Method for forming configurable beams and associated devices
The frequency domain beam forming method addresses resource-intensive challenges in beam forming systems by using adapted FIR architectures for efficient fractional delay and equalization, achieving precise beam formation across multiple channels in broadband and multimode applications.
Patent Information
- Application Number
- FR2022001784
- Authority / Receiving Office
- FR · FR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2022-03-01
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2042-03-01
AI Technical Summary
Existing beam forming systems face challenges in achieving precise beam formation with fractional delays and equalization across multiple reception channels, especially in broadband and multimode applications, leading to resource-intensive FIR filters that are computationally prohibitive.
A method and system for forming configurable beams using a beam forming calculator that operates in the frequency domain, employing FIR architectures adapted to implement operations such as unpointing, decimation, and multiplexed delays and weightings, reducing computational and memory resources by leveraging spectral domain filtering.
This approach significantly reduces computational and memory resources while enabling precise beam formation with fractional delays and equalization across multiple channels, supporting broadband and multimode operations efficiently.
Smart Images

Figure 00000049_0000 
Figure 00000050_0000 
Figure 00000051_0000
Abstract
Description
Title of the invention: Method for forming configurable beams and associated devices
[0001] The present invention relates to a method for forming Q beams by a beamforming system comprising P reception channels. It also relates to a calculator and a system for forming Q beams, a computer program product and a computer-readable medium involved in implementing the forming method.
[0002] In a multi-antenna system, each antenna receives a set of signals from multiple listening directions with delays that depend both on the angle 0 of the listening direction and on the position of the antenna within the antenna system.
[0003] Ideally, beamforming in direction 6 consists of compensating for delay differences between antennas to enable coherent summation of the same signal between the different reception channels with the dual aim of gaining spatial selectivity and signal-to-noise ratio.
[0004] An example of such a beamforming system or beamformer is visible in [Fig.l]. The system comprises a set of antennas 1, each connected to a digital reception chain 2 and a beamforming unit 3 making it possible to generate N beams here.
[0005] Whether analog or digital, the realization of a beamformer capable of listening in different directions involves the realization of programmable delays, which, in practice, is very difficult to achieve.
[0006] Usually, it is considered that the signals to be listened to are narrowband around a carrier f0 assumed to be known. In this case, “simple” phase shifts ensure the formation of the listening beam; if rk is the delay to be compensated on the chain of index k, these phase shifts are worth 0k = 2irfork.
[0007] However, as soon as the band of the listening signal increases, this way of doing things is no longer possible and beam formation involving the realization of "true" delays (as opposed to phase shift, more often designated by the corresponding English term "true time delay") is then indicated.
[0008] Thus, the context of the present invention is that of so-called digital reception chains whose output data are the digitized values of periodic time samples of the signal at a rate noted fECH.
[0009] If the delay to be applied to the signal is a multiple of the sampling period T Ech = 1 / Ïech, the problem is simple since it is sufficient to digitally delay of an integer number of samples. On the other hand, if the delay to be achieved, counted in number of sampling periods, includes a decimal part, or even simply a fractional part, the problem is more complicated since it implicitly corresponds to an interpolation of the signal.Indeed, on the one hand, the interpolation filter depends on the delay to be ensured (a different filter per delay) and, on the other hand, for a given precision, the duration of the temporal response of this filter is all the greater as the band in which we want to ensure this delay with precision is closer to the Nyquist band associated with the sampling (typically inversely proportional to the difference between the Nyquist band and the useful band); it is known that this is linked to the discontinuity at the limits of the sampling band of the 0¾¾ response associated with a delay r since r is not a multiple of the sampling period l / fECH-
[0010] Furthermore, in any beamforming, the quality of the realization degrades. since the reception channels do not have exactly the same response. It is known that the differences in response, both in spectral amplitude and in spectral phase, must be corrected prior to good beam formation; the correction data to be made on the amplitude and spectral phase of each of the channels are acquired by calibration and are commonly available in the spectral domain. This correction therefore constitutes a filtering specific to each channel.
[0011] However, as for the delay, the complex gain to which this correction corresponds is generally neither continuous nor continuously derivable at the limits of the sampling band so that, here again, the duration of the time response of this correction filter is all the greater as the band in which one wishes to ensure this correction with precision is closer to the Nyquist band associated with the sampling.
[0012] Furthermore, the band in which channel formation is desired often depends on the mode of the intended application. Thus, in a multimode system, the sampling frequency of the processed signals must be, in the sense of the Shannon criterion, compatible with the largest of the bands covered. In this case, it is not desirable, for narrowband modes, to maintain the full input rate; decimation must then be carried out and for this purpose involves filtering the useful band, this useful band possibly being centered at different locations in the sampling band. Still in this case, it may be desirable to simultaneously form several beams covering the same useful band.
[0013] It may also be desirable, in other configurations, to simultaneously form a large number of narrow beams and a smaller number of wider beams. The ideal is then to be able to carry, by multiplexing, indifferently and simultaneously on a set of output channels clocked at fECn, sets of beams of the same band, the nth channel carrying Kn mul- beams fECH / Kn band diplexes (K can vary from one channel to another).
[0014] It is therefore desirable for a wideband, multimode beamformer to meet several distinct objectives.
[0015] It is appropriate to form a set of beams of given and potentially different useful bands; this involves carrying out 3 different filtering operations, the equalization of the reception channels specific in essence to each reception channel, the delay specific to each beam / reception channel pair and the selection of the useful band, specific to each beam formed. As explained above, these filtering operations can correspond to very long time responses which are carried out, in an approximate manner in practice, by transversal type filters with a priori finite impulse response which are all distinct. Such filters are often designated by the acronym FIR referring to the English terminology "Finite Impulse Response" which means "finite impulse response". The FIR filters according to the state of the art implement operations including the use of time shifts of the signal, gains and summations.The number of operations is equal to the length of the impulse response of the FIR filter considered (the length being expressed in number of samples). Thus, such a large number of computing resources are involved that they can very quickly become prohibitive.
[0016] One objective is therefore to form groups of individually configurable beams with reduced resources involved.
[0017] For this purpose, the description describes a method for forming Q multiplexed beams on Q' groups of beams indexed by q', q' being an integer varying between 1 and Q', Q' being an integer greater than 1, and the multiplexing rate being equal to , Kq- being a respective integer for each value of q', the method being implemented by a computer of a beam forming system comprising P reception channels each suitable for receiving signals coming from a respective antenna, each antenna having a respective aperture, Q and P being two integers greater than or equal to 2, the method comprising:
[0018] - the passage in the spectral domain of signals originating from signals originating from each antenna,
[0019] - the implementation of a set of operations in the spectral domain, called depointing operations, the set of depointing operations comprising:
[0020] - for each reception channel, a generation operation by duplication of a contribution to each of the Q' beam groups to be generated,
[0021] - for each contribution, a decimation operation by , and a maintenance on 2ÀV periods to constitute a group of contributions multiplexed by 2ÂA
[0022] - the implementation for each group of contributions of an operation multiplexed introduction by 2^' of a set of delays, and of a possible set of 2^ operations of weighting of the opening of the antenna, and
[0023] - an operation of summing the contributions to a beam to be generated, resulting from all reception channels, multiplexed for each channel on Q' groups of contributions, the summation operation being carried out for each beam to be generated.
[0024] Today, beam formation is carried out with conventional FIR architectures operating in the time domain, whereas here we use an FIR architecture working in the frequency domain, which was the subject of applications FR 3049131, FR 3102264, FR 3102262 and FR 3102265, adapting it in a clever manner to obtain a substantial gain in resources and individually configurable beam groups.
[0025] On reading what follows, the reader will understand that, in the case of beamforming with FIRs working in the frequency domain, two distinct gains are obtained: a gain on the calculation operators (multiplier in particular) and a gain on the memory resources through pooling.
[0026] According to particular embodiments, the method for forming Q multiplexed beams comprises one or more of the following characteristics, taken in isolation or in all technically possible combinations:
[0027] - the operations implemented in the spectral domain further comprise a channel equalization operation, the equalization operation being implemented before the depointing operations, these operations preferably being carried out in a multiplicative manner.
[0028] - the operations implemented in the spectral domain further comprise a band limiting operation, the band limiting operation being implemented after the depointing operations, these operations preferably being performed multiplicatively.
[0029] - the method comprises, for each reception channel, the formation of two reception channels calculation, the input signal of the first calculation channel being obtained by implementing the shift of the signal received by the antenna by M points, M being an integer greater than or equal to 2 to obtain a shifted signal and calculating the sum of the received signal and the shifted signal, and the input signal of the second calculation channel being obtained by implementing the shift of the received signal by M points to obtain a shifted signal and calculating the difference between the received signal and the shifted signal.
[0030] - the transition to the spectral domain is carried out by applying a transform of Discrete Fourier at M points on each of the calculation channels.
[0031] - the method comprises a step of returning to the time domain, the return step in the time domain involving the application of 2^ Fourier transforms discrete inverse points multiplexed by 2^' on each of the calculation channels, the method comprising a spectral shift operation of the second calculation channel by multiplicative application of a signal e[+j floor(n / 2«^jr / ( M / 7^')], n being an integer ranging from 0 to Ml and the floor function designating the integer part function, and an operation of summing the two outputs of the two calculation channels.
[0032] The description also describes a computer, in particular a programmable logic circuit, adapted to implement the operations of a training method as previously described.
[0033] The description also proposes a system for forming Q multiplexed beams on Q' groups of beams indexed by q', q' being an integer varying between 1 and Q', Q' being an integer greater than 2, Q' being an integer greater than 1, and the multiplexing rate being equal to 9^', Kq being equal to a respective integer for each value of q', the forming system comprising P reception channels each suitable for receiving signals coming from a respective antenna, Q and P being two integers greater than or equal to 2, and the forming system comprising a computer as previously described.
[0034] The description also describes a computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the operations of a method or a method as previously described.
[0035] The description also provides a computer-readable medium comprising instructions which, when executed by a computer, cause the computer to implement the operations of a method or a method as previously described.
