FMCW radar sensor with synchronized high-frequency module

Abstract

A method for encoding and storing digital data including a plurality of real parameters in a signal processing unit of a radar sensor, in which at least one real parameter r is stored in an exponential representation of the following form: r = m·b<supgt;‑k< / supgt>, where m is a digital mantissa of length p, b is an integer base, and k is a positive number encoded as a digital number of length q, characterized in that an exponential representation with b > 2 is used for the compressed storage of the parameter r.

Description

Technical Field

[0001] The invention relates to a method for encoding and storing digital data comprising a plurality of real variables in a signal processing unit of a radar sensor, in which method at least one real variable r is stored in an exponential representation of the following form:

[0002] r=m·b- k ,

[0003] Where m is the mantissa of length p, b is the integer base, and k is a positive number encoded as a digital number (digitale Zahl) of length q.

[0004] In particular, the present invention relates to the encoding and storage of digital data in a radar sensor for a motor vehicle. Background Art

[0005] In radar sensors for motor vehicles, the radar signals transmitted are mostly periodically frequency modulated signals, for example a sequence of frequency ramps transmitted in succession or also a sequence of so-called OFDM symbols (OFDM=Orthogonal Frequency Division Multiplex). The signal received from the located object is mixed down to a low-frequency baseband and digitized by means of an analog / digital converter after preamplification. Thus, a time signal in the form of a vector x(n) is obtained per modulation period, the components of which are real or complex numbers depending on the demodulation method used and which are represented digitally. The number of components, i.e. the dimension of the vector, corresponds to the number of measurement time points at which the signal is evaluated within the modulation period.

[0006] The vector x(n) can be converted by at least one Fourier transformation (e.g. FFT) into a vector V(k), the complex components of which describe the amplitude and phase of the received signal as a function of the frequency k. The dimension of this vector corresponds to the number of frequency bins on the frequency axis and thus determines the frequency resolution.

[0007] In many known radar systems for motor vehicles, each modulation period includes a plurality of successive frequency ramps or OFDM symbols, which can be counted using the index y. In this case, the vector V y (k) A further fast Fourier transform on index y is performed to form a two-dimensional matrix. Then, each cell of the matrix represents a combination of distance d and relative velocity v, and the cell in which the parameter obtained by the two-dimensional FFT takes the maximum value represents an object located at distance d and having a relative velocity v.

[0008] If the radar sensor has an array with multiple receiving antennas and evaluates the signals of the receiving antennas in separate receiving channels, the azimuth and / or elevation angle at which the object is located can also be determined based on the amplitude and phase relationships of the signals obtained for the same object in different channels.

[0009] As the requirements for the range resolution and angle resolution of radar sensors increase, the number of receiving channels increases, and the dimensions of the vectors x(n) and V(k) to be processed in each receiving channel also increase, so that a considerable amount of data must be processed in each modulation cycle. This requires not only a fast processor, but also a high storage capacity for temporary storage of digital input parameters and intermediate results obtained in each processing step. The increase in the demand for storage capacity leads to an increase in chip size and higher costs as well as higher energy requirements.

[0010] According to ANSI / IEEE Standard 754-1985, the digital representation of real numbers is represented by r = m·2- k The storage requirement for each real number is then given by the sum of the length p of the mantissa m and the length q of the exponent k. To represent a complex number, two real numbers are required, ie, the real part and the imaginary part.

[0011] US Pat. No. 9,541,637 B2 and WO 2015 / 185058 A1 describe methods by which the storage requirements are intended to be kept within limits by means of data compression. Summary of the invention

[0012] The object of the present invention is to specify a method by means of which the memory requirement can be further reduced or the positioning accuracy can be improved given a given memory space.

[0013] This object is achieved according to the invention by means of a method of the type mentioned at the outset, in which an exponential representation with b>2 is used for the compressed storage of variables.

[0014] By using a larger integer base instead of the standard base b = 2, a higher resolution can be achieved in a significant portion of the quantity to be stored, given the total length of the mantissa and the exponent. Conversely, this means that, given a given resolution requirement, the length of the mantissa and / or the length of the exponent can be reduced and storage space can therefore be saved.

[0015] Advantageous embodiments and refinements of the invention are described in preferred embodiments.

