Filters and Electronics

Abstract

A filter and electronic device, wherein the filter is provided with a passband, a transition band and a plurality of filtering subbands, and the filter comprises: a first frequency shifting module, used for performing a first frequency shifting operation on an input signal to obtain a frequency-shifted input signal, so as to prevent the transition band from falling into the boundary of adjacent filtering subbands; a subband separation module, used for separating each subband signal from the frequency-shifted input signal according to the filtering subband; an edge filtering module, used for filtering the separated subband signal, so as to retain the subband signal located in the passband, partially filter out the subband signal located in the transition band, and completely filter out other subband signals, so as to obtain a filtered subband signal; a subband merging module, used for merging the filtered subband signals; and a first inverse frequency shifting module, used for performing an inverse operation of the first frequency shifting operation on the merged subband signal, so as to obtain an output signal. The filter can realize narrow transition band filtering, and has low complexity.

Description

Technical Field

[0001] The present invention relates to the field of signal processing, and in particular to a filter and an electronic device. Background Art

[0002] Digital filters usually include two structures: Finite Impulse Response (FIR) filter and Infinite Impulse Response (IIR) filter. FIR filter is the most commonly used filter, which is easy to achieve linear phase requirements and has stable characteristics. IIR filter has much smaller implementation resources than FIR filter, but because of stability issues and difficulty in achieving linear phase, it is not widely used in practice.

[0003] In various types of data communications, digital filters are often used to process signals (for example, voice signals). Digital filters are a common device for implementing frequency selection in the digital domain, which can allow specific frequency components in the signal to pass through while greatly attenuating other frequency components. The frequency range between the passband and stopband of the filter is called the transition band. In order to make full use of spectrum resources, the filter is usually required to have a relatively narrow transition band. However, when the transition band of the filter is very narrow, the order of the FIR filter is very high, and the design of the filter is complex, and the resource requirements are also high during implementation.

[0004] In summary, how to provide a low-complexity and narrow-transition-band digital filter has become a problem that needs to be solved urgently. Summary of the invention

[0005] The technical problem solved by the present invention is how to provide a digital filter with low complexity and narrow transition band.

[0006] To solve the above technical problems, an embodiment of the present invention provides a filter, which is provided with a passband, a transition band and multiple filter sub-bands, and includes: a first frequency shifting module, used to perform a first frequency shifting operation on an input signal to obtain a frequency-shifted input signal, so as to prevent the transition band from falling into the boundary part of adjacent filter sub-bands; a sub-band separation module, used to separate each sub-band signal from the frequency-shifted input signal according to the filter sub-band; an edge filtering module, used to filter the separated sub-band signals, so as to retain the sub-band signals located in the passband, partially filter out the sub-band signals where the transition band is located, and completely filter out other sub-band signals, so as to obtain the filtered sub-band signals; a sub-band merging module, used to merge the filtered sub-band signals; a first inverse frequency shifting module, used to perform an inverse operation of the first frequency shifting operation on the merged sub-band signals, so as to obtain an output signal.

[0007] Optionally, when the first frequency shift operation is performed on the input signal, the offset frequency f offset Determined according to the following formula: Among them, f s is the sampling frequency of the input signal, M is the number of sub-band signals in the input signal, and the deviation of the passband frequency from the center frequency of the adjacent filter sub-band is It is expressed by the following formula: Among them, f p is the frequency of the input signal, and round() means rounding the value in the brackets to the nearest integer.

[0008] Optionally, the edge filtering module includes: a zero gain unit, used to set the gain of the other sub-band signals to zero so as to completely filter out the other sub-band signals; a filtering unit, used to perform asymmetric filtering on the sub-band signal where the transition band is located so as to partially filter out the sub-band signal where the transition band is located to obtain an output signal of the filtering unit, wherein the filtering unit has a delay; a delay unit, used to perform delay processing on the sub-band signal located in the passband to obtain an output signal of the delay unit, and the output signal of the delay unit is aligned with the output signal of the filtering unit in time domain; wherein the filtered sub-band signal includes the output signal of the filtering unit and the output signal of the delay unit.

[0009] Optionally, the filtering unit includes: a second frequency shifting subunit, used to perform a second frequency shifting operation on the subband signal where the transition band is located; the low-pass filter, used to perform asymmetric filtering on the output signal of the second frequency shifting subunit, the coefficient of the low-pass filter is a real number; and a second inverse frequency shifting subunit, used to perform an inverse operation of the second frequency shifting operation on the output signal of the real filter.

