Method for calculating compensation coefficients for an analog reception chain, device and corresponding program

Abstract

Method for calculating compensation coefficients of an analog reception chain, device and corresponding program The present invention relates to a method for calculating static compensation coefficients of an analog reception chain, a method implemented by an electronic device during a calibration phase of a reception device comprising the analog reception chain and a digital compensation module implementing an FIR filter whose coefficients are the calculated static compensation coefficients, the method comprising: - an estimation step (A01), using a predetermined calibration signal, of a frequency response of the analog reception chain; - a selection step (A02), using the frequency response, of a bandwidth over which the static compensation coefficients are calculated;a calculation step (A03) of the static compensation coefficients of the analog reception chain, in the time domain, as a function of the bandwidth and the inverse of the frequency response over this bandwidth. Figure for the abbreviation: 3;

Description

Title of the invention: Method for calculating compensation coefficients of an analog reception chain, device and corresponding program

[0001] 1. Scope of the invention

[0002] The invention relates to the reception of radio frequency signals. More particularly, the invention relates to the reception of radio frequency signals within a receiving device comprising an analog chain and a digital chain. One object of the invention is to compensate for distortions in the analog reception chain of a digital receiver comprising an analog reception chain and a digital reception chain. 1. Prior Art#

[0003] In many areas of communication, there is a need to meet the demand for increased communication speeds. This is the case, for example, in the civilian sector as well as in the military sector. This increase in speeds is necessitated by the ever-increasing use of communication devices to receive ever-larger amounts of data.

[0004] The increase in data rates must be achieved while guaranteeing constant reception quality. Thus, the continuous increase in military communication data rates requires the use of increasingly demanding waveforms in terms of managing radio distortions, particularly in the analog chain of the digital receiver.

[0005] It should be noted that in a digital receiver, the analog chain refers to the series of analog components and processes that manage the reception and conversion of the analog radio frequency (RF) signal into a digital format. This is the initial step during which the incoming RF signal is conditioned and prepared for further digital processing.

[0006] An analog chain often includes the following components: - an antenna which is responsible for capturing the RF signal from the waves and converting it into an electrical signal; - a low noise amplifier (LNA) which amplifies the weak electrical signal received from the antenna while introducing minimal noise. - a bandpass filter that eliminates unwanted signals and noise outside the desired frequency range; - a mixer that combines the filtered RF signal with a local oscillator signal to perform a downward frequency conversion: it produces an intermediate frequency (IF) signal, which is easier to process and digitize; - an intermediate frequency (IF) filter that eliminates unwanted signals and noise outside the desired frequency range, thus improving signal quality and isolating the specific frequency band of interest; - an automatic gain control (AGC) circuit that adjusts the gain of the received signal to maintain a constant signal strength; - an analog-to-digital converter (ADC) that converts the analog IF signal into a digital representation that can be processed by digital signal processing (DSP) components in the digital chain.

[0007] Once the analog chain has processed the signal and converted it into a digital format, the digital chain performs the necessary signal processing (DSP). The analog chain thus plays a crucial role in reception by ensuring the fidelity and quality of the received signal. It minimizes noise, filters out unwanted frequencies, and optimizes signal strength before digital processing takes place, ultimately contributing to the overall performance of the digital receiver. The digital chain processes the signal, for example, by performing PSK phase demodulation, 64QAM quadrature amplitude demodulation, or any other type of phase and amplitude demodulation.

[0008] To meet these requirements at the analog chain level, complex and expensive implementations are necessary. These implementations can be costly in terms of their execution: for example, to guarantee minimal distortion at the radio receiver chain level, a known method consists of sorting and selecting electronic components. This sorting is often done manually, by testing each electronic component individually to be used in manufacturing the radio receiver. The resulting additional cost is therefore significant.

[0009] These implementations can be technologically complex, for example at the level of gain configuration in the analog chain; at the level of filtering and equalization (use of static equalizers, which are mainly analog all-pass filters to compensate for phase but not amplitude and only in baseband); or in terms of physical implementation and design of the analog chain, requiring minimizing the signal path length and limiting sources of interference.