[0016] In an advantageous embodiment, the base b used is a power of 2, for example b=4 or b=8. The conversion of the exponential representation from one base (for example b=2) to another base (for example b=4) can then be performed very simply and requires almost no additional computing time. Alternatively, the variable can also be generated in the digitization of the analog / digital converter in an exponential representation with b<2.

[0017] According to an advantageous development of the invention, the variable is converted into an exponential representation of the following form:

[0018] r=m * b -f(k) ,

[0019] Among them, m * is the mantissa, f is a function of k selected from multiple functions, and the selection of function f is performed based on the value distribution of the parameter to be stored.

[0020] Depending on the value distribution of the parameter to be stored, it may happen that: b is not needed at all in the exponential representation -1 The function f can then be chosen so that higher powers appear in the exponential representation instead of these unwanted powers, which higher powers enable a higher resolution to be achieved. A digital number k of length q is then used to encode another more suitable selected power instead of b. -1 of 2 q to achieve higher resolution without additional storage requirements.

[0021] The selection of the function f can be performed before the radar sensor is put into operation. The selection criterion is then the expected value distribution of the variable to be stored, which is most likely to be expected taking into account the characteristics, conditions of use and tasks of the radar sensor.

[0022] However, the selection of function f may also be changed dynamically during operation of the radar sensor, for example in the event of a change in the required characteristics of the radar sensor or alternatively also based on a random or continuous statistical analysis of the value distribution of the data to be stored. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In the following, embodiments are further explained based on the accompanying drawings. The accompanying drawings show:

[0024] Figure 1 A simplified circuit diagram of a radar sensor is shown, to which the present invention can be applied;

[0025] Figure 2 Show according to Figure 1 A block diagram of a digital signal processing unit in a radar sensor;

[0026] Figure 3 An example of encoding a digital parameter in a standard format is shown;

[0027] Figures 4 to 6 An example of encoding a digital parameter according to a first embodiment of the present invention is shown;

[0028] Figure 7 The memory architecture of the encoding method according to another embodiment of the present invention is shown;

[0029] Figures 8 to 10 An example of encoding data according to the method of the second embodiment of the present invention is shown;

[0030] Fig.11 Shown to illustrate the Figures 8 to 10 A schematic diagram of a variant of the method;

[0031] Fig.12 and 13 A histogram is shown for illustrating another variant of the method according to the second exemplary embodiment;

[0032] Fig.14 A block diagram of signal processing stages for a method according to a second embodiment is shown. DETAILED DESCRIPTION

[0033] Figure 1 An FMCW radar sensor is shown as an example, which has a transmitting and receiving device 10 with four antenna elements 12, 14, 16, 18, which together form a planar group antenna. The radar sensor is installed in a motor vehicle in such a way that antenna elements 12 to 18 are located next to each other at the same height, so that a certain angular resolution capability of the radar sensor in the horizontal direction (in azimuth) is achieved.

[0034] The high-frequency part 20 for controlling the antenna elements is formed, for example, by one or more MMICs (Monolithic Microwave Integrated Circuits) and has an oscillator 22, which feeds the transmission signal into each antenna element. The frequency of the transmission signal is modulated periodically in the form of a sequence of rising and / or falling frequency ramps. For example, each modulation period includes a sequence of so-called fast linear frequency modulations, i.e. frequency ramps with the same slope, which have a certain frequency offset from each other. The radar echoes received by the antenna elements 12 to 18 are respectively coupled out by means of a circulator (Zirkulator) 24 and supplied to a mixer 26, in which the radar echoes are mixed with the transmission signal provided by the oscillator 22. In this way, baseband signals b1, b2, b3, b4 are obtained for each of the antenna elements, and the baseband signals are supplied to an electronic control and analysis device 28.

[0035] The control and analysis device 28 includes a four-channel analog / digital converter 30, which digitizes and records the baseband signals b1 to b4 obtained by the four antenna elements. The digital time signals thus obtained are then further processed channel by channel in a signal processing unit 32. For example, the time signal of each ramp is converted into a spectrum by means of a fast Fourier transform, which is then Fourier transformed again on the ramp index. In this way, a two-dimensional spectrum is obtained, from which the distance d and the relative speed v of the located object can be read.

[0036] The parameters obtained by Fourier transformation are complex numbers that describe the amplitude and phase of the received signal. Since the amplitude-phase relationship of the signals received from the same object in different receiving channels depends on the azimuth of the relevant object, the azimuth θ of the object can also be determined with a certain accuracy in the angle estimation module 34.