[0010] Optionally, the filtering unit includes a low-pass filter with complex coefficients, which is used to perform asymmetric filtering on the sub-band signal where the transition band is located, so as to partially filter out the sub-band signal where the transition band is located, and obtain an output signal of the filtering unit.

[0011] Optionally, the first frequency shift module includes: a digitally controlled oscillator, used to generate an oscillation signal; and a multiplier, used to multiply the input signal and the oscillation signal to perform a first frequency shift operation on the input signal.

[0012] Optionally, the filter further includes: a speed reduction module, and the sub-band signal output by the sub-band separation module is transmitted to the edge filtering module via the speed reduction module, and the speed reduction module is used to perform speed reduction processing on each sub-band signal.

[0013] Optionally, the sub-band separation module is used to perform analysis filtering and fast discrete transformation processing on the frequency-shifted input signal to obtain each of the sub-band signals; the sub-band merging module is used to perform fast inverse discrete transformation processing and comprehensive filtering on the filtered sub-band signals to merge the filtered sub-band signals.

[0014] An embodiment of the present invention further provides an electronic device, comprising the filter as described above.

[0015] Compared with the prior art, the technical solution of the embodiment of the present invention has the following beneficial effects:

[0016] In the filter provided in the embodiment of the present invention, a first frequency shift module is added to move the transition band of the filter away from the boundary of two adjacent filter sub-bands, the edge filter module performs different filtering operations on different sub-band signals, and then the transition band of the filter is moved back to its original position through the first inverse frequency shift module. The specific operation of the edge filter module is to retain the sub-band signal located in the passband, partially filter out the sub-band signal where the transition band is located, and completely filter out other sub-band signals to obtain the filtered sub-band signal. Compared with the DFTFB in the prior art, the filter in the embodiment of the present invention can be applied to filtering with any bandwidth requirements, it can run at low speed, and requires very few additional resources. At the same time, the number of filter sub-bands included in the filter can be very small, which can effectively reduce the delay of filtering.

[0017] Furthermore, the edge filtering module retains the subband signal in the passband, performs asymmetric filtering on the subband signal in the transition band to partially retain the subband signal in the transition band, and completely filters out other subband signals. Thus, the filter of the present invention can flexibly support different filtering requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a structural schematic diagram of a FRM filter in the prior art;

[0019] Figure 2 It is a structural schematic diagram of a DFTFB in the prior art;

[0020] Figure 3 for Figure 2 A schematic diagram of an output spectrum of DFTFB;

[0021] Figure 4 A schematic diagram of the structure of a filter according to an embodiment of the present invention;

[0022] Figure 5 for Figure 4 A schematic structural diagram of the first frequency shift module in a specific embodiment;

[0023] Figure 6A spectrum diagram of filtering performed by a filter according to an embodiment of the present invention;

[0024] Figure 7 A schematic diagram of the structure of an analysis filter bank according to an embodiment of the present invention;

[0025] Figure 8 A schematic diagram of the structure of a comprehensive filter bank according to an embodiment of the present invention

[0026] Fig. 9 A schematic diagram of a filter bank of an edge filtering module according to an embodiment of the present invention;

[0027] Fig.10 A structural schematic diagram of a filtering unit according to an embodiment of the present invention;

[0028] Fig.11 This is a spectrum diagram of a filtering unit according to an embodiment of the present invention. DETAILED DESCRIPTION

[0029] As mentioned in the background art, in a traditional digital filter, when the transition band of the filter is very narrow, the order of the FIR filter is very high, and the design of the filter is complex, and the resource requirements for implementation are also high.

[0030] The performance parameters of the filter include: passband bandwidth (in -f p ~f p denoted by -f), cut-off bandwidth (also called stop-band bandwidth, c ~f c ), passband ripple δ p and out-of-band suppression δ c Among them, the channel ripple δ p It represents the difference between the maximum and minimum amplitudes of the passband in the frequency response of the filter, and the out-of-band suppression δ c Indicates the degree to which the filter suppresses signals outside the passband. In the design of digital filters (such as FIR filters), the filter transition band f c -f p The size of the passband ripple δ p , out-of-band suppression δ c Determines the length of the filter, that is, the order. Transition band f c -f p The narrower the transition band, the higher the order of the filter, the more resources are needed to implement the filter, and the higher the cost. Even when the design requires a very narrow transition band, the digital filter (such as FIR filter) may reach hundreds or even thousands of orders, and the filter complexity is very high. Directly implementing this type of filter will consume a lot of resources.

