Frequency spectrum regeneration method and receiver

Abstract

To implement spectrum reproduction processing using software.SOLUTION: A frequency spectrum reproducing method includes: sampling, without using BPF, a receiving frequency spectrum with K kinds of sampling frequencies; performing Fourier transformation independently to obtain frequency spectra from 0 to a Nyquist frequency group; setting a frequency pitch width of K frequency spectra equally; generating a vector (y) by combining frequency spectrum elements; generating a coefficient matrix (A) using frequency elements corresponding to a frequency corresponding to a frequency group given by a resolution corresponding to the frequency pitch width with frequencies from 0 to a system upper limit frequency and a folding frequency in the Nyquist frequencies, and matrix elements corresponding to the elements of (y), as coefficients corresponding to frequency characteristics from a signal branch point before folding in the Nyquist frequencies to an analog-digital converter, and other elements as zero; and calculating a solution (x) of (y)=(A)(x) with an overdetermined algorithm using (y) and (A).SELECTED DRAWING: Figure 9

Description

[Technical Field]

[0001] The technology described herein relates to a frequency spectrum recovery method and receiver. [Background technology]

[0002] In recent years, with the spread of wireless Internet of Things (IoT), real-time spectrum monitoring technology for wireless IoT frequency bands has been developed to avoid interference between different wireless IoT systems. Wireless IoT frequency bands include, for example, the 920 MHz band, the 2.4 GHz band, and the 5-6 GHz band.

[0003] FIG. 1 is a block diagram showing a schematic configuration of a receiver 600 as a conventional example.

[0004] The receiver 600 includes an antenna 6, a plurality of (three in the illustrated example) band pass filters (BPFs) 7, a simultaneous undersampling processor 8, and a radio frequency (RF) spectrum restoration processor 9. The simultaneous undersampling processor 8 includes a plurality of (three in the illustrated example) analog-digital converters (ADCs) 81 and a plurality of (three in the illustrated example) clocks (CLKs) 82.

[0005] In the receiver 600, signals received by the antenna 6 are passed through a plurality of BPFs 7, each of which passes a signal in a predetermined frequency band. The signals that have passed through the BPFs 7 are converted from analog to digital signals by a plurality of ADCs 81 in a simultaneous undersampling processor 8. Each ADC 81 is controlled by a CLK 82 (in other words, CLK #1 to #3). The digital signals output from the simultaneous undersampling processor 8 are then passed through an RF spectrum restoration processor 9, where the original spectrum is reproduced.

[0006] FIG. 2 is a diagram for explaining spectrum regeneration processing by the receiver 600 shown in FIG.

[0007] The signal output from each BPF 7, indicated by reference symbol A1, is reconstructed to the original spectrum, indicated by reference symbol A2, by the compressed sensing algorithm.

[0008] FIG. 3 is a diagram illustrating the details of the spectrum reproduction process shown in FIG.

[0009] In the example shown in Fig. 3, an observation spectrum including three RF complex spectra is input as indicated by reference symbol B1. As indicated by reference symbol B2, the RF complex spectra are convolved (in other words, folded back). Then, as indicated by reference symbol B3, linear equation generation and compressed sensing are performed to reconstruct the original spectrum. [Prior art documents] [Patent documents]

[0010] [Patent Document 1] Patent Publication No. 2021-106340 Summary of the Invention [Problem to be solved by the invention]

[0011] However, the receiver 600 shown in FIG. 1 includes a BPF7, and current technology does not provide a tunable filter with low loss (high Q) and low distortion. As a result, it is not possible to freely set the bandwidth with a single device, and the receiver cannot be made into a Software Defined Radio (SDR).

[0012] In one aspect, the technology described herein aims to realize spectrum reclamation processing using software. [Means for solving the problem]

[0013] In one aspect, a frequency spectrum recovery method is a frequency spectrum recovery method using an undersampling reception scheme, in which a received frequency spectrum is generated from time axis data sampled at K different sampling frequencies without using a band pass filter for removing aliasing noise, by sampling at the K different sampling frequencies and independently Fourier transforming each of the samples to obtain frequency spectra from 0 to a Nyquist frequency group, setting frequency step widths of 1 to K frequency spectra to be equal, generating a vector (y) by combining each frequency spectrum element, and combining frequencies corresponding to a frequency group given at a resolution equivalent to the frequency step width from 0 to an upper limit frequency of a system and frequency elements corresponding to aliasing frequencies in the Nyquist frequency group and the vector (y ) is set to a coefficient corresponding to the frequency characteristic from the signal branch point before aliasing in the Nyquist frequency group to the analog-digital converter (1 if there is no frequency characteristic), and other elements are set to zero to generate a coefficient matrix (A). Using the vector (y) and the coefficient matrix (A), a solution (x) of (y) = (A)(x) is calculated by an over-determined algorithm. As a condition for the solution (x) of the over-determined algorithm, when a solution xe when aliasing in the Nyquist frequency group is zero or even and a solution xo when aliasing in the Nyquist frequency group at a frequency in the same desired frequency band group is odd, the condition that xo and xe are complex conjugates is implemented in the iterative algorithm of the over-determined algorithm, and a predetermined band is extracted from the vector of the solution (x) of the over-determined algorithm. [Effects of the Invention]