[0037] exist Figure 2 , the basic components of the signal processing unit 32 are shown as a block diagram. The signal processing unit has an input stage 36, which receives the digital data from the analog / digital converter 30. In a compression stage 38, the data are compressed according to a method that will be explained further below. The data are then stored in a compressed form in a memory 40. If the data is to be processed further, it is decompressed in a decompression stage 42 and then further processed in a processing stage 44.

[0038] The signal processing unit 32 can also have a plurality of linked processing stages 44, for example two FFT stages for a two-dimensional Fourier transform. The processing results of the first stage are then recompressed and stored in a further memory, which is then accessed by the downstream processing stage via a further decompression stage. In certain application cases, the processing stage 44 or a downstream processing stage can also be designed so that it can directly process the compressed data, as it is stored in the memory 40. The decompression stage 42 is then bypassed or the data is passed on unchanged. It is also possible that the processing stage changes the compressed data directly in the memory 40. The downstream processing stage then accesses the same memory 40.

[0039] According to the IEEE 754-1985 standard, r = m·b- k The actual parameter is encoded and stored in an exponential representation of the form. Here, b=2, m is a mantissa of length p, for example, 8 bits, and k is a positive integer of length q, for example, 3 bits. Figure 3 In FIG. 1 , the digitization of the decimal number 0.8408203125 is shown as an example. In the binary fixed-point representation, this number is represented by the bit sequence 01101011101, which has a length of 11 bits. Figure 3 In this bit sequence, the value of the relevant binary position is specified as a negative power of 2. The most significant bit has the value 2 0 , the next bit has the value 2 -1 ,etc.

[0040] Figure 3 Table 46 in shows the conversion of this fixed-point number into a standard exponential representation with p=8, q=3 and base b=2. Since the exponent k has a length q=3, k can take values ​​from 0 to 7. Accordingly, Table 46 has eight rows. Since the mantissa has a length p=8, the three least significant bits of the original bit sequence must be discarded when converting to the exponential representation. The eight-bit mantissa is entered into the following row of Table 46: this row (in the case of a positive number) is located one position to the left of the first bit different from 0 in the original 11-bit sequence, that is, in the given example in row k=0. Therefore, the stored parameter can be represented as

[0041] r=m 0 2 -0 +m 1 2 -1 +…+m 7 2 -7 ,

[0042] Among them, m i (i=0 to 7) is the i-th digit of the mantissa m.

[0043] In this example, the mantissa is specified in two's complement format. In this format, in the case of a positive number, the most significant bit (at the left end of the sequence) must be equal to 0, and in the case of a negative number, the most significant bit must be equal to 1. If the original bit sequence has more than one leading zero, the leading zeros (except the last one) can be deleted and the exponent k is increased by 1 for each deleted zero.

[0044] As an example, Figure 3 Table 48 in shows the encoding of the decimal number 0.02813720703125. The relevant bit sequence has six leading zeros. Five zeros are deleted and the remaining mantissa (next 8 bits) is entered into row k=5.

[0045] Accordingly, in the case of negative numbers, the leading ones (except the last one) can be dropped. Since additional low-order bits in the mantissa can be recorded for the dropped leading bits, a higher precision is achieved in the representation of real numbers in the exponential representation.

[0046] For the case shown in Table 46, in the case of the original decimal number 0.840..., the value of the decimal number after conversion to exponential representation is 0.8359375. Comparison of the two decimal numbers shows that a quantization error of the order of 0.004 is caused by limiting the mantissa to 8 bits.

[0047] The decimal number corresponding to the original bit sequence and the exponent representation is also indicated for the case shown in Table 48. It can be seen that here the quantization error is significantly reduced due to the scaling with k=5 bits.

[0048] exist Figure 4 In Table 50, Figure 3 The example of the same decimal number in (Table 46) shows the encoding in the exponential representation of the method according to the invention. The particularity of the method is that it does not process with the usual base b=2, but with base b=4. Therefore, an increase of 1 in the exponent k corresponds to a multiplication by a factor of 2 -2 . Correspondingly, in table 50, the rows are shifted by two positions relative to each other. Figure 3 The same dynamic range (2 0 to 2 -14 In this case, an exponent k in the range of 0 to 3 is sufficient, which can be encoded by a 2-bit word. The length q can thus be reduced from 3 to 2. As a result, either storage space is saved or the freed bits are used to increase the length p of the mantissa from 8 to 9, thereby achieving higher precision and reducing the quantization error.