[0010] Document US10998926 proposes a solution for compensating the gain flattening of a receiving chain using a filter, the compensation being limited to the amplitude. This solution, however, requires the use of several signals and is complex to implement, since it requires working on a large number of points, which increases the computational cost on the one hand and the processing latency in operational conditions, which is prohibitive for processing that must take place in real time.

[0011] It is therefore necessary to propose a solution for compensating distortions in an analog reception chain that is simple and economical in terms of implementation, while also addressing the problem of increasing data rates to be processed, and that is adapted to real-time constraints. 1. Summary of the invention#

[0012] To this end, the invention relates to a method for calculating static compensation coefficients for an analog reception chain, a method implemented by an electronic device (2) during a calibration phase of a reception device (3) comprising the analog reception chain and a digital compensation module implementing an FIR filter whose coefficients are the calculated static compensation coefficients, the method comprising:

[0013] - an estimation step (A01), using a predetermined calibration signal, of a frequency response H of the analog reception chain;

[0014] - a selection step (A02), using the frequency response H, of a band BP passing on which the static compensation coefficients are calculated;

[0015] - a calculation step (A03) of the static compensation coefficients of the chain analog reception, in the time domain, as a function of the bandwidth BP and the inverse of the frequency response HBP over this bandwidth.

[0016] Depending on specific implementation methods, the process includes one or more of the following characteristics: - ; the estimation step (A01) of the frequency response H of the chain of Analog reception includes: • a step of determining a reference wobbled signal, in a frequency band of a predetermined width; • a step of obtaining a signal spectrum at the output of the analog reception chain, as a function of the reference wobbled signal; • a step of calculating the frequency response H of the analog reception chain as a function of the signal spectrum at the output of the analog reception chain and the reference wobbled signal spectrum; - The bandwidth selection step (A02) on which the static compensation coefficients are calculated includes a step of selecting, within the frequency response H, a portion of the bandwidth HBP corresponding to the portion of the frequency response H whose amplitude exceeds a predetermined amplitude value; - said amplitude value is predetermined by: • calculating the average value of the amplitude response of the frequency response H of the analog receiving chain; • subtracting, from this average value of the amplitude response of the frequency response H of the analog reception chain, a predefined tolerance value; - the calculation step (A03) of the static compensation coefficients of the analog reception chain as a function of the bandwidth BP and the inverse of the frequency response HBP on this bandwidth includes at least one iteration of a calculation of coefficients of a current finite impulse response (FIR) filter in the time domain as a function of the inverse Hi of the frequency response HBP on the bandwidth BP, said static compensation coefficients being constituted by the coefficients of the FIR filter; - Calculating the coefficients of a finite impulse response (FIR) filter includes: • a step of obtaining a number of points NFFTk current of an inverse Fourier transform, N FFTk being a power of 2 between a minimum value and a maximum value; • with the number of points N FFTk current of the inverse Fourier transform ifft, a step in solving the equation:

[0017] hik{t) NFFTk)

[0018] with:

[0019] hïk^t} being the impulse response;

[0020] being the inverse response of the transfer function in the bandwidth BP;

[0021] this calculation delivering NFFTk values ​​of FIR filter coefficients; • a digital filtering step of a predetermined signal using the FIR filter whose NFFn coefficients were obtained in the previous step; • a step to evaluate the quality of the FIR filter according to an EVM performance criterion; - the number of points N FFTk^i between 16 and 512.

[0022] The invention further relates to an electronic device (2) for calculating static compensation coefficients of an analog receiver chain, a device intended to be used during a calibration phase of a receiver device (3) comprising the analog receiver chain and a compensation module digital implementing an FIR filter whose coefficients are the calculated static compensation coefficients, the electronic device comprising: - means of estimating, using a predetermined calibration signal, a frequency response H of the analog reception chain; - means of selecting, using the frequency response H, a bandwidth BP on which the static compensation coefficients are calculated; - means of calculating the static compensation coefficients of the analog reception chain, in the time domain, as a function of the bandwidth BP and the inverse of the frequency response HBP on this bandwidth.