[0031] In order to solve the high implementation complexity problem of narrow transition band digital filters, many methods have been proposed, such as Frequency Response Masking (FRM) filters and Discrete Fourier Transform Filter Bank (DFTFB).

[0032] Among them, FRM filter is a very popular narrow transition band filter. Figure 1 , Figure 1 FIG. 1 is a schematic diagram of the structure of a FRM filter 10 in the prior art. The FRM filter 10 includes a filter H(z -N ), shielding filter H a (z -1 ) and H b (z -1 ). Filter H(z -N ), each uniformly spaced N coefficients has only one non-zero value, so the resources required by the FRM filter 10 are much smaller than those of the traditional FIR filter. At the same time, the group delay of the FRM filter 10 is not much different from the delay of the traditional FIR filter. For example, the group delay of the FRM filter is generally not greater than 110% of the delay of the traditional FIR filter. Although the FRM filter 10 has the advantages of small resources and small delay. However, the FRM filter 10 also has the following three disadvantages.

[0033] Disadvantage 1: Poor flexibility. It is not easy to select the value of N for different design requirements. Specifically, different filter bandwidths correspond to different optimal N values ​​and different orders of H. a (z -1 ) and H b (z -1 ). This means that when the system needs to flexibly support different requirements, the FRM filter needs to support different N values, and H a (z -1 ) and H b (z -1 ) also needs to have enough coefficients to meet all the requirements. Disadvantage 2: The value of N cannot be too large. Specifically, when the value of N is large, H a (z -1 ) or H b (z -1 ) will also have a lot of coefficients. And it is also difficult to optimize the parameters at this time. This means that for the case of a particularly narrow transition band, the FRM method still requires more resources. Although the current multi-stage FRM filter can reduce more resources, its implementation structure is very complex and less flexible. Disadvantage 3, the optimal parameters of FRM require powerful mathematical optimization tools to obtain.

[0034] Please refer to Figure 2 , Figure 2 , which is a schematic structural diagram of a DFTFB 20 in the prior art. The DFTFB 20 may include the following modules: an analysis filter module 201, a Fast Fourier Transform (FFT) module 202, a weighting module 203, an Inverse Fast Fourier Transform (IFFT) module 204, and a synthesis filter module 205. The analysis filter module 201 and the FFT module 202 separate the input signal into multiple sub-bands. The weighting module 203 filters the signal by taking weighting values (the values of the weighting values can be 0 or 1) for different sub-bands. The IFFT module 204 and the synthesis filter module 205 then combine all the sub-bands into an output signal, and the spectrum of the output signal is as Figure 3 shown. Both the analysis filter module 201 and the synthesis filter module 205 adopt a polyphase filter structure, which can achieve filtering after deceleration, so the resource requirements are very small. The FFT module 202 and the IFFT module 204 both have very effective implementation methods. Therefore, the resource requirements of this method are very low. The DFTFB filter has the advantage of requiring less implementation resources than the FRM filter. However, this method has the following two disadvantages:

[0035] Disadvantage 1: Low flexibility. Specifically, the DFTFB 20 can only support the requirements of certain specific filtering bandwidths. It can be seen from Figure 3 that the DFTFB 20 can only support filtering with a certain bandwidth and a transition band of f2 - f1. The certain bandwidth can be expressed as: fs×k / M + f1, where the value of the frequency point k is expressed as -M / 2 ≤ k < M / 2, and for filtering with a transition band frequency of f2 - f1, the transition band frequency range is [f1, f2]. Among them, fs is the sampling rate of the input signal; M is the number of filtering sub-bands (sub-bands can also be called sub-carriers); f1 and f2 are the passband frequency and the cut-off frequency of the prototype filter of the analysis filter 201, respectively. This flexibility problem can be alleviated by increasing the number of filtering sub-bands M, but it still cannot be completely solved, and it will bring the delay problem of the following Disadvantage 2.

[0036] Disadvantage 2: When the number of sub-bands M is very large, the delay of the system is very large. Specifically, the lengths of the prototype filters for analysis and synthesis filtering are proportional to M, which also means that the delay of the entire system is also proportional to the number M.

[0037] Combined with Figures 1 to 3From the introduction of FRM filter 10 and DFTFB20, it can be seen that the existing narrow transition band digital filters cannot take into account the characteristics of low latency, small resource occupation, and low structural complexity.