[0049] In table 50, the mantissa has a length of 9 bits, so that only the last two bits need to be removed from the original 11-bit word. The associated decimal number is also in Figure 4 It is illustrated in , and shows that the quantization error has been reduced to about 0.001.

[0050] However, the method applied in this example does not lead to a reduction in the quantization error for every real number. Figure 5 In Figure 5 In the example, Table 52 first shows the conventional exponential representation of an 11-bit word with base b=2 and with p=8 and q=3, which has two leading zeros in this example. In the conventional method, the leading zeros are deleted and the exponent is increased to k=1 for this purpose. Therefore, only the last two bits need to be removed from the original 11-bit word.

[0051] Figure 5 Table 54 in shows the method according to the invention for the same 11-bit word with b=4, p=9 and q=2. If k is to be increased from 0 to 1, the two leading zeros will have to be deleted. However, then the most significant bit in the remaining mantissa will be 1, which in the two's complement format will correspond to a negative number. Therefore, in Table 54, the encoding must be done with k=0 and, although the mantissa is extended to 9 bits, the same data loss as in Table 52 results.

[0052] If multiple real parameters are to be encoded and their values ​​are approximately evenly distributed, then in about half the cases the number of leading zeros will be an odd number, as in Figure 4 In one half of the data, the number of leading zeros (or ones) will be even, and a reduction in the quantization error will be achieved, while in the other half, the number of leading zeros (or ones) will be even, thus not having any advantage over conventional methods. However, in the statistical mean, a significant improvement in accuracy is achieved in half of the data by the method according to the invention.

[0053] exist Figure 6 In FIG. 5 , Table 56 shows a variant of the method according to the invention with base b=4 and with a conventional length p=8 for the mantissa and a conventional length q=3 for the exponent k. Overall, for the storage of the mantissa and the exponent, the same as in Figures 3 to 5 Although no higher precision can be achieved by extending the mantissa, a much larger dynamic range can be covered with the same 11 bits as in the prior art, i.e. 2 0 to 2 -21 .

[0054] The processing stages 44 in the signal processing unit 32 are each designed for a specific task setting, and the basic features of the data structure of the data to be processed and stored are known in advance. Therefore, it can be determined individually for each individual processing stage 44 which variant of the method according to the invention is to be worked with. If a large dynamic range is required, q will be increased. If the expected dynamic range is smaller, p can be increased and thus achieve a higher accuracy. In some cases, it is also possible to work with an even larger base (e.g. b=8) according to the purpose, wherein the base should preferably be a power of 2.

[0055] Figure 7 The encoding method according to the second embodiment of the present invention is shown. In this embodiment, k does not directly indicate the exponent in the exponential representation, but k is only the independent variable of the function f(k), and the function value of the function then determines the exponent. The exponential representation therefore has the following form:

[0056] r=m * b f(k) ,

[0057] Among them, the last digit is m * mark.

[0058] exist Figure 7 60, 62, 64, which define different functions f. In the value tables 58, 60 and 62, the function values ​​or the powers b formed therefrom are explicitly stated. f(k) For the assignment of independent variable k.

[0059] The advantage of this method is that even in the case of a given length q=2 of the independent variable k to be stored, it is not limited to the four lowest powers 4 0 , 4 -1 , 4 -2 and 4 -3 (as in value table 58), but optionally another set of four powers of 4 may also be used, as shown by way of example in value tables 60 and 62. The choice of function f and the associated value table may then depend on which set of powers best suits the expected or discovered structure of the data to be stored. This is further elaborated below.

[0060] In this example, the data are stored in memory 40 in different data blocks 66, 68, 70 and a pointer 72 is additionally stored for each data block, which points to one of the value tables 58 to 64. The data in each data block are encoded and decoded using the function f indicated by the pointer 72.

[0061] exist Figure 874 to be stored, which in this simplified example consists of only five bit sequences, each of which has a length of 22 bits. However, in this example, the dynamic range is only 14 bits, since all low-order bits are 0. The first three bit sequences correspond to real numbers in the order of magnitude 20, while the last two bit sequences indicate real numbers in the order of magnitude 2. -4 or 2 -6 The smaller number in .