[0014] In one aspect, the spectrum regeneration process can be implemented using software. [Brief explanation of the drawings]

[0015] [Figure 1] FIG. 1 is a block diagram schematically illustrating the configuration of a receiver as a conventional example. [Figure 2]2 is a diagram illustrating a spectrum regeneration process performed by the receiver shown in FIG. 1. FIG. [Figure 3] 3 is a diagram illustrating details of the spectrum reproduction process shown in FIG. 2. FIG. [Figure 4] 10(a) to 10(c) are graphs showing the RF spectrum of the original signal in the case where there is no BPF in the related example. [Figure 5] 10(a) to 10(c) are graphs showing the results of RF spectrum reconstruction after applying compressed sensing in a related example where there is no BPF. [Figure 6] 10 is a graph showing the characteristics of a BPF for removing aliasing noise in a related example. [Figure 7] 10(a) to 10(c) are graphs showing the RF spectrum of the original signal when a BPF is present in a related example. [Figure 8] 10(a) to 10(c) are graphs showing the results of RF spectrum reconstruction after applying compressed sensing in the case where a BPF is present in a related example. [Figure 9] FIG. 2 is a block diagram illustrating a configuration example of a receiver according to an embodiment. [Figure 10] FIG. 10 is a block diagram schematically illustrating an example of the configuration of a receiver according to a modified example. [Figure 11] FIG. 11 is a diagram schematically illustrating the relationship between the RF complex spectrum and the observed spectrum in the receiver shown in FIGS. 9 and 10. [Figure 12] 11 is a graph showing the characteristics of an LPF for band limiting in the receiver shown in FIG. 10. [Figure 13] 11(a) to 11(c) are graphs showing undersampling spectra in the receivers shown in FIGS. [Figure 14] 11(a) to 11(d) are graphs showing the RF spectrum of the original signal in the receiver shown in FIGS. [Figure 15] 11(a) to 11(d) are graphs showing the results of RF spectrum reproduction in the receiver shown in FIGS. DETAILED DESCRIPTION OF THE INVENTION

[0016] Hereinafter, embodiments will be described with reference to the drawings. However, the embodiments described below are merely examples, and are not intended to exclude various modifications and applications of techniques not explicitly stated in the embodiments. In other words, the present embodiments can be implemented with various modifications within the scope of the spirit thereof.

[0017] Furthermore, each drawing does not necessarily include only the components shown in the drawing, but may include other components. In the drawings below, parts with the same reference numerals indicate the same or similar parts unless otherwise specified.

[0018] [A] Related Examples Figures 4(a) to 4(c) are graphs showing the RF spectrum of the original signal in the related example without BPF7. Figures 5(a) to 5(c) are graphs showing the RF spectrum reconstruction results after applying compressed sensing in the related example without BPF7.

[0019] Figures 4 and 5 respectively show the RF spectrum of the original signal and the RF spectrum regeneration result when the BPF 7 is omitted from the receiver 600 shown in Figure 1. The graphs shown in Figures 4 and 5 each show the relationship between frequency Freq. (MHz) and radio wave intensity p (dBm).

[0020] Figures 4(a) and 5(a) show the spectrum in the 920 MHz band, Figures 4(b) and 5(b) show the spectrum in the 2.4 GHz band, and Figures 4(c) and 5(c) show the spectrum in the 5 GHz band.

[0021] As shown in Figures 5(a) to 5(c), without BPF7, the spectrum cannot be accurately reproduced in any band. In particular, as shown in Figures 5(b) and 5(c), the spectrum cannot be completely reproduced in the 2.4 GHz and 5 GHz bands.

[0022] FIG. 6 is a graph showing the characteristics of the BPF 7 for removing aliasing noise in a related example.

[0023] 6 shows the relationship between frequency Freq. (MHz) and gain (dB), where symbols C1 to C4 indicate suppression bands and symbols C5 to C7 indicate pass bands.