[0023] Furthermore, the invention relates to a computer program comprising software instructions which, when executed by a programmable electronic device, implement a method for calculating static compensation coefficients of an analog reception chain as defined above.

[0024] The invention finally relates to a receiving device (3) comprising an analog receiving chain and a digital distortion compensation module for the analog chain implementing an FIR filter, the digital compensation module being configured from coefficients calculated according to the method as defined above. 1. Brief Description of Figures#

[0025] The invention will become clearer upon reading the following description, given solely by way of non-limiting example, and made with reference to the drawings in which: - [Fig. 1] illustrates a system implemented to calibrate a receiver using a device for calculating the coefficients of a compensation filter; - [Fig.2] illustrates a device for calculating the coefficients of a filter compensation; - [Fig.3] explains the different steps implemented to calculate the correction filter coefficients; - [Fig.4] illustrates a wobbled signal of constant amplitude (between -1 and 1) and of frequency varies over time; - [Fig. 5] illustrates a comparison of the spectra of the input wobbled signal with the exit one; - [Fig. 6] illustrates an amplitude response of a wobbled output signal from the analog chain, response obtained from a frequency response H of the analog chain; - [Fig.7] illustrates a phase response of a wobbled output signal from the analog chain, response obtained from a frequency response H of the analog chain; - [Fig.8] illustrates the number of points needed (16) to correct a first analog chain in a first example of a situation; - [Fig.9] illustrates the number of points needed (256) to correct a second analog chain in a second example situation; - [Fig. 10] illustrates the symbol distribution for an output signal of an analog chain before correction by the method of the invention; and - [Fig. 11] illustrates the distribution of symbols for an output signal of an analog chain after correction by the method of the invention. 1. Description of an embodiment 1. Review of principles# #

[0026] As previously stated, one object of the invention is to perform digital compensation for distortions in an analog reception chain. This compensation, according to the invention, is simple to implement insofar as it does not entail any significant modification of the digital chain and is performed over a limited bandwidth. It therefore requires little computation and is compatible with real-time processing constraints.

[0027] Compensating for these distortions reduces implementation costs by relaxing the linearity constraints of the filters in the analog chain. The solution implemented essentially consists of inverting the amplitude response and / or the phase response (these responses being estimated during calibration) in the digital chain in order to correct the distortion of the overall analog chain and thus comply with a given template (for example, in terms of EVM - "error vector magnitude") and avoids resorting to more expensive methods such as component sorting.

[0028] According to the invention, digital compensation is achieved using a compensation filter defined by static compensation coefficients. More particularly, the compensation filter is a finite impulse response (FIR) filter. The static compensation coefficients of the FIR filter are calculated only once for a typical radio chain. These coefficients are calculated during an initial calibration phase of the analog chain. This calculation can be performed again during a subsequent recalibration of the chain (for example, during maintenance). Thus, the static compensation coefficients are fixed once and for all after calibration. The advantage is that it is not necessary to significantly modify the digital chain, which is the case when one wishes to achieve Dynamic compensation, in which the coefficients are automatically recalculated based on the received signal. The other advantage, as explained previously, is to limit not only the implementation cost but also the processing latency caused by this compensation in the receiver that implements the analog chain and the FIR filter.

[0029] According to the invention, these static compensation coefficients are calculated based on a frequency response of the entire receiving chain. The objective is to compensate for the overall distortions of the analog receiving chain, and not only those introduced by the analog filters, although the latter are primarily responsible for all the distortions.

[0030] More specifically, according to the invention, distortions are compensated only in the bandwidth defined at -2dB from the center frequency. Then, the inverse response to be applied in the time domain is calculated. Indeed, a correction directly in the frequency domain would be simpler but more expensive to implement and therefore prohibitive. The inventors have, instead, performed an estimation of an equivalent time-domain filter, which is easier to implement and therefore more economical for the digital chain.

