illustrate, an example is given with 11 gates in range, which is almost extremely unlikely.

As a specialist can see, as a result of this part of the algorithm, the array will record exactly the signal period equivalent to the sampling frequency Ft · N. Or formally let element zero be captured at time t0, then R [i] = U (t0 + i · Ts), where U (t) is the periodic reflected signal.

The second part of the algorithm is data processing, which ensures that the conditional signal start falls at the beginning of the array - the element R [0].

In a straightforward approach, the block diagram shown in Fig. 4 is used, in which each receiver (29) contains a phase detector (26), to which a digital signal is supplied from the transmitter master generator (1) (the one that is currently emitting) and the master oscillator of this receiver (21). If the active edges of both signals arrive at the same time, the phase detector (26) generates a logic signal (206), according to which the described algorithm of the digital block (24) starts working, and the value captured by the I / O (22) at this edge gets into cell R [0 ]. Since a probe pulse is radiated into the ether along the active front of the master oscillator (1) of the transmitter, a signal corresponding to the very beginning of the pulse radiation will be transmitted to the ether in R [0]. This is exactly what is required.

However, the straightforward approach is extremely difficult to implement, given the fact that the time error should be significantly less than the pulse duration, and the pulse duration in order of magnitude is a hundred picoseconds. Thus, the phase detector (26) should have both a random and a systematic error of less than 10 ps. The situation is complicated by the fact that the propagation time of the reference signal of the transmitter (9) to the phase detector (26) is guaranteed to be many times greater than this value, the delay on any active element with the current state of the art (for example, matching buffer that outputs a signal to the cable) is also several times the required 10 ps. Thus, the calibration of the system becomes inevitable when all these delays are taken into account and compensated in one way or another. The delay data will vary from one device instance to another, so a calibration operation will be required for each device instance. If the device uses more than one transmitter (9), then the task arises of constructing a multiplexer that passes the input signal from the transmitter (9) that is currently emitting pulses to the input of the phase detector (26), this in turn introduces additional errors by time. Finally, the problem arises of compensating for the thermal drift of active elements.

In a preferred embodiment, to avoid the described problems, this phase detector (26) is not used at all. The receiver (29) starts the execution of the algorithm at any time when necessary, and the period of the received signal is fixed in the array R. Next, the fact is used that, in addition to the reflected signal, a signal transmitted directly from one of the transmitters (9) passes to the receiver (29). The shape of this signal is known, the direct signal is usually much more powerful than the reflected one, therefore it can be identified in the R array.

After the accumulation data collection part, there is a part of the algorithm that processes the signal. This part receives an input array R, obtained as a result of the work of the accumulation part. A direct signal is detected in the array R by digital signal processing methods and determines the cell number sr, into which some conditional point selected at the beginning of the signal fell. In a preferred embodiment, such a conditional start point of the signal is considered to be a transition through zero between two maximums (positive and negative) of the pulse that has passed directly.

After the number sr becomes known, a cyclic shift of the array R by sr is performed, that is, each cell i is rewritten to the cell with the number i-sr. Here and in operations with the cell number of the array R, arithmetic is used modulo N, that is, if a negative number is obtained by subtraction, then N. is added to the result. Thus, the cell capturing the conditional start of the signal falls into cell R [0], the signal of which was in the array until the cell sr hits the end of the array. This corresponds to the tasks of the algorithm, because the signal before the start of the emission of the pulse is the end of the reflected signal from the previous pulse. The result is the desired array R containing a record of the period of the received signal, starting at R [0] with the start of the pulse emission by the transmitter.

If we consider the algorithm more strictly, as a result of the above algorithm, a certain point is located in the cell R [0], which is no more than Ts / 2 from the conditional moment of the signal start, we denote this deviation as dTa. From the point of view of the processing algorithm, it looks as if all points of the response signal are recorded at times with the same deviation dTa from the nominal. The worst case is, for example, when all points are captured at Ts / 2 earlier than the nominal values, which may degrade the quality of the algorithm for reconstructing a three-dimensional picture.