[0036] Characteristics and advantages of the invention will appear on reading the description which follows, given solely by way of non-limiting example, and made with reference to the appended drawings, in which:
[0037] - [Fig.l] [Fig.l] is a schematic representation of a system for training beams,
[0038] - [Fig.2] [Fig.2] is a schematic representation of the resources of an ar long time response filter architecture controlled in the spectrum,
[0039] - [Fig.3] [Fig.3] is a schematic representation of an example of architecture modified from the architecture shown in [Fig.2] in which certain resources are shared,
[0040] - [Fig.4] [Fig.4] is a schematic representation of another example of modified architecture compared to the architecture represented in [Fig.2] in which certain resources are shared,
[0041] - [Fig.5] [Fig.5] is a set of two graphs showing the time response of a Hilbert filter (real part on top and imaginary part on bottom),
[0042] - [Fig.6] [Fig.6] is a graph showing the frequency response of a filter of Hilbert (real part on top and imaginary part on bottom),
[0043] - [Fig.7] [Fig.7] is a set of two graphs showing the elimination effect total of the second Nyquist band in the spectrum of a Hilbert filter,
[0044] - [Fig.8] [Fig.8] is a set of two graphs allowing to compare the filter of Hilbert of [Fig.6] (left graph) with the filter required when the input signal is complex to avoid problems related to the discontinuity of the frequency response of the delay at the limits of the sampling band (right graph),
[0045] - [Fig.9][Fig.lO] Figures 9 and 10 are schematic representations of a example of a beamforming system comprising parts A to H, ([Fig.9] corresponding to the left part and [Fig. 10] to the right part),
[0046] - [Fig. 11] [Fig. 11] is a schematic representation of parts A to D of the beamforming system of Figures 9 and 10,
[0047] - [Fig. 12] [Fig. 12] is a schematic representation of parts E to H of the system beamforming of figures 9 and 10,
[0048] - [Fig. 13] [Fig. 13] is a schematic representation of the decomposition of a M-point IFFT into a 2-point DFT and an M / 2-point IFFT,
[0049] - [Fig. 14] [Fig. 14] is a schematic representation of an example of a system of beamforming with multiplexing of 2 beams decimated by 2,
[0050] - [Fig. 15] [Fig. 15] shows the modification of the IFFT to 2N points in view of the mul triplexing of 2K beams decimated by 2K,
[0051] - [Fig. 16] [Fig. 16] is a schematic representation of an example of a system of beamforming with multiplexing of 2K beams decimated by 2K,
[0052] - [Fig. 17] [Fig. 17] is a schematic representation of a resetting device frequency of the decimated band,
[0053] - [Fig. 18] [Fig. 18] is a schematic representation of another example of beamforming system with multiplexing of 2K beams decimated by 2K,
[0054] - [Fig. 19] [Fig. 19] is a block diagram of beamforming,
[0055] - [Fig.20][Fig.21] Figures 20 and 21 are a schematic representation of another example of a beamforming system with multiplexing of 2K beams decimated by 2K ([Fig.20] corresponding to the left part and [Fig.21] to the right part),
[0056] - [Fig.22] [Fig.22] is a schematic representation of an example of a device calibration,
[0057] - [Fig.23] [Fig.23] is a schematic representation of an example of a device of time resetting,
[0058] - [Fig.24] [Fig.24] shows the modification of the 2N-point FFT for the multiplexed processing of 2K beams decimated by 2K,
[0059] - [Fig.25] [Fig.25] schematically represents a system for training beams in transmission with multiplexing of 2K beams decimated by 2K, and
[0060] - [Fig.26] [Fig.26] represents a detail of a part of the system of [Fig.24], and
[0061] - [Fig.27] [Fig.27] represents a detail of another part of the system of [Fig.24]
[0062] To facilitate the further reading of the description, it is separated into chapters and sections. However, the chapters and sections are not independent and the concepts used in one chapter are also valid for the others.
[0063] In order, the following will be discussed:
[0064] 1 - SOME PRELIMINARY NOTIONS
[0065] 1.1- PRODUCTION OF LONG TIME RESPONSE FILTERS
[0066] 1.2-REALIZATION OF A DELAY
[0067] 1.3- IQ DEMODULATION - HILBERT FILTERING
[0068] 2 - DESCRIPTION OF EXAMPLES OF TRAINING METHODS
[0069] 2.1 - GENERAL PRINCIPLE OF THE TRAINING PROCESS
[0070] 2.2 - DESCRIPTION OF AN EXAMPLE OF A PROCESS FOR FORMING BEAMS
[0071] 2.2.1 - Case of figures 9 and 10
[0072] 2.2.2 - Generalization
[0073] 2.3 - DESCRIPTION OF AN EXAMPLE OF A PROCESS FOR FORMING BEAMS
[0074] 2.3.1 - Decimation by 2 (for example, passage from real to I / Q)
[0075] 2.3.2 - Case of bands not between 0 and -1 + 2X K
[0076] 2.3.3 - Decomposition into successive partial beam formations
[0077] 2.3.4 - General case
[0078] 3 - DESCRIPTION OF A CALIBRATION METHOD
[0079] 4 - TRANSPOSITION OF TRAINING METHODS TO BROADCASTING
[0080] In particular, there will be mentioned, with particular reference to paragraph 2.2, a method for forming Q beams by a computer of a beam forming system comprising P reception channels each suitable for receiving signals coming from a respective antenna, each antenna having a respective aperture, Q being an integer greater than or equal to 1 and P being an integer greater than or equal to 2, the method comprising:
[0081] - the passage in the spectral domain of signals originating from signals originating from each antenna,
[0082] - implementing operations in the spectral domain, the operations comprising de-clocking operations, de-clocking operations including:
[0083] - for each reception channel, a generation operation by duplication of a contribution to each of the Q beams to be generated,
[0084] - the implementation for each contribution of an operation of introduction of a delay and a possible antenna aperture weighting operation, and
[0085] - an operation of summing the contributions to a beam to be generated from all reception channels, the summation operation being carried out for each beam to be generated.
[0086] According to particular aspects, the method for forming Q beams comprises one or more of the following characteristics, taken in isolation or in all technically possible combinations:
[0087] - the operations implemented in the spectral domain further comprise a channel equalization operation, the equalization operation being implemented before the depointing operations, these operations preferably being carried out in a multiplicative manner.
[0088] - the operations implemented in the spectral domain further comprise a band limiting operation, the band limiting operation being implemented after the depointing operations, these operations preferably being performed multiplicatively.
[0089] - the method comprises, for each reception channel, the formation of two reception channels calculation, the input signal of the first calculation channel being obtained by implementing the shift of the signal received by the antenna by M points, M being an integer greater than or equal to 2 to obtain a shifted signal and the calculation of the sum of the received signal and the shifted signal, and the input signal of the second calculation channel being obtained by implementing the shift of the received signal by M points to obtain a shifted signal and the calculation of the difference between the received signal and the shifted signal, this difference being translated in frequency by multiplicative application of a signal e JnPi / M , n being an integer ranging from 0 to Ml.
[0090] - the transition to the spectral domain is carried out by applying a transform of Discrete Fourier at M points on each of the calculation channels.
[0091] - the method comprises a step of returning to the time domain, the return step in the time domain comprising the application of an inverse discrete Fourier transform at M points on each of the calculation channels, the method comprising a spectral shift operation of the second calculation channel by multiplicative application of a signal e+JnPi / M and an operation of summing the two outputs of the two calculation channels.
[0092] The description also describes: - a computer, in particular a programmable logic circuit, adapted to implement the operations of a training process such as previously describe. - a Q beamforming system comprising P reception channels each capable of receiving signals from a respective antenna, Q being an integer greater than or equal to 1 and P being an integer greater than or equal to 2, the forming system comprising a computer as previously described. - a computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the operations of a method or a method as previously described. - a computer-readable medium comprising instructions which, when executed by a computer, cause the computer to implement the operations of a method or a method as previously described.
[0093] There will also be mentioned, with particular reference to paragraph 2.3, a method for forming Q multiplexed beams on Q' groups of beams indexed by q', q' being an integer varying between 1 and Q', Q' being an integer greater than 1, and the multiplexing rate equal to Kq' being equal to a respective integer for each value of q', the method being implemented by a computer of a beam forming system comprising P reception channels each suitable for receiving signals coming from a respective antenna, each antenna having a respective aperture, Q and P being two integers greater than or equal to 2, the method comprising:
[0094] - the passage in the spectral domain of signals originating from signals originating from each antenna,
[0095] - the implementation of a set of operations in the spectral domain, called depointing operations, the set of depointing operations comprising:
[0096] - for each reception channel, a generation operation by duplication of a contribution to each of the Q' beam groups to be generated,
[0097] - for each contribution, a decimation operation by ^<ï, and a maintenance on 9^' periods to constitute a group of contributions multiplexed by 9^,
[0098] - the implementation for each group of contributions of an operation multiplexed introduction by 9^' of a set of 9^' delays, and of a possible set of 2^' antenna aperture weighting operations, and
[0099] - an operation of summing the contributions to a beam to be generated, resulting from all reception channels, multiplexed for each channel on Q' groups of contributions, the summation operation being carried out for each beam to be generated.
[0100] According to particular aspects, the method for forming multiplexed Q beams comprises one or more of the following features, taken in isolation or according to all technically possible combinations:
[0101] - the operations implemented in the spectral domain further comprise a channel equalization operation, the equalization operation being implemented before the depointing operations, these operations preferably being carried out in a multiplicative manner.
[0102] - the operations implemented in the spectral domain further comprise a band limiting operation, the band limiting operation being implemented after the depointing operations, these operations preferably being performed multiplicatively.
[0103] - the method comprises, for each reception channel, the formation of two reception channels calculation, the input signal of the first calculation channel being obtained by implementing the shift of the signal received by the antenna by M points, M being an integer greater than or equal to 2 to obtain a shifted signal and the calculation of the sum of the received signal and the shifted signal, and the input signal of the second calculation channel being obtained by implementing the shift of the received signal by M points to obtain a shifted signal and the calculation of the difference between the received signal and the shifted signal.
[0104] - the transition to the spectral domain is carried out by applying a transform of Discrete Fourier at M points on each of the calculation channels.