[0031] Figure 1 illustrates a system that can be implemented to calibrate a receiver 3 using the calculation device 2 for the coefficients of a compensation filter. For example, the calculation device 2 is implemented during a calibration phase and it accommodates the analog RX chain of the receiver 3. Signals are sent to this analog RX chain, and transformed signals are analyzed by a network analyzer AR, in particular the output signals of the analog RX chain of the receiver 3 during calibration. The input and output signals of the analog RX chain are provided to a calculation component CC for the static coefficients of the compensation filter, which performs the calculation of these coefficients according to the method of the invention. When the coefficients of this filter are obtained, they are implemented in the filter of the receiver 3.

[0032] The receiver 3 includes the analog RX chain, the digital filter which performs the compensation, using the components which are implemented after calibration, so that the corrected signal can be processed by the rest of the baseband digital chain.

[0033] Figure 2 describes in more detail an electronic device for calculating static compensation coefficients for an analog receiver chain. This device can be implemented during the calibration phase of a receiver comprising the analog receiver chain and a digital compensation module.

[0034] In the example of [Fig. 2], the electronic device for calculating static compensation coefficients 2 comprises an electronic memory unit 10, at less a processing unit 70 and a communication interface 80 with remote devices, via a chosen communication protocol, for example a wired protocol and / or a radio communication protocol. The elements of device 2 are adapted to communicate via a communication bus 90.

[0035] In the example of [Fig.2], a frequency response estimation module 20, a bandwidth determination module 40 and a static compensation coefficient calculation module 60 are each implemented as a software program, or a software component, executable by the processor 70. The memory 10 of the electronic device for calculating static compensation coefficients 2 is then capable of storing a frequency response estimation software program 20, a bandwidth determination software program 40 and a static compensation coefficient calculation software program 60. The processor 70 is then capable of executing each of these software programs.

[0036] In an alternative not shown, the frequency response estimation module 20, the bandwidth determination module 40 and the static compensation coefficient calculation module 60 are each implemented as a programmable logic component, such as an FPGA (Field Programmable Gate Array), or an integrated circuit, such as an ASIC (Application-Specific Integrated Circuit).

[0037] When the electronic device for calculating static compensation coefficients 2 is implemented in the form of one or more software programs, i.e., in the form of a computer program, also called a computer program product, it is further capable of being stored on a computer-readable medium (not shown). The computer-readable medium is, for example, a medium capable of storing electronic instructions and being connected to a bus of a computer system. By way of example, the readable medium is an optical disc, a magneto-optical disc, ROM, RAM, any type of non-volatile memory (e.g., FLASH or NVRAM), or a magnetic card. A computer program comprising software instructions is then stored on the readable medium.

[0038] The method implemented for calculating these static compensation coefficients is described in relation to [Fig.3] and comprises the following steps: - estimation A01, using a predetermined calibration signal SCP, of a frequency response H of the analog receiving chain, delivering an amplitude response and / or a phase response; - selection A02, using the frequency response H, of a bandwidth BP on which the static compensation coefficients CSC will be calculated; - Calculation (iterative) A03 of the static compensation coefficients of the analog reception chain, in the time domain, as a function of the bandwidth BP and the inverse of the frequency response HBP in this bandwidth: this is an inverse filter calculation in the bandwidth, in the time domain allowing for example a minimization of EVMs, in an example of implementation.

[0039] Once these static compensation coefficients are established, they are saved in memory and made accessible to the FIR filter implemented in the digital chain of the receiver being calibrated. This FIR filter will then, under operational conditions, filter the signal it receives using the calculated coefficients.

[0040] The FIR filter, calculated iteratively in the step of calculating the static compensation coefficients, represents an asymmetrical, i.e., non-real response. Therefore, it is not possible to apply it directly in the baseband. The inventors thus found it advantageous to apply it either at an intermediate frequency before the baseband conversion or artificially at a low frequency sufficient to cover the filter's bandwidth. Let f be the artificial intermediate frequency such that f > Bw / 2, and Bw represents the filter's bandwidth.