This situation is illustrated in figure 1. Above the time axis, the required moments of signal capture are marked with vertical lines (3). Bottom of the time axis, vertical lines (4) show the moments when the signal was actually captured. Figure 1 shows the worst case when the deviation of the capture moments dTa is maximum and is half from Ts. In a preferred embodiment, the principle of the proportional-integral controller is used to bring the deviation dTa to zero with the required accuracy. For this, a voltage-controlled crystal oscillator with a small adjustment range (50 ppm in the preferred embodiment) is used as the receiver clock (21). The controller, implemented in the form of an algorithm executed by block (24), provides a control action (voltage) to this generator (21) in order to bring dTa to zero. To determine the discrepancy (in terms of automatic control systems), a cell R [j] is selected such that it receives a pulse that has passed directly, and at this point the derivative of the signal is not equal to zero. As a residual, the difference between the value in R [j] and the value that should be captured with perfect synchronization of the clocks (1) and (21) is used. In a preferred embodiment, the cell R [0] is selected, into which the transition through zero between the two maximums (positive and negative) of the direct signal should fall, and zero is selected as the value required for perfect synchronization. That is, the task of the control loop is to maintain zero in the cell R [0], or, what is the same, to keep the moment of capture of the cell zero coinciding with the moment of transition through zero of the signal between the two signal maximums.

In the absence of this control loop, dTa will not only not be zero, but will linearly increase with time, and the rate of change will be proportional to the deviation of the generator frequencies from the nominal. The disclosed method based on a proportional-integral controller not only keeps dTa near zero, but also ensures that the frequency ratio Tr / Tt is maintained as M / N. The deviation of the transmitter generator frequency from the required Ft · N / M will manifest itself in the slow drift dTa with time. This drift is detected by the controller and is eliminated by correcting the frequency of the receiver clock (21).

The use of a proportional-integral controller to form a corrective action on the receiver clock (21) to synchronize it with the clock (1) of this transmitter is not mandatory in the disclosed invention. A specialist who knows the current level of technology in the field of automatic control systems can also choose another way to form a corrective effect.

Benefits of Air Sync

For synchronization, the same channel is used as for the useful signal, so most errors are compensated. For example, the generator of the radio frequency signal (2) in FIG. 2 gives a signal with a certain delay relative to the active edge of the signal of the master digital generator (1), this delay is individual for each generator and can change over time, for example, when the device heats up. In this circuit, this delay is compensated, and its slow changes do not impair the operation of the circuit.

No special synchronization path is required, the device is simplified and cheaper. The path of the transmitter (9) and receiver (29) is designed to work with signals of the band of the order of 10 GHz; such a path has high characteristics, including fluctuations in the propagation delay of the signal, however, this is provided by expensive components and expensive design. This path is the most expensive component of the device. To create a synchronization circuit with the same characteristics, it will take the same effort, which can almost double the price of the device.

Since the same path is used for synchronization and for receiving a useful signal, the random error is the same. Random error, unlike systematic, is not compensated. However, in this case, the improvement of a single path leads to an improvement in both synchronization and reception of the signal. In the case of a straightforward implementation, there would be two paths, and the accuracy of the system would be determined by the characteristics of the worst of them.

The system contains a built-in automatic frequency control system that holds the required frequency ratio Fr / Ft = M / N; no additional circuits are required.

As a result, it turns out that no wires are required between the transmitter (9) and the receiver (29), which greatly simplifies and cheapens the design. If one transmitter (9) is used, it is only required to emit a signal periodically from it all the time the scan is performed. The receivers (29) must conduct a measurement cycle for each position and transmit data to the data processing device. In this case, the receivers (29) must synchronize the data collection cycles with the movement of mechanics, and since the movement is relatively slow, ordinary digital signals can be used for this synchronization. As a result, there is no need to conduct special synchronization wires between the blocks, but only a wire for data exchange is required, while standard digital interfaces such as RS-485, USB, Ethernet can be used. It is also possible to use wireless data transfer interfaces, such as WiFi, which allows for multi-position radar. Such an opportunity is hardly demanded in the main field of application of this invention - inspection cabins, however, it can be extremely valuable in many other applications of radio vision and UWB radar. It is very important that for the connection of remote devices, you can use standard both wired and wireless interfaces, rather than a special UWB connection, as proposed in a series of patents.