[0105] - the method comprises a step of returning to the time domain, the step of returning in the time domain comprising the application of inverse discrete Fourier transforms at M / p^ points multiplexed by qK'ï on each of the calculation channels, the method comprising a spectral shift operation of the second calculation channel by multiplicative application of a signal e[+j floodn / p^p^ M / 2^)], n being an integer ranging from 0 to Ml and the floor function designating the integer part function, and an operation of summing the two outputs of the two calculation channels.
[0106] The description also describes: - a computer, in particular a programmable logic circuit, adapted to implement the operations of a training method as previously described. - a system for forming Q beams multiplexed on Q' groups of beams indexed by q', q' being an integer varying between 1 and Q', Q' being an integer greater than 2, Q' being an integer greater than 1, and the multiplexing rate being equal to , Kg' being equal to a respective integer for each value of q', the forming system comprising P reception channels each suitable for receiving signals coming from a respective antenna, Q and P being two integers greater than or equal to 2, and the forming system comprising a computer as previously described. - a computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the operations of a method or a method as previously described. - a computer-readable medium comprising instructions which, when executed by a computer, cause the computer to implement the operations of a method or a method as previously described.
[0107] There will also be mentioned, with particular reference to paragraph 3, a method for calibrating a beamforming system comprising P reception channels each capable of receiving signals originating from a respective antenna, Q being an integer greater than or equal to 1 and P being an integer greater than or equal to 2, the formation system comprising a computer capable of implementing the following steps:
[0108] - the passage using spectral analysis in the spectral domain of signals coming from each antenna,
[0109] - the implementation of operations in the spectral domain,
[0110] the calibration method comprising:
[0111] - for each reception channel, the injection in the time domain of a replica of a digital signal used to constitute a calibration signal, this replica being synchronized with the spectral analysis means applied to the signals of the reception channels, the injection being carried out in an identical manner for each reception channel,
[0112] - for each reception channel, the passage in the spectral domain of this signal digital by means identical to those applied to the signals of the reception channels, said identical means being synchronized with each other, to constitute an identical reference signal on each reception channel, the passage being carried out in an identical manner for each reception channel,
[0113] - the measurement, for each reception channel, of values of the reference signal and of values of the signal propagated on the reception channel, and
[0114] - the deduction of the difference between the reference signal and the signal propagated on the channel of reception based on measured values.
[0115] According to particular aspects, the calibration method comprises one or more of the following characteristics, taken in isolation or in all technically possible combinations:
[0116] - the method further comprising the following steps:
[0117] - the generation of a digital calibration signal
[0118] - conversion of the calibration signal into analog,
[0119] - the frequency transposition of the converted calibration signal,
[0120] - the amplification of the transposed calibration signal, and
[0121] - balanced distribution to the different reception channels.
[0122] - when several reception channels are housed on the same electronic card, the constitution of the reference signal is shared in common with all the channels of the same card (a single generation of reference signal per card).
[0123] - the measured values include the power correlation between the spectra of the signal propagated on a reception channel and the reference signal.
[0124] - the measured values include the power spectral density of the signal of reference and / or the power density of the signal propagated on the reception path.
[0125] - the measured values are obtained by integration over several successive frames.
[0126] - the deduction step includes the estimation of the phase of the correlation of power between the spectrum of the signal on a reception channel and the reference signal, to obtain an estimated phase, the phase of the difference between a reception channel and the reference signal being the estimated phase.
[0127] - the deduction step includes the estimation of the amplitude difference between a channel of reception and the reference signal by calculating the square root of the ratio between the power spectral densities of the signal on a reception channel and the reference signal.
[0128] - the deduction step includes the estimation of the amplitude difference between a channel of reception and the reference signal by calculating the ratio between the power spectral density of the reference signal and the power correlation between the spectrum of the signal on said reception channel and the reference signal. This operation is done at low cost.
[0129] - the reference signal is noise.
[0130] The description also describes:
[0131] - a device for calibrating a Q beamforming system comprising P reception channels each suitable for receiving signals from a respective antenna, Q being an integer greater than or equal to 1 and P being an integer greater than or equal to 2, the calibration device being suitable for implementing a calibration method as previously described.
[0132] - a computer program product comprising instructions which, when the program is executed by a computer, causing it to implement a calibration process as previously described.
[0133] - a computer-readable medium comprising instructions which, when executed by a computer, lead it to implement a calibration process as previously described.
[0134] There will also be mentioned, with particular reference to paragraph 4 and transposed to paragraph 2.2, a method for forming P multi-beam emission channels from Q beams, the method being implemented by a computer of a transmission beamforming system comprising P transmission channels each provided with an antenna, Q being an integer greater than or equal to 1 and P being an integer greater than or equal to 2, the method comprising:
[0135] - the passage into the spectral domain of signals from the beams,
[0136] - implementing operations in the spectral domain, the operations comprising pointing operations, pointing operations including:
[0137] - a generation operation for each beam of a contribution for each P transmission channels by implementation for each contribution of an operation of introducing a delay and a possible weighting of the opening of the antenna, and
[0138] - an operation of summing the contributions of a corresponding transmission channel to each beam, the summation operation being carried out for each emission channel to be generated.
[0139] According to particular aspects, the method for forming P emission paths comprises one or more of the following characteristics, taken in isolation or in all technically possible combinations:
[0140] - the operations implemented in the spectral domain further comprise a channel equalization operation, the equalization operation being implemented after the pointing operations, these operations being implemented multiplicatively.
[0141] - the operations implemented in the spectral domain further comprise a band-limiting operation, the band-limiting operation being implemented before the pointing operations, these operations being implemented multiplicatively.
[0142] - the method comprises, for each beam, the formation of two calculation channels,
[0143] the input signal of the first calculation channel being obtained by implementing the shift of the beam signal of M points, M being an integer greater than or equal to 2 to obtain a shifted signal and the calculation of the sum of the received signal and the shifted signal, and
[0144] the input signal of the second calculation channel being obtained by implementing the shift of the signal of the beam of M points to obtain a shifted signal and the calculation of the difference between the received signal and the shifted signal, this difference being spectrally shifted by multiplicative application of a signal e jnPi / M, n being an integer ranging from 0 to M-1.
[0145] - the transition to the spectral domain is carried out by applying a transform of Discrete Fourier at M points on each of the calculation channels.
[0146] - the method comprises a step of returning to the time domain, the return step in the time domain involving the application of a Fourier transform discrete inverse at M points on each of the calculation channels, the method comprising a spectral shift operation of the second calculation channel by multiplicative application of a signal e+JnPi / M and an operation of summing the two outputs of the two calculation channels.
[0147] The description also describes: - a computer, in particular a programmable logic circuit, adapted to implement the operations of a training method as previously described. - a system for forming P multi-beam transmission channels from Q beams, the forming system comprising P transmission channels each equipped with an antenna, Q being an integer greater than or equal to 1 and P being an integer greater than or equal to 2, the forming system comprising a computer as previously described. - a computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the operations of a method for forming P multi-beam emission channels from Q beams or a method for forming P multi-beam emission channels from Q beams as previously described. - a computer-readable medium comprising instructions which, when executed by a computer, cause the computer to implement the operations of a method for forming P multi-beam emission paths from Q beams or a method for forming P multi-beam emission paths from Q beams as previously described
[0148] There will also be mentioned, with particular reference to paragraph 4 and transposed to paragraph 2.3, a method for forming P multi-beam transmission channels from Q beams multiplexed on Q' groups of beams indexed by q', q' being an integer varying between 1 and Q', Q' being an integer greater than 1, and the multiplexing rate being equal to 2^, being equal to a respective integer for each value of q', the method being implemented by a computer of a transmission beamforming system comprising P transmission channels each provided with an antenna having a respective aperture, Q and P being two integers greater than or equal to 2, the method comprising:
[0149] - for each group q' of beams, the passage in the spectral domain of the signals composing the group of multiplexed beams, to obtain multiplexed spectra,
[0150] - implementing operations in the spectral domain, the operations comprising pointing and interpolation operations including:
[0151] - an operation of forming as many replicas of the spectra as there are channels of emission to form,
[0152] - for each replica corresponding to each of the P transmission channels, a operation of generating a contribution to the corresponding transmission channel by implementing, for each contribution, an introduction operation multiplexed by 2^' of a set of delays, and a possible set of weightings of the antenna opening,
[0153] - an integration operation on points,
[0154] - an interpolation operation by inserting 2^' _ ] zeros, and
[0155] - an operation of summing the contributions to a corresponding transmission channel to each group of beams, the summation operation being carried out for each transmission channel to be generated.
[0156] According to particular aspects, the method for forming P emission paths comprises one or more of the following characteristics, taken in isolation or in all technically possible combinations:
[0157] - the operations implemented in the spectral domain further comprise a channel equalization operation, the equalization operation being implemented after the pointing operations, these operations being implemented multiplicatively.
[0158] - the operations implemented in the spectral domain further comprise a band-limiting operation, the band-limiting operation being implemented before the pointing operations, these operations being implemented multiplicatively.
[0159] - the method comprises, for each group of beams, the formation of two paths calculation, the input signal of the first calculation channel being obtained by implementing the shift of the beam signal by M points, M being an integer greater than or equal to 2x2^' to obtain a shifted signal and calculating the sum of the beam group signal and the shifted signal, and the input signal of the second calculation channel being obtained by implementing the shift of the beam group signal by M points to obtain a shifted signal and calculating the difference between the beam group signal and the shifted signal, this difference being translated into frequency by multiplicative application of a signal e[+j noodn / p^'^ M / 9^)], n being an integer ranging from 0 to M-1 and the floor function designating the integer part function, M being an integer greater than or equal to
[0160] - the transition to the spectral domain is carried out on each of the calculation channels by application of a discrete Fourier transform to M^' / points multiplexed by fa'.
[0161] - the method comprises a step of returning in the time domain, the step of returning in the time domain comprising the application of an inverse discrete Fourier transform at M points on each of the calculation channels, the method comprising a spectral shift operation of the second calculation channel by multiplicative application of a signal e+jnPi / M and an operation of summing the two outputs of the two calculation channels.