[0024] The pass band indicated by reference symbol C5 is the 920 MHz band, the pass band indicated by reference symbol C6 is the 2.4 GHz band, and the pass band indicated by reference symbol C7 is the 5 to 6 GHz band.

[0025] Figures 7(a) to 7(c) are graphs showing the RF spectrum of the original signal when a BPF is used in a related example. Figures 8(a) to 8(c) are graphs showing the RF spectrum regeneration results after applying compressed sensing when a BPF is used in a related example.

[0026] 7 and 8 respectively show the RF spectrum of the original signal and the RF spectrum regeneration result of the receiver 600 having the BPF 7 shown in Fig. 1. The graphs shown in Fig. 7 and 8 respectively show the relationship between frequency Freq. (MHz) and radio wave intensity p (dBm).

[0027] Figures 7(a) and 8(a) show the spectrum in the 920 MHz band, Figures 7(b) and 8(b) show the spectrum in the 2.4 GHz band, and Figures 7(c) and 8(c) show the spectrum in the 5 GHz band.

[0028] As shown in FIGS. 8(a) to (c), when BPF7 is used, the spectrum can be reproduced in all bands.

[0029] Thus, in the related example, in order to realize spectrum regeneration, it is necessary to install the BPF 7 as hardware in the receiver 600.

[0030] [B] Embodiment FIG. 9 is a block diagram schematically illustrating an example of the configuration of a receiver 100 according to an embodiment.

[0031] The receiver 100 includes an antenna 1, a low noise amplifier (LNA) 21, one or more (k in the illustrated example) ADCs (analog-to-digital converters) 31, one or more (K in the illustrated example) CLKs 32, one or more (K in the illustrated example) Fast Fourier Transforms (FFTs) 4, and a spectrum regeneration processing unit 5.

[0032] In the receiver 100, a signal received by the antenna 1 is amplified by the LNA 21. The amplified signal is converted from an analog signal to a digital signal by a plurality of ADCs 31. Each ADC 31 is controlled by a CLK 32. The digital signal output from each ADC 31 is fast Fourier transformed by a plurality of FFTs 4. Then, the original spectrum including the noise of the entire band is regenerated by the spectrum regeneration processor 5 using the least squares method.

[0033] The spectrum reproduction processing unit 5 is an example of a processing unit, and may be realized by a field programmable gate array (FPGA), a dedicated circuit, or the like.

[0034] In the receiver 100, a filter for suppressing aliasing noise is not required, and SDR spectrum regeneration processing can be achieved.

[0035] FIG. 10 is a block diagram schematically illustrating an example of the configuration of a receiver 100a according to a modified example.

[0036] In the receiver 100a, a low pass filter (LPF) 22 is provided between the LNA 21 and the plurality of ADCs 31 of the receiver 100 shown in FIG.

[0037] In a normal RF device, the LPF 22 is not actually required because the system bandwidth is determined by the upper limit bandwidth of the LNA 21, the upper limit bandwidth of the line, etc., but it is installed to set the bandwidth in the system during simulation. In an actual circuit, the LPF 22 may be installed from the perspective of setting the upper limit frequency of the system in order to shorten the calculation time.

[0038] FIG. 11 is a diagram schematically showing the relationship between the RF complex spectrum and the observed spectrum in the receivers 100 and 100a shown in FIGS.

[0039] 9 and 10 correspond to the observed spectra indicated by symbols D1 to D3. The spectrum regeneration processing unit 5 convolves each RF spectrum with the observed spectrum and performs linear equation generation in accordance with the condition shown in the following equation 1. In this embodiment, calculations are performed under overdetermined conditions, so the condition of equation 1 (half of the sum of all sampling frequencies is equal to or greater than the upper limit frequency of the system) is necessary.

number

[0040] fs k is the frequency of each CLK32. k If is the sampling number of each ADC31, the frequency resolution RBW = fs k / N k is constant, that is, all sampling frequencies are N k times, and in this embodiment, N k is expressed as an arithmetic series as follows: In this embodiment, the number of ADCs 31 is set to K=15.

[0041] Nsmax=8192 fsmax=1GHz dNs=2 Ns k = Nsmax-dNs(k-1) fs k =Ns kfsmax / Nsmax

[0042] The x, y vectors and matrix A for compressed sensing are defined by the following equations 2 to 5.

number

number

number

number

[0043] In equation 3, the complex undersampling spectrum sampled by ADC31 at clock 1 is y 1 m’ (m (1) =1 to M (1) ) and the complex undersampling spectrum sampled by ADC31 at clock K is y K m K ’ (m (K) =1 to M (K) ) where M-1 corresponds to the discrete frequency after FFT4 divided by RBW, and corresponds to Ns / 2 when the number of clocks is Ns.