[0041] In some implementation cases, however, it is not possible to operate on the artificial intermediate frequency f;. In this case, an alternative is to use a real (i.e. symmetric) filter which is less suitable and less precise but nevertheless allows for gains in terms of performance.

[0042] To obtain the time response, the assumed known frequency response is used. This overall frequency response is reduced to its bandwidth and then inverted. Its impulse response is then deduced via an inverse Fourier transform. This process requires a certain number of points, for example from 16 to 512, in order to be sufficiently precise.

[0043] It should be noted that one of the particularities of the method described above lies in its radio application, which aims to optimize the specific performance criterion of EVM. Other equalization techniques are more studied but are very focused on waveforms and propagation channels, and are therefore not suitable for the problem solved by the invention. 1. Example of implementation#

[0044] An example of the implementation of the method described above is then presented. 1. Estimation of the frequency response of the string#

[0045] First, as previously stated, device 2 performs an estimation of the frequency response of the analog chain. In order to obtain this overall frequency response, it is necessary to calibrate the radio equipment 3 to The FIR filter coefficients are calculated. This calibration is performed by electronic device 2 by injecting a so-called chirp waveform, that is, a signal whose frequency varies over time. This excites the different frequencies of interest in the band used by the analog chain of receiver 3. Electronic device 2 then recovers the signal at the output of the analog chain of receiver 3, unaffected by the (only) distortions.

[0046] According to the invention, the wobbled signal c(t) which is implemented is defined as follows:

[0047] _ Aexp (j2pi(f1 + ) xt)

[0048] With fi the starting frequency of the wobbled signal and A / the frequency band traversed. A signal of this type is exemplified in [Fig.4].

[0049] In this particular case, it is useful to know the bandwidth to be estimated in order to determine the correct bandwidth of the wobbled signal. For a chain with global bandwidth filtering B, we thus define A / = B and f _ with fc theo J ] J ^theo 2 the theoretical center frequency of the chain to be characterized. For example, the bandwidth to be estimated may be between 1.2 and 2 MHz, in a real case shown in relation to [Fig.4] and 5.

[0050] A comparison of the spectra of the input wobbled signal, denoted wobbled In, with that of the output, denoted wobbled Out, is exemplified in [Fig. 5]. In this example of [Fig. 5], the attenuations at the edge of the band, which represent the filtering effects of the analog reception chain, are noted.

[0051] The next operation consists of dividing these two spectra to obtain the frequency response H of the chain. This frequency response includes both the phase and the amplitude:

[0052] ( \

[0053] In which: - fft is the fast Fourier transform function defined such that:

[0054] = ^F^x{nyj^Nm. avcc k, n = 0... NFFT-1 and NFFT the number of FFT points used which is a power of 2. In calibration phase, there is no limit on the number of points which must be chosen to obtain the most faithful representation possible of the frequency response; - S- [«] — S- (nT^ represents the discrete signal obtained by sampling at the rate f =}lTe of the time signal emitted at the input of the analog reception chain; $ 11n 1 = rcPæscntc 'c discrete signal obtained by sampling at rate y — \ j Tg of the time signal received at the output of the

[0055]

[0056]

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063]

[0064]