[0038] To solve the above technical problems, an embodiment of the present invention provides an improved filter, which can be applied to a communication system, and the input signal can be various appropriate signals, such as a voice signal or a communication carrier. The filter of the embodiment of the present invention is described below in conjunction with the accompanying drawings.

[0039] See also Figure 4 , Figure 4 This is a structural schematic diagram of a filter 40 according to an embodiment of the present invention. The filter 40 has a passband, a transition band and multiple filtering sub-bands. The filter 40 may include a first frequency shift module 401, a sub-band separation module 402, an edge filtering module 403, a sub-band merging module 404 and a first inverse frequency shift module 405.

[0040] The first frequency shift module 401 is used to perform a first frequency shift operation on the input signal to obtain a frequency-shifted input signal, so as to prevent the transition band from falling into a boundary portion of adjacent filter sub-bands.

[0041] Optional, see Figure 5 The first frequency shift module 401 may include a numerically controlled oscillator (NCO) 4011 and a multiplier 4012. The numerically controlled oscillator 4011 is used to generate an oscillation signal; the multiplier 4012 is used to multiply the input signal and the oscillation signal to perform a first frequency shift operation on the input signal.

[0042] When the transition band of the filter 40 is located at the boundary of two adjacent filter sub-bands, the subsequent edge filter module 403 cannot achieve a satisfactory filtering target without the first frequency shift module 401 and the first inverse frequency shift module 405. The function of the first frequency shift module 401 is to move the transition band of the filter 40 away from the boundary of two adjacent filter sub-bands. Figure 6 Provide explanation.

[0043] The edge filtering module 403 includes a filter bank, which includes a plurality of filter circuits. The filtering range of each filter circuit may correspond to a filter sub-band, and there is a boundary between the filter sub-bands of two adjacent filter circuits. More specifically, there is an overlapping portion between adjacent filter sub-bands, and the overlapping portion is the "boundary portion".

[0044] Figure 6This is a spectrogram of the filtering of a filter according to an embodiment of the present invention. The horizontal axis of the coordinate axis is the frequency (denoted by f), and the vertical axis of the coordinate axis is the amplitude of the signal (denoted by h). Among them, fs is the sampling rate of the input signal, and the frequency band of the input signal is [-fs / 2, fs / 2]. The frequency band of the input signal may include multiple sub-band signals, and [-f3, f3] represents one of the sub-band signals. If the passband frequency band of the filter is [-fa, fa], when the filter 40 is a narrow transition band filter, the transition band of the filter (the transition band frequency point is close to -fa or fa) may be located at the junction of two adjacent filtering sub-bands. Figure 6 The frequency at the junction of two adjacent filtering sub-bands can be expressed as: fs×k / M + [f1, f2]. Among them, fs is the sampling rate of the input signal, the value of the frequency point k is expressed as -M / 2 ≤ k < M / 2, [f1, f2] is the frequency band of the junction part described in the figure, and M is the number of the filtering sub-bands.

[0045] The first frequency shift module 401 can shift the transition band out of the junction of two adjacent filtering sub-bands. Further, it can be shifted to any frequency outside the junction part. The first frequency shift operation can be expressed by the following formula: f’ = f0 + f offset , where f0 is the frequency before the first frequency shift operation (for example, the frequency of a certain frequency point in the input signal), and f’ is the frequency value after the first frequency shift operation. Thus, the frequency points in the transition band of the input signal can be shifted to any frequency except the junction part.

[0046] Furthermore, the transition band can be shifted to the vicinity of the center frequency of a certain filtering sub-band. For example, the transition band of the filter 40 is shifted to the vicinity of the center frequency of the filtering sub-band closest to this junction part. As Figure 6 shown, the passband frequency band of the filter 40 is shifted from [-fa, fa] to [-fb, fb].

[0047] At this time, the offset frequency f offset is determined according to the following formula (1):

[0048]

[0049] where f s is the sampling frequency of the input signal, M is the number of filtering sub-bands, and the deviation between the passband frequency point and the center frequency of the adjacent filtering sub-band is expressed by the following formula (2):

[0050]

[0051] where f pis the frequency of the input signal, and round() means rounding the value in the brackets to the nearest integer.

[0052] Still reference Figure 4 The sub-band separation module 402 is used to separate each sub-band signal from the frequency-shifted input signal according to the filtering sub-band.

[0053] For details, please refer again to Figure 6 ,according to Figure 6 The frequency band where each filter sub-band is located in the frequency band, the frequency-shifted input signal is divided into M sub-band signals. Each sub-band signal after separation is located in the frequency band where a filter sub-band is located, and M is a positive integer greater than or equal to 1.