[0162] The description also describes: - a computer, in particular a programmable logic circuit, adapted to implement the operations of a training method as previously described. - a system for forming P multi-beam transmission channels from Q beams multiplexed on Q' groups of sub-beams indexed by q', q' being an integer varying between 1 and Q', Q' being an integer greater than 2, transmission beam forming system comprising P transmission channels each equipped with an antenna having a respective aperture, Q and P being two integers greater than or equal to 2, the forming system also comprising a computer as previously described. - a computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the operations of a method for forming P multi-beam transmission channels from Q multiplexed beams or a method for forming P multi-beam transmission channels from Q multiplexed beams, the method being as previously described. - a computer-readable medium comprising instructions which, when executed by a computer, cause the computer to implement the operations of a method for forming P multi-beam transmission channels from Q multiplexed beams or a method for forming P multi-beam transmission channels from Q multiplexed beams, the method being as previously described. 1 - SOME PRELIMINARY NOTIONS
[0163] These preliminary notions are part of the technological background.
[0164] 1.1 - PRODUCTION OF LONG TIME RESPONSE FILTERS
[0165] When the lengths of the time responses of an FIR type filter are very large, carrying out the filtering becomes problematic or even impossible given the very large number of resources that this implies, especially if the FIR must be programmable. To get around this problem, one solution is to carry out this operation in the spectrum. To do this, we move from the time domain to that of frequencies by Fourier transformation, the filtering operation then becoming multi plicative, then we return to the time domain by inverse Fourier transformation.
[0166] This makes it possible to define a generic filter architecture which has been the subject of a patent application published under the number FR 3 049 131 A1, the content of which is briefly summarized below, with further information being contained in the description of this application. In accordance with [Fig.2] which refers to a filter architecture with long time response controlled in the spectrum, this architecture makes it possible to produce, using FFTs of durations N, an FIR of impulse response of duration N entirely contained in the domain of positive times.
[0167] As a reminder, a Fourier transform is implemented by a fast Fourier transform, often referred to by the acronym FFT for "Fast Fourier Transform". More precisely, the transition from the time domain to the frequency domain is obtained by using an FFT, while the transition from the frequency domain to the time domain is obtained by using an IFFT. The acronym IFFT refers to "Inverse Fast Fourier Transform".
[0168] In the FIR of [Fig.2], the pk coefficients represent the spectral density obtained by a 2N-point FFT of the FIR response, with negative times representing N zeros on the 2N values of the time frame:
[0169] Note that, if the time response associated with the filtering is entirely contained in the negative times (as happens when filtering adapted to a signal by the conjugate spectrum), then the output recombination between the two channels (even and odd frequencies) is carried out by a subtraction instead of the addition indicated in [Fig.2].
[0170] It can be seen that, in this architecture, the block 1 + zx (moving from the position indicated 2 to the position indicated 3 in [Fig.2]) can be translated at will to any of the positions 1 to 5. It can also be translated to one or other of the positions a or b but on condition of being replaced by a block 1 - z \ This makes it possible to define several equivalent architectures including, in particular, the two which are proposed respectively in [Fig.3] and in [Fig.4] and which make it possible to share the storage resources z \
[0171] These FIR architectures are particularly interesting because they make it possible to considerably reduce the computational resources required compared to a classic FIR as soon as the number of coefficients increases, typically in a ratio M to log(M); this is however paid for by a certain increase in memory resources (factor 4 for the architectures of both figures 3 and 4 and 5 for the architecture of [Fig.2]).
[0172] The computing resources required for a response of maximum duration equal to N = 2M are:
[0173] • additions: 3 + 4x2xM = 8M + 3
[0174] • multiplications: 4 + 4xM / 2 = 2M + 4
[0175] Elements on Radix decomposition
[0176] To understand the rest of the description, it is appropriate to explain general points on the decomposition of a discrete Fourier transform (DFT) in Radix (x being the number of factorizable points).
[0177] The N-point discrete Fourier transform of a sequence of N points xn (N being an integer) is:
[0178] In the case where N is the product N = Ni x N2 of two integers Ni and N2, it can be written: ~ ='X
[0179] There are two ways to decompose the discrete Fourier transform, both ways corresponding to the cascade of a discrete Fourier transform with Ni points and a discrete Fourier transform with N2 points. It is the order in which these discrete Fourier transforms are carried out that differs.
[0180] One of the discrete Fourier transforms involved is weighted by a local oscillator allowing a fine shift of the spectrum; the exponential term allowing this translation is called the “tweedle factor”: hveedle factor =«
[0181] In principle, the output index varying most rapidly is mb since it corresponds to the "fast" discrete Fourier transform, i.e. not multiplexed. Thus, the output order of the frequencies is not the natural order: in fact, the m with a value of 0 modulo Ni first come out, then the m with a value of 1, and so on up to the m with a value of Ni - 1, always 0 modulo Ni; this order is called "bit reverse" because it corresponds to reversing the binary representation of the index to obtain the output rank (for a number of points in power of 2).
[0182] The realization of the inverse discrete Fourier transform, which recovers the frequency data in “bit reverse”, is carried out in a dual manner, by reversing the operations.
[0183] According to a first case, the discrete Fourier transform at Ni points (“fast”) is followed by the N2-point Fourier transform (“multiplexed”), which corresponds to: tweedfe factor FFT to Ni (Radix N;) N2-point FFT multiplexed by Ns (Radix N,)
[0184] The realization of the Fourier transform is carried out as the cascade of a non-multiplexed Ni-point Fourier transform and a N2-point Fourier transform multiplexed by Nb. The “tweedle factor” is placed at the head in front of the first Fourier transform stage.
[0185] According to a second case, the discrete Fourier transform with N2 points (“multiplexed”) is followed by the Fourier transform with Ni points (“fast”), which corresponds to: facto N7 points multipfexed by N, (Radix N ! N-point FFT (Radix N, )
[0186] This time, the implementation is carried out as a cascade of a discrete Fourier transform with N2 points multiplexed by Ni and a non-multiplexed discrete Fourier transform with Ni points. The "tweedle factor" is placed between the two discrete Fourier transform stages. Since the first discrete Fourier transform carried out is multiplexed, this decomposition accommodates parallelized input data; the multiplexing then becomes parallel processing.
[0187] Following one of the classical versions of radix2 decomposition, the 2M-point FFT of this sequence an is written: = (-ir
[0188] In particular:
[0189]
[0190]
[0191]
[0192]
[0193]
[0194]
[0195]
[0196]
[0197]
[0198]
[0199] either : = FFTV 1F FFT^z^^ where FFTm denotes the M-point FFT. It should be noted that this spectral density can also be written in the following two ways which will simplify its implementation: And = FFT^j-1 / ¾ + By denoting / ¾ the spectral response of the filter, the filtered response is obtained by inverse transformation: The response is calculated only for points 0 < k < Ml (due to the folding inherent in the FFT) to finally obtain: / K ..A......) where IFFTm denotes the inverse FFT at M points. 7.2 - REALIZATION OF A DELAY It is known that the complex gain corresponding to a delay r is H(f) = c As soon as we want to carry out this operation on a signal sampled for a delay different from a multiple of the sampling period we come up against the problem of aliasing of the frequency response which in principle is zero when fECHT is not an integer. We must then try to produce the filter having a periodic complex gain of period fECn and response H(f) = e over this period. It may be recalled here that, in principle, shifting a digitized signal xm by a delay r amounts to interpolating the signal x by the Shannon interpolator filter whose time response is sinc(fECHt) where sinc(x) = sin(irx) / (jix) and replacing the sample number m of value xm by ym = x(m TECh - t). Mathematically this is written: r*«sin (mT^„ —,r - ri v '■ , 44 gin JotTL..,_ r -î- -I v = --^0——---Y -r^? - *7*™. = Y -tv--
[0200] The time response of the digital filter ensuring the delay r is therefore: x —r I st ïi 'Wr,'-r (—-0 / ?.. ~ —--—- =---x —--:--- -r)
[0201] In principle, this filter is not feasible since its time response extends from -oc to +½. By truncating and weighting its response between -N and +N, we produce a set of coefficients corresponding to a filter which ensures a differential delay r with respect to a reference delay equal to N TECh- The frequency response of this filter is then the folding of the frequency response of the weighting.
[0202] In any case, the slowness of the decrease of hn with the index n (hn ~ 1 / n) leads to very long lengths since both broad band (close to the Nyquist band) and large dynamic range (good suppression of image bands) are sought.
[0203] The equations presented above indicate that this slowness is a direct consequence of the fact that the complex gain gW? associated with a delay T is discontinuous at the limits of the sampling band since r is not a multiple of the sampling period l / fEcu- In the same way, the complex gain to which the equalization corrections of the reception channels correspond is in general neither continuous nor continuously derivable at the limits of the sampling band so that, here again, the duration of the time response of this correction filter is all the greater as the band in which one wishes to ensure this correction with precision is closer to the Nyquist band associated with the sampling.
[0204] It results from all these observations that this filter leads to an infinite time response hn slowly decreasing with the index n due to the discontinuity at the edge of the band.
[0205] Now it is known that the weighting of a response makes it possible to limit its extension in the dual space; it is indeed common to weight a temporal response to limit the spectral congestion inherent in the discontinuity at the limits of the sampling domain. In a dual manner, the weighting in the spectral domain, which corresponds to a filtering, makes it possible to greatly limit the temporal extension associated with the realization of a non-integer delay (in terms of sampling periods), by reducing by filtering the band in which the delay will be operational.
[0206] Thus, the wider the band in which the delay is precisely ensured, the greater the length of the time response of the FIR filter producing it.
[0207] 1.3 - IQ DEMODULATION - HILBERT FILTERING
[0208] Furthermore, it is possible to sample the real signals on carrier (and no longer in baseband after analog I / Q demodulation). In this case, in principle, the positive frequencies of the useful signal are strictly contained in one of the Nyquist bands of the sampling (i.e. between k fECH / 2 and (k+1) fECH / 2 where k is arbitrary). After sampling, the reconstruction of the analytical signal then passes through a filtering, called Hilbert, which makes it possible to eliminate the spectrum of negative frequencies. Usually, as indicated in [Fig.5] representing the time response of a Hilbert filter (real and imaginary parts), the time response of this filter includes a real part limited to a single central coefficient and an antisymmetric imaginary part decreasing in 1 / n and of which one sample out of two is zero.