[0044] In Equation 4, the complex spectrum in the full band is shown, and the pair of x1, x2, the pair of x3, x4, ... x 2n-1 ,x 2n The sets of are the same frequency spectrum corresponding to zero and even foldback and odd foldback, respectively. Note that N is the degree of freedom, and the following condition of Equation 6 is added (n is 0 or a positive even number).

number

[0045] In equation 5, if the frequency of the x vector is zero or an integer multiple of the Mxfy_d frequency (half the sampling frequency), set both to 1. If the frequency of the x vector is zero or an odd fold of the Mxfy_d frequency, set the element after the spectrum pair to 1. If the frequency of the x vector is zero or an even fold of the Mxfy_d frequency, set the element before the spectrum pair to 1. Set all others to 0.

[0046] The spectrum interval is defined by the following equation 7. Tt is the sampling time (constant), and the frequency intervals of the x and y vectors are the same.

number

[0047] In this patent, it is considered that the spectrum of x is completely filled under the overdetermined condition, and the x vector is found from the formula 2. In this embodiment, the least squares method is used as the algorithm, but other overdetermined condition type algorithms may also be used.

[0048] In this way, the FFT 4 does not use a bandpass filter for aliasing noise removal. When generating a received frequency spectrum from time-axis data sampled at K different sampling frequencies, it samples at K different sampling frequencies and independently Fourier transforms each to obtain a frequency spectrum from 0 to the Nyquist frequency group. The spectrum regeneration processor 5 sets the frequency step widths of 1 to K frequency spectra to be equal and generates a vector (y) combining the respective frequency spectrum elements. The spectrum regeneration processor 5 sets the matrix elements corresponding to the frequencies corresponding to the frequency groups given at a resolution equivalent to the frequency step width from 0 to the system's upper limit frequency, the frequency elements corresponding to the aliasing frequencies in the Nyquist frequency group, and the elements of the vector (y), as coefficients corresponding to the frequency characteristics from the signal branch point before aliasing in the Nyquist frequency group to the ADC 31 (1 if there is no frequency characteristic), and generates a coefficient matrix (A) with zeros for the other elements. In the above example, (A) is formed assuming that there is no frequency characteristic before input to each ADC 31. The spectrum regeneration processor 5 uses a vector (y) and a coefficient matrix (A) to calculate a solution (x) of (y) = (A)(x) using an overdetermined algorithm (least squares method in this embodiment). The spectrum regeneration processor 5 implements, as a condition for the solution (x), a condition that xo and xe are complex conjugates, where xe is a solution when the aliasing in the Nyquist frequency group is zero or even, and xo is a solution when the aliasing in the Nyquist frequency group is odd at a frequency in the same desired frequency band group, within the least squares iterative algorithm. The spectrum regeneration processor 5 extracts a predetermined band from the vector of the least squares solution (x). While the least squares method is used as the overdetermined algorithm in this embodiment, other overdetermined algorithms may also be used.

[0049] FIG. 12 is a graph showing the characteristics of the LPF 22 for band limiting in the receiver 100a shown in FIG.

[0050] 12 shows the relationship between frequency Freq. (MHz) and gain (dB), where symbol E1 is the suppression band and symbol E2 is the pass band.

[0051] The passband indicated by symbol E2 is a band of 6 GHz or less, which allows the frequency band determined by the system to be set.

[0052] 13(a) to 13(c) are graphs showing undersampling spectra in the receivers 100 and 100a shown in FIGS.

[0053] The graphs shown in Figure 13 are for m and |y m | indicates the relationship.

[0054] Figure 13(a) shows the spectrum after sampling by the ADC 31 of clock 1 (fs = 1000 MHz), Figure 13(b) shows the spectrum after sampling by the ADC 31 of clock 2 (fs = 999.0234 MHz), and Figure 13(c) shows the spectrum after sampling by the ADC 31 of clock 3 (fs = 998.0469 MHz). Note that the noise is NF (dB) + G (dB) = 85 dB.

[0055] 13(a) to (c), the spectrum can be reproduced for any clock even without a BPF. Note that the spectrum shown is three representative spectra out of a total of 15.

[0056] Figures 14(a) to 14(d) are graphs showing the RF spectrum of the original signal in the receivers 100 and 100a shown in Figures 9 and 10. Figures 15(a) to 15(d) are graphs showing the results of RF spectrum reproduction in the receivers 100 and 100a shown in Figures 9 and 10.

[0057] The graphs shown in FIGS. 14 and 15 each show the relationship between frequency Freq. (MHz) and radio wave intensity p (dBm).