[0054] refer to Figure 4 The edge filtering module 403 is used to filter the separated sub-band signals to retain the sub-band signals located in the passband, partially filter out the sub-band signals located in the transition band, and completely filter out other sub-band signals to obtain filtered sub-band signals.

[0055] Please see again Figure 6 , the sub-band signal located in the passband refers to the sub-band signal between the frequency points -fb and fb, and does not include the sub-band signal at the frequency point fb and the sub-band signal at -fb; the sub-band signal at the transition band refers to the sub-band signal at the frequency point fb and the sub-band signal at -fb, and other sub-band signals refer to sub-band signals other than the sub-band signal located in the passband and the sub-band signal at the transition band. The sub-band signal at the transition band refers to Figure 6 The sub-band signal where the intermediate frequency -fb is located and the sub-band signal where the frequency fb is located.

[0056] refer to Figure 4 The sub-band merging module 404 is used to merge the filtered sub-band signals. The sub-band merging module 404 should perform the opposite operation to the sub-band separation module 402.

[0057] In a non-limiting example, the sub-band separation module 402 is used to perform analysis filtering and fast discrete transform (FFT) processing on the frequency-shifted input signal to obtain each of the sub-band signals; the sub-band merging module 404 is used to perform inverse fast discrete transform (IFFT) processing and comprehensive filtering on the filtered sub-band signals to merge the filtered sub-band signals.

[0058] Among them, analysis filtering can be performed by an analysis filter bank, which implements polyphase filtering. Figure 7 , Figure 7 Schematic diagram of the structure of an analysis filter bank according to an embodiment of the present invention. The prototype filter is represented by the following formula: Among them, z -1 Represents the unit delay of the digital filter; the prototype filter h(z -1 ) is decomposed into M polyphase filters h 0 (z -1 ), h 1 (z -1 ), …, h i (z -1 ), …, h M-1 (z -1 ); Figure 7 The “N↓” in represents downsampling the signal by a factor of N, that is, the input signal of each polyphase filter is slowed down by a factor of N.

[0059] Correspondingly, the M filtered sub-band signals are merged by the sub-band merging module 404. Further, the comprehensive filtering can be performed by the comprehensive filter bank. Figure 8 , Figure 8 The schematic diagram of the structure of a comprehensive filter bank according to an embodiment of the present invention is shown in FIG. 1 , wherein the comprehensive filter bank can realize polyphase filtering. The prototype filter thereof is represented by the following formula: Among them, g is the prototype filter of comprehensive filtering, g i is the polyphase filter decomposed from the prototype filter g, 0≤i≤M-1, N is the upsampling multiple value, upsample(g i , N) represents the g i Perform N-fold upsampling (in Figure 8 Indicated by "N↑" in the -i Represents the unit delay of the digital filter.

[0060] In another non-limiting example, the sub-band separation module 402 is used to perform analysis filtering and discrete cosine transform (DCT) processing on the frequency-shifted input signal to obtain each of the sub-band signals; the sub-band merging module 404 is used to perform inverse discrete cosine transform (IDCT) processing and comprehensive filtering on the filtered sub-band signals to merge the filtered sub-band signals. The analysis filtering and comprehensive filtering can refer to the previous description.

[0061] In another non-limiting example, the sub-band separation module 402 may include a half-band filter, that is, the frequency-shifted input signal is separated into the sub-band signals by using the half-band filter.

[0062] The first inverse frequency shift module 405 is used to perform an inverse operation of the first frequency shift operation on the combined sub-band signal to obtain an output signal.

[0063] Optionally, the first inverse frequency shift module 405 may also include an NCO 4011 and a multiplier 4012. The manner in which the first inverse frequency shift module 405 performs the inverse operation of the first frequency shift operation on the merged sub-band signal can refer to the relevant description of the first frequency shift operation performed by the first frequency shift module 401, which will not be repeated here.

[0064] The oscillation signal generated by the NCO (4011) in the first frequency shift module 401 and the oscillation signal generated by the NCO in the first inverse frequency shift module 405 have the same frequency and opposite phase. For example, the input of the NCO 4011 is the parameter f offset , the amplitude of the oscillation signal generated by NCO is 1 and the frequency is f offset The oscillation signal is a single frequency signal. The input of the NCO in the first inverse frequency shift module 405 is the parameter -f offset , the amplitude of the oscillation signal generated is 1 and the frequency is -f offset , this oscillation signal is a single frequency signal.