[0209] In doing so, as indicated in [Fig.6] which illustrates the frequency response of a Hilbert filter, the complex gain is -6 dB at the zero frequency and at the Nyquist frequency; moreover, the level of rejection of negative frequencies is all the higher as the duration of the response is significant.
[0210] If the real signal does not occupy the entire Nyquist band or if the complex signal does not occupy the entire sampling band, it is customary to filter the useful band (denoted BU in [Fig.6]) occupied by the signal to allow a subsequent decimation step aimed at bringing the sampling frequency as close as possible to the useful band. In this case, the response of the filter resembles that of the Hilbert filter, except that its band is narrower.
[0211] 2 - DESCRIPTION OF EXAMPLES OF FORMING METHODS
[0212] 2.1 - GENERAL PRINCIPLE OF THE TRAINING PROCESS
[0213] The training method can be applied indifferently directly to a real signal after analog-digital conversion, or to a complex signal, after return to baseband, I / Q demodulation and partial decimation. The method is intended to jointly carry out the different steps, the equalization of the reception channels, the delay and the weighting associated with the depointing and the limitation of the occupied band.
[0214] For this, the method is carried out by the filtering structures presented in the paragraph entitled “Production of filters with long time responses” which makes it possible to control the filtering in the spectral domain and to make the operations independent and multiplicative.
[0215] It is also possible to duplicate the delay operation as many times as necessary to create the contribution of a reception channel to as many beams as required.
[0216] It should be noted that it is desirable that the response of the Hilbert filter or that of the filter limiting the spectrum to the useful band sufficiently filters out the discontinuities associated with the delay and equalization operations in order to avoid these introducing notable artifacts caused by these spectral discontinuities. This is shown schematically in [Fig.7] which shows the modification of the Hilbert filter for total elimination of the second Nyquist band (left graph before modification and right graph after modification).
[0217] In the case where the input signal is complex and the signal spectrum occupies almost the entire sampling band (fECH), it is appropriate to filter in order to avoid problems related to the discontinuity of the frequency response of the delay at the limits of the sampling band. In this case the time response of the filter is a sub-sampling by 2 of that of a filter similar to the Hilbert filter, but designed at double frequency. [Fig.8] illustrates this by showing the filter required when the input signal is complex to avoid problems related to the discontinuity of the frequency response of the delay at the limits of the sampling band.
[0218] 2.2-DESCRIPTION OF AN EXAMPLE OF A METHOD FOR FORMING BEAMS 2.2.1 - Case of figures 9 and 10
[0219] It is proposed to exploit the filter architectures presented in Figures 2 to 4 as a constituent element of a beamforming system architecture.
[0220] The beamforming system is a system capable of forming Q beams, Q being a non-zero integer.
[0221] Each beam can thus be identified by an index q, an index which is an integer between 1 and Q.
[0222] It should be noted that in a particular embodiment of this scenario, Q can be equal to 1.
[0223] The beamforming system comprises P reception chains each equipped with an antenna. P is an integer greater than or equal to 2.
[0224] A reception chain is an assembly formed by an antenna and a processing circuit.
[0225] The term receiving channel is sometimes used for the processing circuit. This term will be used in the remainder of the description.
[0226] Each of the P reception channels is thus capable of receiving signals coming from a respective antenna.
[0227] Furthermore, each of the P reception channels comprises means for carrying out operations which will be described later. In particular, each reception channel comprises spectral analysis means, the beamforming here having the particularity of comprising operations in the spectral domain, and more specifically the depointing operations.
[0228] Each reception channel can be identified by an index p which is between 1 and P.
[0229] It should be noted:
[0230] • Cp(m) the complex correction to be applied on the reception channel p in order to equalize it with the other reception channels with a resolution equal to fECH / (2M), the index m which, modulo 2M, is between 0 and 2M-1, is representative of the frequency,
[0231] • the delay to be applied on the reception channel p to form the beam q and, if f0 is the carrier frequency brought back to the continuous after sampling, ([^(m) = -2ir ( f0 + m fECH / (2M) ) Tp^-1 is the corresponding spectral phase,
[0232] • Opta' a possible real weighting coefficient of the antenna aperture, for each index q of the formed beam, the weighting law cip^' being intended to improve the level of the lobes of the directivity diagram of the antenna thus formed digitally, and
[0233] • HW(m) the complex gain associated with the band limitation specific to each shaped beam (Hilbert filter or limiting the spectrum to the useful band); this complex gain comes from a 2M-point FFT of its causal time response.
[0234] This can be interpreted as a frequency-dependent matrix operation that links the vector of signals from the receiving channels to the vector of the formed beams. This operation is decomposed as the product of three matrices, the first and last of which are diagonal:
[0235] • [ C ] the diagonal matrix P x P of the equalization corrections of the reception channels,
[0236] • [ aei<]) ] the full matrix Q x P of the spectral phases associated with the delays and the opening weighting coefficients, and
[0237] • [ H ] the diagonal matrix Q x Q of the band-limiting filters.
[0238] If [ X ] is the vector P x 1 of power spectral densities (sometimes called under the acronym DSP) of the reception channels and if [ Y ] is the vector Q x 1 of the power spectral densities of the channels formed, the operation is then written: [ Y ]Qxi = [ H ]qzq x [ o®* W x [ C x [ X ]pxi (matrices and vectors depend on the frequency of an antenna).
[0239] It will be noted that these three filtering functions, which are applied in a multiplicative manner due to the filtering structure, since they are operated in the spectral domain, are linked in a very particular order which is that of the matrix decomposition cited:
[0240] • the channel equalization [C] is applied first, after the FFTs which form the passage in the spectrum, then
[0241] • the delay as well as the opening weighting of the antenna । aej$ j, operation duplicated on each reception channel as many times as there are beams to be formed, and finally
[0242] • the specific band limitation [ H ] applied to each beam formed, that is to say after summation of the different contributions from all reception channels.
[0243] Figures 9 and 10 describe a beamforming system in the particular case where P and Q are equal (here equal to 4), that is to say that it is a 4 beamformer from 4 reception channels.
[0244] In these figures, for greater readability, the lines are increasingly wider depending on the index p or q considered.
[0245] The beamforming system is referenced 10 and comprises 4 reception channels 12b 122j 123, 124.
[0246] Schematically, the training system 10 comprises, for each reception channel 12, an input 14 operating in the time domain, a processing circuit 16 operating in the frequency domain and an output 18 operating in the time domain.
[0247] Input 14, processing circuit 16 and output 18 implement operations and can be interpreted as computer program products.
[0248] In particular, they can be implemented at least in part by a computer, such as a programmable logic circuit. Such a circuit is often referred to by the acronym FPGA, an English acronym for the expression “Field-Programmable Gate Array”.
[0249] In another example, they form an application-specific integrated circuit. Such a circuit is often referred to by the acronym ASIC (an acronym for "Application-Specific Integrated Circuit").
[0250] This means that these elements perform a beamforming function by calculation, this function sometimes being designated by the acronym FFC. The acronym DBF is sometimes used in reference to the corresponding English term “Digital BeamForming”, meaning “digital beamforming”).
[0251] In the example described, the processing circuit 16 comprises two calculation channels 20 and 22 which are each connected to the input 14 by an FFTM application unit 24 and to the output 18 by an IFFTM application unit 26.
[0252] The two calculation channels 20 and 22 carry out similar operations on the points having respectively an even index (calculation channel 20) and an odd index (calculation channel 22).
[0253] On each channel, there is an equalization unit 28, a delay and weighting unit 30 (shown larger in parts A to H of figures 11 and 12), a summation unit 32 and a band limitation unit 34. These notations are only introduced on the first calculation channel 20 of the first reception channel 12i to guarantee a certain readability of figures 9 and 10.
[0254] The equalization unit 28 applies the aforementioned matrix C, with the notation Cp(2m) if this is the first way 20 (reference to the fact that even indices are processed) and the notation Cp(2m+1) if it is the second way 22 (reference to the fact that odd indices are processed).
[0255] As visible in figures 11 and 12, the delay and weighting unit 30 applies delay and weighting coefficients as many times as there are beams to be formed, i.e. 4 here.
[0256] The summing unit 32 of the first channel 20 sums the contributions obtained after the delay and weighting unit 30 coming from the first channels 20 of each reception channel 12.
[0257] For the case of the first channel 20 of the first reception channel 12b this means that the summing unit 32 comes to sum the output of the delay and weighting unit 30 of the first channel 20i (the index 1 indicates that it is that of the first reception channel 12J, the output of the delay and weighting unit 30 of the first channel 202, the output of the delay and weighting unit 30 of the first channel 203 and the output of the delay and weighting unit 30 of the first channel 204.
[0258] Similarly, the summation unit 32 of the second channel 22 sums the contributions obtained after the delay and weighting unit 30 coming from the second channels 22 of each reception channel 12.
[0259] For the case of the second channel 22 of the first reception channel 12b this means that the summing unit 32 comes to sum the output of the delay and weighting unit 30 of the second channel 22b the output of the delay and weighting unit 30 of the second channel 222, the output of the delay and weighting unit 30 of the second channel 223 and the output of the delay and weighting unit 30 of the second channel 224.
[0260] This being true for each summing unit 31, it is therefore clear that a relatively complex interconnection 36 is visible between the delay and weighting units 30 and the summing units 32.
[0261] The band limiting unit 34 applies the matrix H and the same notations as for the matrix C are used.
[0262] Output 18 performs a recombination of the calculation channels 20 and 22 using a shift unit 38 and an adder 40.