Coherent accumulation

In the simplest case, for a given position of the transmitter (9) and receivers (29), only one cycle of the described algorithm for the operation of the digital unit is required. In this case, it is required to provide sufficient transmitter power (9) to ensure a given signal-to-noise ratio of the reflected signal. However, in many cases it is advisable to conduct several cycles of recording the period of the reflected signal and add the corresponding values of the array R, this operation is a coherent accumulation of the reflected signal. This operation improves the signal-to-noise ratio, which reduces the power of the emitted generator.

There are several reasons why this may be desirable.

Sanitary standards were calculated for continuous radiation. Brief powerful pulses of electromagnetic radiation in this frequency range are practically not found in nature, their effect has been little studied. And although there is no data on the dangers of short pulses under conditions of their compliance with sanitary standards, nevertheless, apparently, it is desirable to avoid them, if possible.

From the point of view of electromagnetic compatibility, a lot of weak pulses are usually preferable.

Powerful pulses can have a negative effect on the electronics of the device.

A powerful transmitter signal (9) can overload the receiver (29) and bring it out of the normal mode.

Moreover, from a theoretical point of view, both approaches are equivalent, the same energy of one powerful pulse and a series of weak ones will provide the same signal / noise with both approaches.

Coherent signal accumulation is carried out by the required number of repetitions of the execution of the digital block algorithm described above. The corresponding cells of the array R are summed, and the array of sums is given out for processing by the algorithm for reconstructing a three-dimensional picture.

At the same time, the control loop that ensures the synchronization of the generators is maintained in each recording cycle of the reflected signal, that is, it uses the values before the averaging operation, which allows the generators to be corrected, as before, once per Tr This is possible if, after reducing the power of the transmitter (9), a direct signal can nevertheless be identified at the receiver (29). Since the direct signal is usually several tens of times stronger than the reflected signal, this condition is usually satisfied.

Signal thinning

Consider the accuracy requirements for generators (1) and (21), which are necessary for the algorithm to function correctly. According to this algorithm, with ideal elements of the receiver circuit (29) (primarily with ideal generators (21)) R [i] = U (t ₀ + i · Ts). In a real circuit, during the operation of the Tr · N algorithm, the phase of both generators deviates from the nominal value. That is, the last pulse is emitted by the transmitter (9) with a deviation dTt from the nominal moment of time, and the last sample is captured by the UVX (22) with the deviation dTr from the nominal. Since it is required to ensure the exact time of signal capture relative to the moment of emission of the pulse, the error is determined by the difference dTr minus dTt. The difference dTt and dTr is a random variable, it is required that its distribution function satisfy the requirements of particular embodiments. In a specific case, a specialist in information processing algorithms determines the requirements for statistics of this value, and a specialist radio engineer is able to calculate and measure this value for given types of generators. Typically, the phase deviation of the generator is considered to have a Gaussian distribution and is described by two parameters - bias and dispersion. And for the successful operation of the algorithm for determining the three-dimensional shape of an object, it is usually required that both parameters be several times smaller than the required time step Ts.

The shift of the random variable dTr is determined by the deviation of the generator frequency from the nominal one - in the case of a free generator, such a deviation is inevitable. Since in the preferred embodiment, a frequency autotune system Fr to Ft functions, these frequencies are kept in the correct ratio and the offset of the difference dTr-dTt is zero.

The values of dTr and dTt represent the deviation of the actual duration of N periods from the ideal N · Tr in the case of the receiver and M · Tt in the case of the transmitter. The variance of this random variable is called accumulated jitter. It is known that for a generator that is not corrected by a feedback circuit, it increases as a root of n, where n is the number of periods between the studied fronts. Since the generator (21) of the receiver (29) is locked by a frequency-locked loop to the transmitter (9), during continuous operation, the dispersion dTr is determined by the properties of the generator (21) and the frequency-locked loop. In the above algorithm, the corrective action is applied once a cycle of the algorithm, that is, during the interval of one cycle of operation, the correction algorithm is not performed, and the accumulated jitter grows according to the law the root of n, like a generator without a feedback circuit. This imposes a restriction on decreasing the step Ts. Reducing the step Ts, on the one hand, strengthens the requirements for the accumulated jitter. On the other hand, a decrease in the step Ts leads to an increase in the duration of the recording cycle of the reflected signal Tt / Ts · Tr and thus increase the accumulated jitter. As a result, a decrease in the step Ts seriously increases the requirements for the characteristics of the generators (1), (21). In order to ensure that the system operates at lower step values Ts, the following improvement is implemented in a preferred embodiment.