[0065] In the filter 40 provided in the embodiment of the present invention, a first frequency shift module 401 is added to shift the transition band of the filter 40 away from the boundary of two adjacent filtering sub-bands, and then different filtering operations are performed on different sub-band signals through the edge filtering module 403, and then the transition band of the filter is moved back to its original position through the first inverse frequency shift module 405. The specific operation of the edge filtering module 403 is to retain the sub-band signal located in the passband, partially filter out the sub-band signal where the transition band is located, and completely filter out other sub-band signals to obtain the filtered sub-band signal. Compared with the DFTFB in the prior art (see Figure 2 ), the filter 40 in the embodiment of the present invention can be applied to filtering with any bandwidth requirement, and can run at a low speed, requiring very few additional resources. At the same time, the number of filtering subbands included in the filter 40 can be very small, which can effectively reduce the delay of filtering.

[0066] In one embodiment, Figure 4 The edge filtering module 403 in may include a filter bank, see Fig. 9 , Fig. 9 FIG. 4 is a schematic diagram of a filter bank of an edge filtering module 403 according to an embodiment of the present invention. The filter bank includes a zero gain unit (with Fig. 9 "G=0" in the figure), filtering unit (with Fig. 9 The "h L (z -1 ) and h R (z -1 )”) and delay units (in Figure 6 The "z -d ” indicates), filter unit hR (z -1 ) can be Figure 6 The subband signal where the intermediate frequency fb is located, the filter unit h L (z -1 ) can be Figure 6 The subband signal where the intermediate frequency -fb is located.

[0067] The zero gain unit is used to set the gain of the other sub-band signals to zero to completely filter out the other sub-band signals; the filtering unit is used to perform asymmetric filtering on the sub-band signals where the transition band is located to partially filter out the sub-band signals where the transition band is located to obtain the output signal of the filtering unit, wherein the filtering unit has a delay; the delay unit is used to perform delay processing on the sub-band signals located in the passband to obtain the output signal of the delay unit, and the output signal of the delay unit is aligned in time domain with the output signal of the filtering unit. The filtered sub-band signal includes the output signal of the filtering unit and the output signal of the delay unit.

[0068] The filtering unit (with Fig. 9 The "h L (z -1 ) and h R (z -1 )” means that ) can include the following two implementation methods to achieve asymmetric filtering.

[0069] The implementation method 1 of the filter unit is as follows Fig.10 As shown, Fig.10 The present invention is a schematic diagram of the structure of a filtering unit according to an embodiment of the present invention. The filtering unit may include a second frequency shift subunit 1011, a low pass filter (LPF) 1012 and a second inverse frequency shift subunit 1013. The second frequency shift subunit 1011 is used to perform a second frequency shift operation on the subband signal where the transition band is located; the low pass filter 1012 is used to perform asymmetric filtering on the output signal of the second frequency shift subunit 1011, and the coefficient of the low pass filter 1012 is a real number; the second inverse frequency shift subunit 1013 is used to perform an inverse operation of the second frequency shift operation on the output signal of the low pass filter 1012.

[0070] Optionally, the second frequency shift subunit 1011 and the second inverse frequency shift subunit 1013 may include an NCO and a multiplier, respectively. The manner in which the second frequency shift subunit 1011 and the second inverse frequency shift subunit 1013 perform the second frequency shift operation and the inverse operation of the second frequency shift operation can refer to the relevant description of the first frequency shift module 401 performing the first frequency shift operation, which will not be repeated here. The oscillation signal generated by the NCO in the second frequency shift subunit 1011 and the oscillation signal generated by the NCO in the second inverse frequency shift subunit 1013 have equal frequencies and opposite phases. For example, Fig.10 The input of the NCO in the second frequency shift subunit 1011 is the parameter f NCO , the amplitude of the oscillation signal it generates is 1 and the frequency is f NCO , this oscillation signal is a single frequency signal. Correspondingly, the input of the NCO in the second inverse frequency shift subunit 1013 is the parameter -f NCO , the amplitude of the oscillation signal generated is 1 and the frequency is -f NCO , this oscillation signal is a single frequency signal.

[0071] It should be noted that in addition to implementing the frequency shifting operation in the embodiment of the present invention through NCO and multiplier, including one or more operations of the first frequency shifting operation, the inverse operation of the first frequency shifting operation, the second frequency shifting operation and the inverse operation of the second frequency shifting operation, other methods that can achieve frequency shifting of the signal can also be used, which will not be repeated here.