[0058] Figures 14(a) and 15(a) show the spectrum in the 920 MHz band, Figures 14(b) and 15(b) show the spectrum in the 2.4 GHz band, Figures 14(c) and 15(c) show the spectrum in the 5 GHz band, and Figures 14(d) and 15(d) show the spectrum over all bands. Note that the noise is NF(dB)+G(dB)=85 dB.

[0059] As shown in FIGS. 15(a) to (d), even without a BPF, the spectrum can be reproduced in any band.

[0060] [C] Other The disclosed technology is not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the embodiments. The configurations and processes of the embodiments can be selected or combined as needed. [Explanation of symbols]

[0061] 100, 100a, 600: Receiver 1,6: Antenna 21: LNA 22:LPF 7:BPF 8: Simultaneous undersampling processing unit 31,81:ADC 32,82:CLK 4: FFT 9: RF spectrum restoration processing section 5: Spectrum regeneration processing section

Claims

1. In a frequency spectrum regeneration method using an undersampling receiving method, No band-pass filter is used to remove aliasing noise. When generating the received frequency spectrum from time-axis data sampled at K different sampling frequencies, Sampling is performed at the K types of sampling frequencies, and each sampling frequency is independently Fourier transformed to obtain a frequency spectrum from 0 to the Nyquist frequency group; The frequency step widths of 1 to K frequency spectra are set to be equal, and a vector (y) is generated by combining the respective frequency spectrum elements; a coefficient matrix (A) is generated in which the frequencies corresponding to the frequency group given at a resolution equivalent to the frequency step width in the range from 0 to the upper limit frequency of the system, the frequency elements corresponding to the aliasing frequencies in the Nyquist frequency group, and the matrix elements corresponding to the elements of the vector (y) are set as coefficients corresponding to the frequency characteristics from the signal branch point before aliasing in the Nyquist frequency group to the analog-digital converter (1 if there is no frequency characteristic), and the other elements are set to zero; Using the vector (y) and the coefficient matrix (A), a solution (x) of (y) = (A) (x) is calculated by an overdetermined condition type algorithm; As a condition for the solution (x) of the overdetermined algorithm, when a solution xe is a solution when the number of folds in the Nyquist frequency group is zero or even, and a solution xo is a solution when the number of folds in the Nyquist frequency group is odd at a frequency in the same desired frequency band group, a condition that xo and xe are complex conjugates is implemented in the iterative algorithm of the overdetermined algorithm; Extracting a predetermined band from the vector of the solution (x) of the overdetermined algorithm; Frequency spectrum reproduction method.

2. placing a low pass filter before sampling the data to determine the upper frequency limit of the system; 2. The method of claim 1, wherein the frequency spectrum is reproduced.

3. Half the sum of all sampling frequencies is greater than or equal to the upper frequency limit of the system; 3. The method for reproducing a frequency spectrum according to claim 1 or 2.

4. All sampling frequencies are N times the frequency resolution, where N is expressed as an arithmetic series: The method for reproducing a frequency spectrum according to any one of claims 1 to 3.

5. In a receiver that performs frequency spectrum regeneration using an undersampling reception method, No band-pass filter is used to remove aliasing noise. When generating the received frequency spectrum from time-axis data sampled at K different sampling frequencies, Sampling is performed at the K types of sampling frequencies, and each sampling frequency is independently Fourier transformed to obtain a frequency spectrum from 0 to the Nyquist frequency group; The frequency step widths of 1 to K frequency spectra are set to be equal, and a vector (y) is generated by combining the respective frequency spectrum elements; a coefficient matrix (A) is generated in which the frequencies corresponding to the frequency group given at a resolution equivalent to the frequency step width in the range from 0 to the upper limit frequency of the system, the frequency elements corresponding to the aliasing frequencies in the Nyquist frequency group, and the matrix elements corresponding to the elements of the vector (y) are set as coefficients corresponding to the frequency characteristics from the signal branch point before aliasing in the Nyquist frequency group to the analog-digital converter (1 if there is no frequency characteristic), and the other elements are set to zero; Using the vector (y) and the coefficient matrix (A), a solution (x) of (y) = (A) (x) is calculated by an overdetermined condition type algorithm; As a condition for the solution (x) of the overdetermined algorithm, when a solution xe is a solution when the number of folds in the Nyquist frequency group is zero or even, and a solution xo is a solution when the number of folds in the Nyquist frequency group is odd at a frequency in the same desired frequency band group, a condition that xo and xe are complex conjugates is implemented in the iterative algorithm of the overdetermined algorithm; Extracting a predetermined band from the vector of the solution (x) of the overdetermined algorithm; A receiver comprising a processing unit.

Images