[0062] In addition, Figure 8 Three tables 76, 78, 80 are shown, which respectively represent the Figure 7 The exponential representation of one of the functions f shown in FIG. The respective associated value tables 58, 60 and 62 are in Figure 8 The same is explained in .

[0063] In order to check the extent to which the index representation according to table 76 is suitable for data set 74, for example, a row can now be selected from table 76 for each bit sequence in data set 74, with which the bits different from 0 can be covered as well as possible. The bits covered in this way are marked in data set 74 by a box 82. It can be seen that the first bit sequence can be completely covered with row k=0 and the last bit sequence can be completely covered with row k=3. In the other three bit sequences, several low-order bits are lost in each case.

[0064] If the same procedure is now repeated with tables 78 and 80, it will be seen that the data loss occurring in these tables in total is greater. Therefore, for encoding data set 74, the function f defined by value table 58 (table 76) will be selected.

[0065] Similarly, Fig. 9 shows an example of a data set 84 that may be best mapped in a value table 60 (table 78), and Fig.10 An example of a data set 86 is shown that may be best mapped in the value table 62 (table 80).

[0066] An error metric may be calculated as a criterion for selecting the most appropriate function f, such as the sum of quantization errors, mean square error, etc.

[0067] Fig.11 A more extensive data set 88 is shown with a total of twenty bit sequences. These bit sequences are at least approximately sorted according to the decreasing magnitude of the real number represented by the bit sequence. This corresponds to a situation that is often encountered in practice. For example, a bit sequence can represent the amplitude of a received radar echo with increasing frequency and correspondingly increasing object distance. The sorting according to decreasing magnitude is automatically achieved based on the fact that the radar echo becomes weaker as the object distance increases.

[0068] In order to minimize the storage requirements and / or improve accuracy, it is now practical to divide such an ordered or partially ordered data set into individual blocks 90, 92, 94 and to select for each block a function f which best matches the data structure of the block. For example, a function f is selected for block 90 in which mainly low powers of four occur in the value range, and a function is selected for block 94 in which mainly high powers of four occur in the value range, so that leading zeros or ones occurring in all bit sequences in block 94 can be reduced.

[0069] If the area of ​​use of the radar sensor is known, the selection of the function f for the different blocks and the base (p=4 or higher) to be used and the length p of the mantissa and the length q of the argument k can be determined before the radar sensor is put into operation. However, in another embodiment, these parameters can also be adapted dynamically during operation of the radar sensor based on the data currently to be processed.

[0070] Instead of defining the function f with the aid of predefined and stored value tables 58 , 60 , 62 , in another embodiment the function f to be applied in each case can also be generated directly during data compression by selecting a power of four that best covers the valid bits in the bit sequence to be compressed.

[0071] The selection or generation of the function f can be performed, for example, based on a statistical analysis in which a histogram is created based on the bit sequences to be stored, which histogram specifies, for each power e of the base b, the number n of bit sequences whose most significant bit (after omitting leading ones or zeros) is located at b -e With b -e-1 An example of such a histogram is Fig.12 and 13 In Fig.12 The value of the most significant bit is either in b 0 To b -1 In the range of b -5 To b -7 Since only four different powers can be described with an argument k of length 2, the four lowest powers are selected from all the powers that occur, in this case b 0 , b -1 , b -5 and b -6 . Bit sequence - the valid bits of the bit sequence are in b -7 Started in - m.b -6 A form of encoding, in which some data loss must be tolerated.

[0072] Fig.13 An example is shown in which the most significant bit is either in b-3 To b -5 In the range of b -9 Therefore, in this case, the power b is chosen -3 、b -4 、b -5 and b -9 .

[0073] Fig.14 The block diagram of the signal processing unit 32' is shown, in which the parameters for data compression can be dynamically adapted. Here, a statistical module 96 is inserted between the input stage 36 and the compression stage 38, which performs a statistical analysis on the data received from the input stage 36, for example by creating Fig.12 and 13 . If necessary, the data set to be stored is also divided into blocks with similar data structures in this statistical module 96. The results of the statistical analysis of the complete data set or of the block just considered are then submitted to a selection module 98, which determines the value table for the function f to be applied and, if necessary, the optimal parameters p and q and, if necessary (if the base should be greater than 4), the base b. The parameters (and the function) determined by the selection module 96 are submitted to the compression stage 38 and used there for data compression.