[0072] See also Fig.11 , Fig.11 is a spectrum diagram of a filter unit according to an embodiment of the present invention, corresponding to Fig. 9 The filter unit h in R (z -1 ). Filter unit h R (z -1 ) is the sub-band signal where the transition band is located. The frequency band of this sub-band signal is [fh1, fh2], and the passband frequency of the filter is fb. The output signal of the second frequency shift sub-unit 1011 (i.e. Fig.11 The frequency band of the subband signal where the transition band is located after the second frequency shift operation) is [fh1', fh2']. The frequency band of the passband of the low-pass filter 1012 is set to [fh1', fh3], and the center point of the passband frequency band of the low-pass filter 1012 is represented by fh0, fh0 = (fh3-fh1') / 2.

[0073] Fig. 9 The filter unit h in L (z -1 ) and Fig.11 The filter unit h is symmetrically distributed with the vertical axis as the axis. L (z -1) of LPF and filter unit h R (z -1 ) is the same as the LPF, but the filter unit h L (z -1 ) NCO With h R (z -1 )'s fNCO are opposite to each other.

[0074] In the implementation mode 1 of the filter unit, the filter unit h in the filter bank can be adjusted. L (z -1 ) and h R (z -1 ) in the NCO parameter value to obtain filters with different filtering results and the same performance, so that the filter of the embodiment of the present invention can flexibly support different filtering requirements.

[0075] In implementation mode 2 of the filtering unit, the filtering unit may include a low-pass filter with complex coefficients, which is used to perform asymmetric filtering on the sub-band signal where the transition band is located, so as to partially filter out the sub-band signal where the transition band is located to obtain the output signal of the filtering unit. Thus, by adjusting the coefficient value of the low-pass filter with complex coefficients, filters with different filtering results and the same performance can be obtained, so that the filter of the embodiment of the present invention can flexibly support different filtering requirements.

[0076] In this embodiment, the edge filtering module retains the subband signal in the passband, performs asymmetric filtering on the subband signal in the transition band to partially retain the subband signal in the transition band, and completely filters out other subband signals. Thus, the filter of the present invention can flexibly support different filtering requirements.

[0077] Please see again Fig.11 The low-pass filter with complex coefficients can be directly used without the second frequency shift operation. Fig.11 The sub-band signal where the transition band is located is filtered to obtain the signal corresponding to the passband of the low-pass filter.

[0078] In one embodiment, the filter may further include: a speed reduction module, through which the sub-band signals output by the sub-band separation module are transmitted to the edge filtering module, and the speed reduction module is used to perform speed reduction processing on each sub-band signal.

[0079] After the sub-band separation module outputs the sub-band signal, the bandwidth of each sub-band signal is much smaller than the bandwidth of the input signal. By reducing the speed of each sub-band signal through the speed reduction module, the computing resources of each module can be reduced. Optionally, to avoid aliasing after speed reduction, the speed reduction multiple N of the speed reduction module must be less than M, usually N = M / 2.

[0080] Compared with the existing FRM filter, the filter of the embodiment of the present invention has the following three advantages.

[0081] Advantage 1: Fewer resources are consumed. Generally speaking, the comparison of resources can be based on the multiplier resources, because the multiplier is the largest part of all consumed resources. The inventors have found through research that the multiplier used in the filter of the embodiment of the invention is smaller than the multiplier used in the existing two-stage FRM filter. Under the condition of achieving the same narrowband filtering, the filter of the embodiment of the invention consumes fewer resources.

[0082] Advantage 2: Higher flexibility. Specifically, the edge filtering module of the filter in the embodiment of the present invention can adjust some parameters (such as adjusting the filter unit h in the filter group) to obtain the edge filtering module of the filter. L (z -1 ) and h R (z -1 ) or the coefficient value of the low-pass filter with complex coefficients), that is, filters with different filtering results and the same performance can be obtained to flexibly support different filtering requirements. For the FRM filter, any change in the parameters will affect the order and value of the filter inside it.

[0083] Advantage 3: Simple design: Specifically, compared with the existing FRM filter, the filter of the embodiment of the present invention only needs a simple filter design tool to complete, while the design of the FRM filter requires a powerful numerical optimization tool to complete, otherwise the result obtained will be very unsatisfactory.

[0084] An embodiment of the present invention further provides an electronic device, the electronic device comprising: Figures 4 to 11 The filter 40 of the illustrated embodiment.