[0065] analog reception chain. With this frequency response H, it is possible to obtain the amplitude response and the phase response. Figures 6 and 7 illustrate these two types of response (amplitude and phase) as examples. 1. Calculation of filtering coefficients# Secondly, the electronic device 2 calculates the static compensation coefficients used in the digital domain to compensate for distortions in the analog radio chain. These static compensation coefficients are determined from the frequency response H. To do this, we consider the following transfer function of a finite impulse response (FIR) filter, with a frequency response H: In which: - represents the delay of k samples; - bo, bi, b 2, ■■■> b M are the coefficients of the FIR filter; M is the order of the FIR filter. The order of the FIR filter indicates the number of coefficients N required for its implementation with N = M+l. According to the invention, the static compensation coefficients are the coefficients of an FIR filter, and it is these FIR filter coefficients that we seek to determine in this implementation. It is therefore necessary to calculate the various coefficients b0, bj, b2, and bM of such a filter. This calculation, however, is not carried out immediately from the previously obtained frequency response H, but from a portion HBP of it which must be determined during the selection step A02, this portion being limited to a so-called correction bandwidth, within which we seek to correct the distortions caused by the analog chain. 1. Determination of the correction bandwidth We select the bandwidth Bw of the filter defined at -2 dB of the amplitude value corresponding to a real center frequency denoted fc. Indeed, this real center frequency fc can be (and often is) different from the theoretical center frequency of the filter, denoted fctheo, precisely because of the dispersions of the analog filters that we wish to correct. To do this, we first consider the average value of the amplitude response in dB:

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074]

[0075]

[0076]

[0077]

[0078]

[0079]

[0080]

[0081]

[0082] As well as the overall amplitude response in dB: G,4 / ) = 201og, jHtd With ll being the absolute value operator. We then select the response and associated frequencies that satisfy a certain tolerance with respect to the value average : HBP(f) = H(fgp) With fBP = f(GrfB(f) > H-tolerance) The tolerance value is, for example, set at 2dB. The center frequency is then calculated as the average frequency of this defined frequency range, 2 dB below the average amplitude value: fc=g.^^ In which: represents the last frequency of the frequency range f bp; - ffijo] represents the first frequency of the frequency range f Bp; Thus, at the end of this calculation step, we have the correction bandwidth and the center frequency of this correction bandwidth. 1. Calculation of coefficients minimizing the EVM# HBP denotes the transfer function in the correction bandwidth. We then calculate the inverse response of the transfer function in the passband. From this, we can deduce the impulse response that can be applied in the time domain. As explained previously, the goal is to calculate the coefficients of a FIR filter (not an IIR filter) for the sake of simplicity and ease of implementation. The calculation of the coefficients of this FIR filter is implemented as follows: N fft ) With : - h^t) being the discrete impulse response in the time domain; - Hi(f) being the inverse response of the transfer function in the bandwidth BP; - ifft the inverse Fourier transform function defined such that: iff^n, N ff ) = W 1 *» avec k ' " = °' " Nrrr 1 e ' Nm k number of FFT points used which is a power of 2.N FFT the number of points to use to calculate the impulse response.

[0083] The number of points N FFT of the inverse Fourier transform directly influences The resolution estimates the response, as do their values. The number of FIR filter coefficients used is equal to the number of points N in the inverse Fourier transform (NFT). The values ​​of the FIR filter coefficients used are equal to the resolution values ​​of the inverse Fourier transform.

[0084] Since the compensation is static, it is possible to determine the number of points required during the calibration phase to meet a certain EVM criterion. In the case of SAW type analog filters, this number of points can, for example, vary from 16 to 512 depending on the EVM performance criterion.

[0085] In other words, depending on the performance criterion, it can be determined that it It will be necessary to calculate, for example, between 16 and 512 points of the inverse Fourier transform, or, for example, between 16 and 512 static compensation coefficients. According to the invention, the number of points of the inverse Fourier transform (i.e., the number of static compensation coefficients) is a power of 2. Furthermore, in order to avoid introducing a time delay that would be due to the use of a filter with too many coefficients (i.e., 2048 and above, for example), the number of points of the inverse Fourier transform is necessarily limited. Thus, determining the static compensation coefficients to be implemented in the filter for a typical analog radio receiver chain involves at least one iteration of the following coefficient calculation steps, consisting of: Obtain a number of points N FFTk of the current inverse Fourier transform (i.e., 16, 32, etc.), N FFTk being a power of 2 between a minimum value and a maximum value (e.g., between 16 and 512), such that N FFTk = 2k (k then being between 5 and 9); Given the number N FFTk points of the current inverse Fourier transform, perform the solution of:

[0086] = Npm)

[0087] this calculation delivers NFFTk values ​​of FIR filter coefficients; To digitally filter a predetermined signal, such as a modulated signal representative of the final application (e.g., PSK or QAM modulation), using the FIR filter whose N FFn coefficients have been obtained in the previous step; Evaluate the quality of the filter according to an EVM performance criterion.