[0074] The compressed data is then stored block by block in the memory 40 together with the parameters used (or a pointer to the set of parameters used).

[0075] In the method described so far, only the encoding of real numbers is considered. However, it is understood that the method can also be used for complex numbers, because each complex number can be represented by two real numbers, for example, by the real part of the complex number and the imaginary part of the complex number or also by the magnitude and phase of the complex number. Then, the above encoding method can be used to encode each or at least one of the two real numbers representing the complex number. For example, the exponential representation can be used for the magnitude and the fixed-point representation can be used for the phase. In many application cases in radar sensors, this representation of complex numbers is particularly advantageous, because phase compensation is often required in data analysis processing, which is simplified to a simple addition of phases when the complex parameters are represented by magnitude and phase. Application examples are, for example, phase compensation in radars with synthetic apertures (SAR=Synthetic Aperture Radar) or phase compensation in radar sensors with OFDM modulation.

Claims

1. A method for encoding and storing digital data comprising a plurality of real variables in a signal processing unit (32, 32') of a radar sensor of a motor vehicle, wherein at least one real variable r is stored in an exponential representation of the form: r=m * ·b -f(k) , in, m * is a digital mantissa having a length p, b is a base, and k is a positive number encoded as a digital number having a length q, f is a function of k selected from a plurality of functions, wherein an exponential representation with b>2 is used for compressed storage of the at least one real parameter r, wherein a complex number is represented by a magnitude of the complex number and a phase of the complex number, and the exponential representation is used for the magnitude, thereby reducing the storage requirements or, given a storage space, improving the positioning accuracy and / or reducing the quantization error based on a statistical analysis of the digital data to be stored, The signal processing unit has an input stage (36) which receives digital data from an analog / digital converter (30), wherein the digital data are compressed in a compression stage (38) and the compressed data are stored in a compressed form in a memory (40) of the radar sensor, and the compressed data are decompressed in a decompression stage (42) and further processed in a processing stage (44), wherein, by using a larger integer base b>2, a higher resolution is achieved in the portion of the actual variable to be stored for the length of the mantissa and the exponent, so that, given a given resolution requirement, the length of the mantissa and / or the length of the exponent can be reduced to save storage space in the radar sensor, The radar sensor comprises an FMCW radar sensor having a transmitting and receiving device (10) with a plurality of antenna elements (12, 14, 16, 18), wherein the plurality of antenna elements together form a planar group antenna. The radar sensor is installed in the motor vehicle so that the antenna elements (12, 14, 16, 18) are located side by side at the same height, thereby achieving a certain angular resolution capability of the radar sensor in the horizontal direction, that is, in azimuth. Wherein, a plurality of functions are stored in advance in the form of a value table (58, 60, 62), and the function f is to be selected from the plurality of functions, The choice of function f and the associated table of values ​​depends upon which set of powers best fits the expected or previously discovered structure of the digital data to be stored.

2. The method of claim 1, wherein b is a power of 2.

3. The method according to claim 1 or 2, in which p and q are determined based on a known or expected data structure of the variables to be stored. 4 . The method according to claim 3 , in which during operation of the radar sensor p and / or q and / or b are dynamically adapted to an expected data structure of the respective variable to be stored.

5. The method according to claim 1, wherein at least one actual parameter r is stored in an exponential representation of the following form: r = m·b -k , in, m is a digital mantissa having a length p, b is a base, and k is a positive number encoded as a digital number having a length q, wherein an exponential representation with b>2 is used for compressed storage of the at least one actual parameter r. 6 . The method according to claim 5 , wherein the selection of the function f is changed during operation of the radar sensor as a function of the data structure of the variables to be stored. 7 . The method according to claim 5 , wherein during operation of the radar sensor, the value of the selected function f is generated as a function of the data structure of the variable to be stored.

8. The method according to claim 5, wherein parameters for data compression and / or functions for data compression are determined based on a statistical analysis of the data to be stored.

9. A radar sensor for a motor vehicle, Features A signal processing unit (32; 32') is designed to carry out the method according to any one of claims 1 to 8.

Images