[0263] It should be noted that, depending on the application cases, the delay to be introduced on the different channels to depoint a beam can be greater than one clock period. For example, for an antenna covering a relative band of 20%, and having 32 x 32 or 1024 elements, and a sampling frequency passing the total band of the antenna, the maximum delay associated with the formation of beams in the two directions is 2 x 16 x (l / fo / 2) = 16 (fEcu / B) x (B / f0) / fECH = 6.4 / fECH. We could then imagine realizing the integer part of this delay digitally in the time domain, but this is no longer possible once we have switched to frequency to simplify the simultaneous production of several distinct beams. This delay must therefore be fully implemented in the spectrum and it must be provided for in the design of the filters, in the sense that their increased time response of the maximum delay must remain lower than N FFï / fECH- 2.2.2 - Generalization
[0264] The elements which have just been described can easily be generalized, which leads to the training process now explained.
[0265] In a general case, the method of forming Q beams is implemented by a beam forming system comprising P reception channels each connected to a respective antenna, Q being a non-zero integer and P being an integer greater than or equal to 2. The method comprises, for each reception channel, the steps of:
[0266] - reception of a signal received by the antenna,
[0267] - generation of two calculation channels by sum and difference of the signal from the channel reception and this same signal delayed by M periods,
[0268] - realization on the difference channel of a frequency shift of fECH / 2M by ap multiplicative plication of a signal e JnPi / M, n being an integer ranging from 0 to Ml, M being an integer strictly greater than 2,
[0269] - implementation of first operations on the first calculation path, the first operations including:
[0270] - the application of a discrete Fourier transform to M points on a signal from the transmitted signal to obtain M points in the spectrum of the signal from the transmitted signal, M being an integer strictly greater than 2, each point of the spectrum of the transmitted signal corresponding to even indices of a spectral analysis at 2*M points of the signal from the transmitted signal and being identified bijectively by an index which is an even number between 0 and 2*M-1,
[0271] - the application in the spectrum of a channel equalization function on the M points, to get equal points,
[0272] - the application in the spectrum of a delay on the equalized points to generate Q series of M delayed points (by means of multiplication with a linear exponential in phase), and
[0273] - implementation of second operations on the second calculation path, the second operations including:
[0274] - the application of a discrete Fourier transform at M points on the signal from the transmitted signal to obtain M points in the spectrum of the signal from the transmitted signal, each point in the spectrum of the signal from the transmitted signal corresponding to odd indices of a 2*M point spectral analysis of the signal from the transmitted signal and being identified bijectively by an index which is an odd number between 0 and 2*M-1,
[0275] - the application in the spectrum of a channel equalization function on the M points, to get equal points, and
[0276] - the application in the spectrum of a delay on the equalized points to generate Q series of M delayed points (by means of multiplication with a linear exponential in phase),
[0277] - formation of each beam from the series of M delayed points generated, the training including, for each bundle:
[0278] - implementing third operations, the third operations comprising:
[0279] - the summation of the series of M delayed points coming from each first channel of calculation of a reception channel, to obtain summed points,
[0280] - applying a band limiting function to the summed points for get limited points, and
[0281] - the application of an inverse discrete Fourier transform to M points on the points limited, to obtain a first output signal,
[0282] - implementing fourth operations, the fourth operations comprising:
[0283] - the summation of the series of M delayed points coming from each second channel of calculation of a reception channel, to obtain summed points,
[0284] - the application of a band limiting function on the summed points to be processed to get limited points, and
[0285] - the application of an inverse discrete Fourier transform to M points on the points limited, to obtain a second output signal, and
[0286] - the frequency translation of the second output signal to obtain a signal of translated output
[0287] - the recombination of the first output signal and the second output signal translated to obtain the formed beam.
[0288] In certain particular cases, the method may also have the following properties:
[0289] - the first and second operations also include the application of a aperture weighting of the antenna of the reception channel considered.
[0290] - the received signal being sampled at a sampling frequency, the function equalization has a resolution equal to the ratio of the sampling frequency by 2M; in a numbering from 0 to 2M-1 the even points are applied on the first calculation channel and the odd points on the second calculation channel.
[0291] - the limiting function is obtained by applying a Fourier transform discrete at 2*M points of a temporal response of 2*M points exclusively contained in the positive times (i.e. zero on the last M points).
[0292] 2.3 - DESCRIPTION OF AN EXAMPLE OF A PROCESS FOR FORMING BEAMS
[0293] In this other example of a beamforming method, the method also makes it possible to optimize the volume of output information by reducing the flow rate of each beam according to its useful band and by multiplexing, on the same output channel, beams of the same band lower than the flow rate of the channel.
[0294] Indeed, if the reception band of a given beam is significantly lower than fECH / 2 for real signals and fECH for complex signals, it is desirable to decimate the signal by an integer or non-integer factor which can be as high as the ratio of the bands; this is only possible in a simple manner, of course, if this factor is an integer greater than 2 and preferably a power of 2.
[0295] 2.3.1 - Decimation by 2 (for example, passing from real to I / O)
[0296] For example, when the input signal is real, it is customary to decimate the signal by two after Hilbert filtering in order to reduce the flow of information; this is made possible by the fact that the negative frequencies have been eliminated from the spectrum. Taking this into account, we will endeavor to restore the signal only for even times at the output of the M-point IFFTs. If Hm is the input of the IFFT and hn its output, we can then write the following relationship, by the very fact that the negative frequencies are eliminated by the filter at a level low enough to be considered as zero in the sense of the instantaneous dynamics: k = yn =y# yh
[0297] Which expresses that the even points, the only ones retained, are constituted by IFFTs at M / 2 points of only positive frequencies.
[0298] However, the natural output of the FFT is carried out in “bit reverse”, which consists, for an M-point FFT, of alternately outputting positive frequencies and negative frequencies, more precisely a point of positive frequency Hm immediately followed by the point of negative frequency Hm+M / 2. The restriction of the output of the FFT to only positive frequencies then appears as a decimation of this output by 2, which means that the fact of outputting the spectrum in “bit reverse” is the expression of the duality which exists between sub-sampling by 2 in the time domain and positive / negative sub-bands in the spectrum.
[0299] This feature can then be advantageously used to time multiplex two different beams; this is made possible thanks to the internal architecture of the DFT with M = 2N points forming an IFFT. DFT is the acronym for the English name “Discrete Fourier Transform” which means “Discrete Fourier Transform”.
[0300] Indeed, it is known to advantageously decompose the inverse DFT with 2N points into N successive stages called “radix2”, each forming a 2-point inverse DFT (which identifies with the 2-point DFT) and separated by time-varying phase rotation stages called "tweedle factors". In such a decomposition, these inverse 2-point DFTs operate on data whose multiplexing rate increases along the 1TFFT (1 for the first stage, 2 for the second, 4 for the third, etc.).
[0301] Due to the “bit reverse” output, Hm and Hm+M / 2 are two points of the spectrum, one in the positive frequencies and the other in the negative frequencies, which are successively delivered at the output of the M-point FFT. Therefore, formally, the output of the M-point IFFT is written: R i ^2 ( 1 X 'y e .M x
[0302] We note that, in fact, the first step of this M-point IFFT consists of a DFT of the 2 consecutive points Hn and Hn+M / 2 whose result, sum and difference, also comes out consecutively. This result is then multiplied by a different “tweedle factor” for the two points thus created.
[0303] It is also noted that the two results thus multiplied form two interlaced series of points which each undergo an M / 2 point IFFT, which leads to the diagram in [Fig. 13], which illustrates the decomposition of an M point IFFT into a 2 point DFT and an M / 2 point IFFT.
[0304] As shown in [Fig. 13], an M = 2N point IFFT can be considered as the cascade of a non-multiplexed 2-point DFT and a multiplexed 2-point M / 2 = 2X 1 point IFFT, which amounts to simultaneously performing 2 separate M / 2 = 2 N1 point IFFTs on two interleaved series of points.
[0305] This therefore makes it possible, by short-circuiting the leading 2-point DFT stage and the "tweedle factor" stage which directly follows it, to use an M-point IFFT operator, either as such or as an M / 2-point IFFT operator multiplexed by 2, i.e. working in a multiplexed manner on 2 interlaced series of points.
[0306] It is this capacity which can be advantageously implemented to multiplex on the same output 2 distinct beams decimated by 2 as shown in [Fig. 14] which illustrates a beam forming system with multiplexing of 2 beams decimated by 2.
[0307] In this case, the digital local oscillator is adapted to be maintained for 2 periods (multiplexing rate here).
[0308] Decimation by 2^ of the band between 0 and -1 + 2^
[0309] By reproducing K times the decomposition presented in the previous paragraph, we can exploit the fact that at the output of an FFT with M = 2N points, the frequency points for 0 < k < -1 + 2K are presented successively.
[0310] For example, by decomposing in the same way the M / 2 point IFFT into a 2 DFT points, a "tweedle factor" stage and an M / 4 point IFFT, we can then also process 4 series of interlaced points in a multiplexed manner since only the frequencies from 0 to -1 + M / 4 are retained and we can therefore decimate by 4 after applying the anti-aliasing filtering which eliminates the other frequencies.
[0311] By generalizing this approach, it is possible to consider more generally that an IFFT with M = 2N points is the cascade of a non-multiplexed inverse DFT with 2K points, a stage of "tweedle factors" and an IFFT with 2XK points multiplexed by 2K, which amounts to simultaneously performing 2K IFFT with 2XK distinct points on 2K interlaced series of points.
[0312] Thus, step by step, by short-circuiting the leading 2K point inverse DFT stage and the tweedle factor stage which follows it directly, it is also possible to process in a multiplexed manner 2K series of points interlaced by the same IFFT multiplexed by 2K, since only the frequency points from 0 to -1 + M / 2K are kept and it is therefore possible to decimate by 2K. The capacity described previously for decimation by 2 and multiplexing of 2 beams can thus be extended to decimation by 2K and multiplexing of 2K different beams.
[0313] One way to achieve this is shown in [Fig. 15] which shows the modification of 1TFFT with 2N points for the multiplexing of 2K beams decimated by 2K. As shown in this [Fig. 15], it is sufficient to provide 1TFFT with 2N points with a set of switches allowing, via appropriate addressing, to enter 1TFFT at the required location to carry out a IFFT processing with 2X K points operating on 2K multiplexed data. The addressing of the different switches is carried out by means of the binary description of 2K - 1.