For the operation of the phase-locked loop system, it is desirable that the data on the phase deviation come as often as possible, this allows you to keep the phase deviations at a lower level. To ensure this, the fact is used that any point of the direct signal with a good signal-to-noise ratio, with a derivative that is not equal to zero, is suitable for the operation of the disclosed method of synchronization retention. In this case, M and N are selected so that the correction operations are distributed evenly over the signal recording cycle, and the interval between corrections is minimal.

In particular, the values of M and N can be selected so that N values of the array R are collected in L passes, i.e., a signal thinned out (decimation) by L times is collected L times, and new points are collected with each pass. The value of L is selected so that for each pass at least one point is captured at which a correction can be made. That is, the value of L is close to the length of the forward signal; the fragment length is measured in the number of steps Ts that fit into it.

Figure 5 shows an example of a typical signal emitted by the transmitter (9), to determine the phase deviation of the generators (1) and (21), you can use part of the signal between the vertical lines 4. Thus, at least one point with a large amplitude is captured for each pass and derivative. To select M and N that meet this requirement, you can use the method described above in the section “Frequency selection in a simplified embodiment”, but at the same time choose the values so that the condition Tt = k · Tr-L · Ts is satisfied and so that k · Tr / L · Ts was not an integer.

3b is a table illustrating a sequence of capturing a period of a signal at N = 10 and M = 17, L = 3. The top line shows the cell number of the array R, the bottom shows the step number of the algorithm at which this cell is filled.

In this example, it is seen that five consecutive signal captures cover the period with an interval between the captured points of 2-3 Ts. Thus, if the region containing the signal direct passage fragment along which the phase deviation can be measured is 4Ts, then the generator (21) will be adjusted every 5 cycles.

This table is provided solely to illustrate the operation of the algorithm. Typical embodiments use hundreds and thousands of gates in range, but a table with so many columns cannot be illustrated. Therefore, for illustration, an example with 17 gates in range is given, which is almost extremely unlikely.

The specialist can choose N and M in another way.

Other embodiments of the invention

In the best embodiment, only the forward signal is used to synchronize the transmitter (9) and receiver (29). However, to further reduce the dispersion of dTt-dTr, it is possible to use the synchronization of the transmitter and receiver generators (9) and (29) according to a common reference generator of the system. A special synchronization circuit in the receiver (9) receives an input signal from this reference generator of the system and controls the generator (21) of the receiver in order to minimize the accumulated jitter. A similar circuit is used in the transmitter (9). As is known from radio engineering, such a “linking” of two generators to one common one can significantly improve the variance of the phase deviation of one of the two slave generators relative to the phase of the other.

In this embodiment, maintaining the ratio of the frequencies of the transmitters and receivers (1) and (21) of the transmitter and receiver, as well as reducing the dispersion dTt-dTr, is carried out by the wired system. And reducing the error dTa to zero and searching for the signal in the array R is carried out using the direct signal, as described in the best embodiment.

To simplify the circuit, a frequency equal to the largest common divider of the clock frequencies of the transmitter and receiver is used as the frequency of the reference generator.

As a reference generator, a generator of any element of the transmitter or receiver can be used. This can be especially effective in an embodiment that includes only one transmitter (receiver), in which case it is advisable to make it a reference generator for other receivers (transmitters).

Alternative embodiment

In an alternative embodiment, the error dTa is determined by the previously disclosed method by identifying in the received signal a pulse transmitted directly from the transmitter (9). In contrast to the preferred embodiment, this error is eliminated by interpolation before mathematical processing of the resulting array R. Formulating the problem mathematically, knowing the signal values at times t (n) = Ts · n + dTa, calculate the signal values at times t (n) = Ts · n Calculation methods are well known in mathematics and are applicable if the period Ts is substantially less than the period of the maximum signal frequency. The last condition is necessary not only for interpolation, but also for subsequent mathematical processing, therefore it is ensured by an appropriate selection of the parameters Tr, Tr, N, M.