[0085] For more information about the working principles and working methods of electronic devices, please refer to the above Figures 4 to 11 The relevant description of the filter 40 in will not be repeated here.

[0086] It should be understood that the term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article indicates that the associated objects before and after are in an "or" relationship.

[0087] The "plurality" appearing in the embodiments of the present application refers to two or more.

[0088] The first, second, etc. descriptions appearing in the embodiments of the present application are only used for illustration and distinction of the description objects. There is no order, nor do they indicate any special limitation on the number of devices in the embodiments of the present application, and cannot constitute any limitation on the embodiments of the present application.

[0089] The "connection" that appears in the embodiments of the present application refers to various connection methods such as direct connection or indirect connection to achieve communication between devices, and the embodiments of the present application do not impose any limitations on this.

[0090] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the claims.

Claims

1. A filter, It is characterized in that The filter is provided with a passband, a transition band and a plurality of filtering subbands, and the filter comprises: The first frequency shift module is used to perform a first frequency shift operation on the input signal to obtain a frequency-shifted input signal to prevent the transition band from falling into the boundary of adjacent filter sub-bands; when the first frequency shift operation is performed on the input signal, the offset frequency of the frequency shift Determined according to the following formula: ;in, is the sampling frequency of the input signal, M is the number of sub-band signals in the input signal, and the deviation of the passband frequency from the center frequency of the adjacent filter sub-band is It is expressed by the following formula: ;in, is the frequency of the input signal, and round() means rounding the value in the brackets to the nearest integer; A sub-band separation module, used for separating each sub-band signal from the frequency-shifted input signal according to the filtering sub-band; An edge filtering module is used to filter the separated sub-band signals to retain the sub-band signals in the passband, partially filter out the sub-band signals in the transition band, and completely filter out other sub-band signals to obtain filtered sub-band signals; A sub-band merging module, used for merging the filtered sub-band signals; The first inverse frequency shift module is used to perform an inverse operation of the first frequency shift operation on the combined sub-band signal to obtain an output signal.

2. The filter according to claim 1, It is characterized in that The edge filtering module comprises: a zero gain unit, used for setting the gain of the other sub-band signals to zero, so as to completely filter out the other sub-band signals; A filtering unit, configured to perform asymmetric filtering on the sub-band signal where the transition band is located, so as to partially filter out the sub-band signal where the transition band is located, and obtain an output signal of the filtering unit, wherein the filtering unit has a delay; A delay unit, used for delaying the sub-band signal in the passband to obtain an output signal of the delay unit, wherein the output signal of the delay unit is aligned in time domain with the output signal of the filter unit; The filtered sub-band signal includes an output signal of the filtering unit and an output signal of the delay unit.

3. The filter according to claim 2, It is characterized in that The filtering unit comprises: A second frequency shift subunit, configured to perform a second frequency shift operation on the subband signal where the transition band is located; A low-pass filter, used for performing asymmetric filtering on the output signal of the second frequency shift subunit, wherein the coefficients of the low-pass filter are real numbers; The second inverse frequency shift subunit is used to perform an inverse operation of the second frequency shift operation on the output signal of the real filter.

4. The filter according to claim 2, It is characterized in that The filtering unit comprises a low-pass filter with complex coefficients, which is used to perform asymmetric filtering on the sub-band signal where the transition band is located, so as to partially filter out the sub-band signal where the transition band is located, and obtain an output signal of the filtering unit.

5. The filter according to any one of claims 1 to 4, It is characterized in that The first frequency shift module comprises: Numerically controlled oscillator for generating oscillation signals: The multiplier is used to multiply the input signal and the oscillation signal to perform a first frequency shift operation on the input signal.

6. The filter according to any one of claims 1 to 4, It is characterized in that The filter further comprises: A speed reduction module, wherein the sub-band signal output by the sub-band separation module is transmitted to the edge filtering module via the speed reduction module, and the speed reduction module is used to perform speed reduction processing on each sub-band signal.

7. The filter according to any one of claims 1 to 4, It is characterized in that The sub-band separation module is used to perform analysis filtering and fast discrete transformation processing on the frequency-shifted input signal to obtain each of the sub-band signals; The sub-band merging module is used to perform fast inverse discrete transform processing and comprehensive filtering on the filtered sub-band signals to merge the filtered sub-band signals.

8. An electronic device, It is characterized in that The electronic device comprises the filter according to any one of claims 1 to 7.

Images