[0314] Such a block can be integrated into the training system of [Fig. 16] to make it work with a decimation / multiplexing rate of 2K. The block is indicated by "block [Fig. 15]" in [Fig. 16] which shows such a generalized training system.
[0315] 2.3.2 - Case of bands not included between 0 and -1 + 2^
[0316] We will note bit_reverse(k,K) the function which associates with k the integer given by the reversal of the bits of the description of k on K bits.
[0317] It can be noted that bit_reverse(bit_reverse) is the identity function.
[0318] Furthermore, we will have for example:
[0319] • bit_reverse(0,K) = 0.
[0320] • bit_reverse(2K *,K) = 1 and,
[0321] more generally, bit_reverse(2K M,K) = 2M-1 for 1 < M < K,
[0322] • bit_reverse(2k+l,K) = bit_reverse(2k,K) + 2K 1 for 0 < k < 2K1 - 1.
[0323] Generally speaking, the band of interest to be kept, consisting of 2N K boxes, m + kx 2X K and m - 1 + (k+1 ) x 2X K with 0 < m < 2K - 1 :
[0324] • the case m = k = 0 corresponds to the previous paragraph in which the band is between 0 and -1 + 2N K,
[0325] • if m = 0 and k 0, the band of interest corresponds to one of the natural sub-bands of the FFT; the boxes that compose it then come out periodically but with an offset equal to bit_reverse(k,K),
[0326] • As soon as m 0, the different boxes constituting the band of interest no longer leave periodically.
[0327] Although these three cases lead to differences in the behavior of the data output order, we can nevertheless bring together all three of these operating cases under a single operating mode.
[0328] To do this, simply:
[0329] • read the spectrum only at the times corresponding to the frequency boxes constituting the useful band, that is to say at indices n such that:
[0330] nM1N = m + k 2X K < bit_reverse(n,2N) < nMAx = m - 1 + (k+1) 2X K
[0331] • hold the value until the next sampled value, and
[0332] • resample the signal thus obtained at times nx 2K - 1.
[0333] An implementation of such an operating mode is shown in [Fig. 17] which represents a device for frequency recalibration of the decimated band comprising in particular a counter, a bit reverse unit and two sample-and-hold units.
[0334] In doing so, a completion latency of 2K clock pulses is introduced. In the example of [Fig. 17], it is assumed that there is no completion latency, these latencies being possibly taken into account by the use of devices known to those skilled in the art. In order not to break the synchronism of the operations which follow and lead to a fixed latency whatever the multiplexing rate, the compensation of this latency can, for example but not exclusively, be carried out by the application of an additional delay equal to 2N -2K clock pulses on the signal thus obtained.
[0335] This creates a folding of the useful band into the sub-sampled band [ 0 - f ECh / 2k ]. In fact, the order of the data at the FFT output is as follows:
[0336] • first the 2K samples 0 of each of the 2K sub-bands, these sub-bands appear arising in the order set by bit_reverse(x,2K), then
[0337] • other samples, delivered in successive packets of 2K values corresponding each has the same index in all sub-bands, the order of the indices being fixed by bit_reverse(x,2K N).
[0338] Therefore, the first sample of interest (i.e. to belong to the useful band) to come out of the FFT is the sample such that (m + kx 2X K) modulo 2ir is zero, i.e. sample 0 of the kth sub-band in the case where m = 0 and of the (k+l)th if m 0.
[0339] When m = 0, the samples travel through the kth sub-band; they are all spaced of 2K. If m 0, the samples of interest belong either to the kth sub-band or to the (k+l)th.
[0340] However, it turns out that at the FFT output the samples are presented in slices of 2K points, a slice representing all the sub-bands for the same given position within the sub-band.
[0341] Consequently, by sampling at the times corresponding to the output times of the 2X K points of the useful band and maintaining the value until the next frequency point, we have at the end of a slice, that is to say at t = nx 2K - 1, a frequency sample whose position in its original sub-band is indeed that corresponding to the slice in question.
[0342] In doing so, the last samples of the kth sub-band were placed at the end of the band thus created; this corresponds to the folding of the useful band, positioned across two sub-bands, into the sub-sampled band [ 0 - fEcH / 2K ].
[0343] This folding corresponds to a frequency transposition of -(k+1) fECH / 2N K. Thus, the center of the processed band is located at [ 2 x (2m - 1) - 2 K ] / fECH / 2 in the sub-sampled band [ 0 - fEcu / 2K ].
[0344] The evolution which follows with regard to the beam forming system with multiplexing of 2K beams decimated by 2K but now occupying the same band located in any way in the sampling band is presented in [Fig. 18], figure in which each block indicated "block of [Fig. 17]" comprises the elements of [Fig. 17].
[0345] It is then clear that, as will be presented in the following paragraphs, the pointing operations of these 2K multiplexed beams must be operated in a manner that is also multiplexed as a set of 2K delays, and as a possible set of 2K weightings of the aperture of the antenna.
[0346] 2.3.3 - Decomposition into successive partial beam formations
[0347] When the number of reception channels becomes large, it is not possible to carry out all the calculations in a single device, both because of the number of inputs / outputs that this implies, and for reasons of computing power. In this case, it is appropriate to distribute all the calculations over several devices. The natural break in this breakdown of the calculations is at the level of the summation of the contributions of the different channels, an associative operation by nature which also generally results in a concentration of the data due to the fact that the number of beams formed is very often much lower than the number of reception channels, especially since, as soon as the useful bandwidth of a beam is reduced, it is appropriate to carry out multiplexing as described previously in this section.The receiving channels are then grouped into subgroups and the beamforming part that concerns them forms partial beams or sub-beams, including recombination. additive is also carried out in an additional grouping device, which can itself be split into several successive concentration stages if the number of inputs / outputs is too large.
[0348] This is shown on the one hand in [Fig. 19] which corresponds to a synoptic representation of the operation and on the other hand in figures 20 and 21 which each represent a part of the beam forming system which would thus be obtained (an example is proposed here with four sub-beam formers and two sub-beam combining stages). 2.3.4 - General case
[0349] The elements which have just been described can easily be generalized, which leads to the training process now explained.
[0350] In a general case illustrated by [Fig. 19], the method for forming Q beams is implemented by a beamforming system comprising P reception channels each connected to a respective antenna, Q being a non-zero integer and P being an integer greater than or equal to 2. The Q beams are implemented by Q' groups of beams calculated in a multiplexed manner by the same operator as described previously. The method comprises, for each reception channel, the steps of:
[0351] - constitution of two calculation channels by sum and difference of the signal from the reception channel and of this same signal delayed by M periods, and
[0352] - on the second calculation channel (difference channel), carrying out a fre- quantity of mr / 2M by multiplicative application of a signal e(-jmr / M) (or e(4nit / M' depending on the notation used, both designating the same operation), n being the current variable ranging from 0 to Ml,
[0353] - application of the FFTs to M points on each calculation channel allows to pass into the frequency domain, and
[0354] - an application of a multiplicative spectral correction function intended to equalize the reception channels,
[0355] - by duplication are created the Q' ways intended to form the Q' groups of multiplexed beams also numbered q' between 1 and Q'.
[0356] - implementation of operations including possibly apé sampling periodic then a resampling, according to the technique of figure 15, to carry out an effective decimation by pA7 of these Q' series of points maintained for periods, which allows to subsequently carry out the multiplexing of 2^« beams. Such multiplexing makes it possible to pass several formed beams in a given time and therefore to increase the capacity.
[0357] All of these operations are intended to obtain Q=sum of ^1' over q' ranging from 1 to Q' contributions of the reception channel to the formation of the Q beams.
[0358] Another way of presenting this method would be to use operations as proposed in the context of section 2.2.1 (case of figures 9 and 10), this equivalent description not being repeated in what follows so as not to make the present description cumbersome.
[0359] On each of these Q contributions partially multiplexed into Q' processing channels each comprising two calculation channels (even and odd frequencies of the spectrum) application of: - a linear spectral phase delay and - an amplitude weighting.
[0360] Then, summation of the processing channels from each of the reception channels to form Q beams partially multiplexed into Q' processing channels. This summation operation can be carried out in accordance with paragraph 2.3.3 by the constitution of intermediate sub-beams.
[0361] On each of the two calculation paths of each of the Q' processing paths referenced by the index q': - application of a multiplicative band-limiting function. - application of an iFFT to multiplexed points by .
[0362] - on the second calculation channels (odd frequencies), transposition of the spectrum by a frequency shift of mr / 2M by multiplicative application of a signal e[+j floor(n / 2Æ'> jjt / ( M / p^ )], n being the current variable going from 0 to Ml and the floor function designating the integer part function,
[0363] On each of the referenced processing channels, summation of the two calculation channels to ultimately form the Q beams partially multiplexed into Q' groups of beams referenced by the index q' and multiplexed by 2^« •
[0364] 3 - DESCRIPTION OF A CALIBRATION METHOD
[0365] As indicated, in any beamforming, the quality of the implementation degrades when the reception channels do not have exactly the same response. It is also known that the differences in response, both in spectral amplitude and in spectral phase, are corrected prior to any beamforming, which the forming system can do via the equalization stage with the cp coefficients. These correction data relate to the amplitude and the spectral phase of the complex gain of each of the channels and they are generally acquired by means of a calibration.
[0366] In principle, beamforming does not require that the channels be equalized, i.e. that their responses be brought back to the same flat response, but simply that they be realigned with respect to each other, i.e. that the complex gain differentials between channels are compensated so that all reception channels are brought back to the same average response.
[0367] On the other hand, channel equalization, which is therefore not a requirement associated with channel formation, can be justified elsewhere, for example to ensure that the appropriate processing (pulse compression) is carried out with the desired filtering function.
[0368] Whether for equalization or simple recalibration, the principle of calibration consists of injecting the same test signal into the different channels and comparing the received signal to a reference.
[0369] Depending on whether the aim is equalization or simple recalibration, the calibration philosophy regarding this reference can vary:
[0370] • for equalization, which consists of calibrating each channel independently, the comparison to the reference must be used to extract the response of each channel; for this the reference which constitutes the test signal must be known, but, as the calibrations of the different channels are independent, it is not required that the same signal be used for all the channels, and
[0371] • for a recalibration, only knowledge of the response differential between the channels is sought, so that knowledge of the calibration signal is not required, since each channel is compared to the other channels; it is sufficient to inject, without needing to know it, the same signal into two channels to access the differential between these channels.