That is, in this embodiment, the deviation of the synchronization of the generators is not eliminated, but is compensated by an additional step of computer processing after the data are accumulated in memory and before they are processed to obtain an image.

Alternative embodiment

In another alternative embodiment, the synchronization of the generators (1) and (21) of the transmitter and receiver is carried out by synchronizing them with a common main generator, the signal from which comes through the wires. Wire synchronization is used to minimize the relative phase noise of the generators (1) and (21) of the transmitter and receiver and to strictly maintain the required frequency ratio. The detection of a forward-pass pulse is used to ensure that the conditional start of the signal enters the cell R [0] of the array transmitted for mathematical processing. For this, the cyclic shift of the array R described in the description of the preferred embodiment is used. The error dTa mentioned in the description of the preferred embodiment is ignored, for which the equivalent TVN sampling step is made small enough so that the time error Tt / N / 2 can be neglected.

Claims (12)

1. The method of recording the reflected signal during radar, which consists in the fact that
- at one or more points in space, ultra-wideband pulses are periodically emitted using at least one transmitter by a signal from at least one clock connected to the corresponding transmitter,
- at more than one point in space using receivers capture electrical signals, record and store them in memory,
- process the captured signals using the processing unit,
characterized in that, using at least one receiver, the instantaneous values of the signals are captured at the moments specified by another corresponding clock generator connected to it with a capture period different from the period of emission of pulses. 2. The method according to claim 1, characterized in that the pick-up period Tr of the signals and the period Tt of the emission of pulses are selected so that Tr / Tt is equal to the irreducible fraction M / N, where M and N are natural numbers, and a signal is recorded of duration Tt with a sampling step Tt / N by rearranging in the memory of any N consecutive captured signal values in memory, at which the signal value with serial number n in the original sequence of captured signal values stored in memory is put in place with the number n · M% N in the resulting sequence atelnosti these values, where% - sign operation receiving remainder. 3. The method according to claim 1, characterized in that they use clock generators of the receivers, each of which is located in close proximity to the component of the corresponding receiver, which captures the reflected signal. 4. The method according to claim 1, characterized in that synchronize the operation of at least part of the clock transmitters with the work of at least part of the clock of the receivers by synchronizing them with a common main generator. 5. The method according to claim 4, characterized in that they use clock generators of transmitters and receivers, the frequency of each of which is a multiple of the frequency of the main generator. 6. The method according to claim 1, characterized in that the captured signals identify the pulse transmitted directly from one of the transmitters, analyze at least one of the values related to this pulse, and form a corrective effect on the receiver clock for synchronization with the clock of this transmitter. 7. The method according to claim 6, characterized in that the formation of the corrective action is carried out using a proportional-integral controller, the input signal of which is a value related to the pulse passed directly, and the residual is the difference between the captured value and the value, which should be Captured with perfect clock synchronization. 8. The method according to claim 6, characterized in that the periods Tt and Tr are selected so that the maximum time interval without the formation of a corrective action is Tr · N / 2 or less. 9. The method according to claim 1, characterized in that a pulse transmitted directly from one of the transmitters is identified in the captured signals, the synchronization deviation parameters of the clock generators of the respective transmitter and receiver are calculated from this pulse, and these parameters are used to compensate for the synchronization deviation. 10. The method according to claim 2, characterized in that to improve the signal-to-noise ratio of the recorded and stored signals, the captured values are averaged, the time interval between the moments of capture of which is equal to or a multiple of Tr · N 11. The method according to claim 6, characterized in that to improve the signal-to-noise ratio of the recorded and stored signals, the values of the captured signals are averaged, the time interval between the moments of capture of which is equal to or a multiple of Tr · N, and one or more values of pulses transmitted directly from one of the transmitters and used to correct the receiver clock is not averaged. 12. The method according to claim 6, characterized in that they use only one transmitter for emitting ultra-wideband pulses, and synchronize the clock generators of all receivers with the clock generator of this transmitter.

Images