[0088] The iterations of the coefficient calculation steps are repeated until the EVM performance criterion is met (for example, to achieve an EVM of less than 5%). When the performance criterion is met, the calculation stops: we then have both the number of coefficients NFFnasi and the value of these coefficients obtained by solving the inverse Fourier transform.

[0089] In this embodiment, the performance criterion is defined as an EVM criterion. Other performance criteria could, of course, be used depending on the requirements. This EVM performance criterion is defined, for a given constellation, with a given frame: 100901 ^=(4-4)^(^-¾) 2 100911 EVMt=lOoJ^ K y iayg - With P avg being the average power of the constellation considered; - Ik the phase measurement of the kth symbol of the frame; - ïk the expected value in phase of the kth symbol of the frame; - Qk the quadrature measure of the kth symbol of the frame; - the expected quadrature value of the kth symbol of the frame.

[0092] Thus, the number of points required (i.e., the number and value of the coefficients) is determined during calibration according to the analog filter(s) that need to be corrected (and therefore according to the radio channel to which this filter(s) relate). For example, with regard to Figures 8 and 9, for a first radio channel, to correctly correct distortions, the number of points may be 16, while for a second radio channel, the number of points may be 256. More specifically, [Fig. 8] illustrates that for a first radio channel, only 16 coefficients are needed to fall below a 5% EVM threshold. [Fig. 9] illustrates that for a second radio channel, distinct from the first, no fewer than 256 coefficients are needed to approach the 5% EVM threshold.The number of calculated coefficients is therefore variable and depends on the radio chain and its uniqueness, particularly the electronic components it comprises. Figures 10 and 11 illustrate the result obtained, in terms of EVM, using a FIR filter with coefficients calculated using the method described above. The advantage of this implementation is that it can accept a certain degree of variability in the radio chain, which is compensated for by the static filter, which has a finite length. 1. Application of the calculated filtering coefficients

[0093] Using the static filtering coefficients implemented in the FIR of the receiver's digital compensation module, a correction is applied to the signal received. This correction is carried out under operational conditions, that is to say during the use of the radio receiver, more particularly, during the use of the receiving device comprising the analog receiving chain and the digital distortion compensation module of the analog chain, the digital module being configured from coefficients calculated during calibration.

[0094] However, in some cases, in order to apply this baseband correction, it is necessary to transpose the impulse response of an intermediate frequency fi in order to be able to filter in the real domain:

[0095] hnco[n] = Zrjn] x exp (j2pi xf; xn) - is 'a discrete impulse response serving as band correction base ; - hncoM is the impulse response ht[ n ] transposed onto an intermediate frequency; - N represents the discrete samples in the time domain

[0096] This intermediate frequency fi is selected at the minimum necessary, i.e., at Qf, in order to avoid excessive oversampling. Indeed, to apply this correction, the sampling frequency (intermediate frequency fi) of the signal to which the correction is applied must be greater than or equal to the filter's bandwidth:

[0097] f. > l- i.

[0098] This impulse response h nco is then applied by convolution to the signal to be corrected, itself shifted by this frequency fi:

[0099] SncJ»]= Soujn]x exp(j2pi xf;xn)

[0100] ç Ll_ ç Ll ^compbcol11] ^411] unco[u]

[0101] with: * the convolution operator.

[0102] We then descend back to baseband:

[0103] she n / -n • r \ SCOmp[n] = x exp( -j2pt x 1;xn)

[0104] The filtered signal is then processed by the next module of the digital chain.

Claims

Demands

1. Method for calculating static compensation coefficients of an analog receiving chain, method implemented by an electronic device (2) during a calibration phase of a receiving device (3) comprising the analog receiving chain and a digital compensation module implementing an FIR filter whose coefficients are the calculated static compensation coefficients, the method comprising: - an estimation step (A01), using a predetermined calibration signal, of a frequency response H of the analog receiving chain; - a selection step (A02), using the frequency response H, of a bandwidth BP over which the static compensation coefficients are calculated;- a calculation step (A03) of the static compensation coefficients of the analog reception chain, in the time domain, as a function of the bandwidth BP and the inverse of the frequency response HBP on this bandwidth.;

2. Method for calculating static compensation coefficients of an analog receiving chain according to claim 1, characterized in that the estimation step (A01) of the frequency response H of the analog receiving chain comprises: - a step of determining a reference wobbled signal, in a frequency band of a predetermined width; - a step of obtaining a signal spectrum at the output of the analog receiving chain, as a function of the reference wobbled signal; - a step of calculating the frequency response H of the analog receiving chain as a function of the signal spectrum at the output of the analog receiving chain and a reference wobbled signal spectrum.

3. A method for calculating static compensation coefficients for an analog reception chain according to claim 1 or 2, characterized in that the bandwidth selection step (A02) BP on which the static compensation coefficients are calculated includes a selection step, within the frequency response H, of a portion HBP corresponding to the portion of the frequency response H whose amplitude exceeds a predetermined amplitude value.

4. Method for calculating static compensation coefficients of an analog receiving chain according to claim 3, characterized in that said amplitude value is predetermined by: - ​​calculating an average value of an amplitude response of the frequency response H of the analog receiving chain; - subtracting, from this average value of the amplitude response of the frequency response H of the analog receiving chain, a predefined tolerance value.

5. Method for calculating static compensation coefficients of an analog receiving chain according to any one of the preceding claims, characterized in that the calculation step (A03) of the static compensation coefficients of the analog receiving chain as a function of the bandwidth BP and the inverse of the frequency response HBp on this bandwidth comprises at least one iteration of a calculation of coefficients of a current finite impulse response (FIR) filter in the time domain as a function of the inverse Hbp of the frequency response Hbp on the bandwidth BP, said static compensation coefficients being constituted by the coefficients of the FIR filter.

6. A method for calculating static compensation coefficients for an analog receiver chain according to claim 5, characterized in that the calculation of the coefficients of a finite impulse response (FIR) filter comprises: - a step of obtaining a current number of points NFFTk of an inverse Fourier transform, NFFTk being a power of 2 between a minimum and a maximum value; - with the current number of points N FFTk of the inverse Fourier transform ifft, a step of solving the equation:

7.

8.

9. hikW ^IF Tk ) with : hik(t) being the impulse response; being the inverse response of the transfer function in the bandwidth BP; this calculation delivers NFFTk values ​​of FIR filter coefficients; - a digital filtering step of a predetermined signal using the FIR filter whose N FFn coefficients were obtained in the previous step; - a step to evaluate the quality of the FIR filter according to an EVM performance criterion. A method for calculating static compensation coefficients for an analog receiving chain according to claim 6, characterized in that the number of points N is between 16 and 512. An electronic device (2) for calculating static compensation coefficients for an analog receiving chain, a device intended for use during a calibration phase of a receiving device (3) comprising the analog receiving chain and a digital compensation module implementing an FIR filter whose coefficients are the calculated static compensation coefficients, the electronic device comprising: - means of estimating, using a predetermined calibration signal, a frequency response H of the analog reception chain; - means of selecting, using the frequency response H, a bandwidth BP on which the static compensation coefficients are calculated; - means of calculating the static compensation coefficients of the analog reception chain, in the time domain, as a function of the bandwidth BP and the inverse of the frequency response HBP on this bandwidth. Computer program comprising software instructions which, when executed by a programmable electronic device, implement a method for calculating static compensation coefficients of an analog receiving chain according to any one of claims 1 to 7.

Images