[0372] Whatever the objective sought, equalization or recalibration, calibration therefore involves comparing two signals. Traditionally, the signals used for calibration are sinusoidal signals and their use makes it possible to obtain amplitude-phase gain information to be recalibrated for each of the frequencies used.
[0373] But it is also possible to use noise as long as a spectral analysis is available in each reception channel; this allows the measurement of the transfer function differential between two reception channels.
[0374] For example, a method for determining the phase differential coefficient of two reception channels, each comprising an analog-to-digital converter, may be envisaged, each analog-to-digital converter being clocked at the same sampling frequency; according to the rules of the trade, the reception channels comprise, before sampling, filtering steps intended to limit the band of the received signals to a single Nyquist band of the sampling function. The method comprises at least one step of:
[0375] - injection of a signal at the input of each of the reception channels, the signal being a optionally white noise signal, to obtain output signals,
[0376] - obtaining the output signal from each analog-digital converter,
[0377] - spectral analysis, by frames, of each output signal, this spectral analysis allowing a plurality of frequency channels to be available, each channel being identified by an integer,
[0378] - estimation, by integration over several successive frames, of the correlation in power between at least two channels of the same index coming from the spectrum of the signals from each of the analog-digital converters, the phase differential between two reception channels being carried by the phase of this correlation, and
[0379] - on each of the reception channels, estimation, by integration over several frames successive spectra obtained, of the PSD of the signal, the amplitude differential between the two reception channels corresponding to the ratio of the PSDs thus estimated.
[0380] The signal to be injected can be obtained by implementing the following steps: - the generation of a digital calibration signal,
[0381] - conversion of the calibration signal into analog,
[0382] - the frequency transposition of the converted calibration signal,
[0383] - the amplification of the transposed calibration signal, and
[0384] - balanced distribution to the different reception channels.
[0385] Thus, such a method corresponds to the fact of using a noise, not necessarily white, but which can be considered as the result of the filtering of a white noise, makes it possible to go back to the amplitude and phase differentials via the inter-correlation between channels of the same rank and via the DSP at the output of each channel (which is obtained as the variance of the output of the different channels).
[0386] It is also possible to replace the output of the spectral analysis of one of the two channels by the spectral analysis of a reference digital signal constituting the test signal after the digital-to-analog conversion; by means of prior calibration of the transmission part, this makes it possible to carry out the absolute calibration of a channel with a view to its equalization.
[0387] In any case, that is to say whether the calibration is intended for equalization or recalibration, whether the type of signal used is coherent (sinusoidal) or not (noise), the calculation of the correction coefficients requires raw data obtained by correlation and / or partial spectral analysis (demodulation combined with low-pass filtering intended to calibrate a single frequency point is similar to a single-channel spectral analysis).
[0388] Furthermore, particularly because beamforming has, in the past, only been practiced in a narrowband application framework which reduces the controls to simple sets of complex gains (phase and amplitude) independent of the frequency, the calibration has, until now, been carried out in the spectral domain with sinusoidal test signals (one calibration point per working frequency).
[0389] However, the need to form broadband beams requires that the calibration also becomes broadband, which requires that the calibration signal itself be composed of several frequencies covering the working domain, whether sequentially or instantaneously.
[0390] Two philosophies therefore exist regarding the nature of calibration.
[0391] According to a first philosophy, the spirit of the old techniques is preserved by remaining in the spectral domain and by replacing the sinusoidal test signal with a succession of sinusoidal frames (this technique is sometimes also called according to the corresponding English term of “step frequency”).
[0392] This involves in particular setting up a sequencing of the calibration process involving both an agile generator for the synthesis of the test signal and all of the reception channels for the measurement.
[0393] The sequencer should also take into account the storage and transmission to a computer of the data acquired during the measurement process. The synchronization of this process also takes into account the stabilization times of each frequency step as well as the correct time alignment of these measurements from one reception channel to another. This last point is not a difficulty if all the reception channels are located on a single circuit, but becomes one when these channels are distributed over several circuits.
[0394] In contrast to this first philosophy is the use of broadband noise in the sense that this type of signal instantly contains all of the spectral components useful for calibration.
[0395] In this case, two approaches are possible, either temporal or frequency.
[0396] The temporal approach consists of calculating the covariance between a reference signal and a set of replicas of the received signal shifted by one clock pulse from each other; in doing so, this approach provides a time response that would have to be deconvolved by the autocorrelation of the reference signal, which is only possible in practice if this signal is white noise. This approach therefore does not allow, in particular, to take one channel as a reference for another channel in order to measure a differential quantity.
[0397] Thus the frequency approach is clearly to be favored; this approach is described previously with the aforementioned determination method.
[0398] This approach describes the possibility of carrying out a differential measurement between two reception channels provided that these channels are equipped with a spectral analysis. The technique consists, by frame-by-frame integration, in calculating the power spectral density pn(k) on each channel and the power correlation cn>m(k) between channels of the same rank k of each of the channels; the amplitude / phase differential between channels is then obtained as the normalization of the correlation by one of the DSPs:
[0399] cn>m(k) / pn(k) or cm>n(k) / pm(k) = cn>m(k)* / pm(k)
[0400] The advantage of this method is that it does not require any particular sequencing since all the frequencies of the spectrum are processed simultaneously.
[0401] Another advantage lies in the fact that the noise does not need to be white because it is sufficient that it can be considered as the result of the filtering of white noise and that all the useful spectral components are present there; thus a simple wideband noise generator (noise diode, high gain amplifier chain) is sufficient knowing that the analog reception chains are responsible, in principle, for limiting its band to the useful band with a selectivity consistent with the need for instantaneous dynamics.
[0402] It will be easily understood that the training system 10 is adapted to this second philosophy, since it involves a spectral analysis on each reception channel and that the resolution required for the calibration is, in principle, that of this spectral analysis.
[0403] Moreover, the presence of the first multiplier used for channel recalibration can be used, via a multiplexer, to produce the product inherent in the correlation calculation. The only additional resource required to ensure calibration is then an integrator multiplexed on M points (M being the length of the FFTs).
[0404] [Fig.22] shows a calibration device of the training system 10, the calibration device making some modifications to the receiving paths to ensure calibration:
[0405] - a multiplexer M for each channel taking as input the equalization function of the tracks (coefficients C(2m) or C(2m+1) omitting the track reference) and a reference signal noted R, and
[0406] - a DC calibration circuit comprising a frame-to-frame integrator.
[0407] These resources are duplicated on the even frequency channel and on that of the odd frequencies.
[0408] The calibration process described above makes it possible to carry out a differential calibration between several reception channels by correlation between the FFT outputs. This process is carried out autonomously as long as all of the reception channels are located on the same circuit.
[0409] If this is not the case, as has been mentioned, it is then necessary to "pass" at least one reception channel from one circuit to the other, which poses both a synchronization problem and increases the need for communication channels between circuits.
[0410] One solution to overcome this problem is to prefer absolute calibration, relative to a digital reference common to all reception channels.
[0411] For this, we will have two elements:
[0412] - a noise generator used for the synthesis of calibration noise; the signal from this generator, or a decimation of this signal (it is desirable that the noise is generated at a higher rate so as not to be degraded by its transmission chain), constitutes the reference signal, and
[0413] - a duplication of this reference generator on each of the circuits allowing form sub-beams from a subgroup of reception channels; the use of replicas thus formed eliminates the need to broadcast the reference signal to all circuits, which simplifies interconnections and thereby makes each circuit completely autonomous. In order to facilitate synchronization between different circuits, we will ensure that the generation of calibration noise carried out by this generator is synchronous with the FFT frames.
[0414] A second aspect of the calibration method consists, during the calibration phase, in substituting a synchronization signal for the signal from the FFTs. The use of the data path to convey this signal intended to reestablish synchronization makes it possible to direct this signal naturally towards the grouping blocks. If the grouping is carried out via several sub-blocks, it will be necessary to regenerate such a synchronization signal in each sub-block for the following sub-block, and, for the last one, for the final block ensuring filtering and time feedback.
[0415] The calibration method as described therefore does not require any specific resources in terms of real-time signal transmission: neither distribution of the common reference, nor exchange of signals between the different sub-beam formation circuits.
[0416] A very particular aspect of noise calibration processes via spectral analysis lies in the fact that the time shift between the measured channel and the reference signal tends to bias the measurement; this is due to the fact that the coherence between the two signals disappears at the edge of the frame.
[0417] Thus, the main effect of this partial de-correlation is a multiplicative bias in amplitude equal to 1 - I r I / T where r is the time shift between the two signals and T the duration of the spectral analysis, here 2NEET / fECH-
[0418] Other, lower order effects may also appear due to the decentering, such as ripples located mainly at the edge of the band. Even if this bias is similar for all the reception channels as soon as the calibration is carried out using a reference signal (and not between channels) and therefore does not impact the differential between channels, it is still desirable, to reduce this effect, that the calibration is carried out in two stages:
[0419] • a first stage, of short duration, intended to “rough out” the result in terms of delay: an order of magnitude of this delay is determined from the average phase slope, then rounded to the nearest multiple of l / fECH. From this initial measurement, it is possible to digitally recalibrate the signals via a recalibration device placed upstream of the processing, and
[0420] • a second stage, longer and more precise, not or very little biased due to the fact that it processes time-shifted signals.
[0421] This implies, of course, that a time recalibration device is introduced at the processing head, before passing into the spectral domain. Such a device, the diagram of which is shown in [Fig.21], comprises an adjustable delay line, directly controlled by the result of the first calibration phase (delay rounded in number of sampling periods).
[0422] This device is directly controlled by the delay to be applied which can be indifferently positive or negative between —2Q 1 and -1 + 2Q1.
[0423] When the delay to be applied is zero, that is to say in the initial position, the absolute delay applied by the device is 2Q 1; the delays are applied in relative relation to this absolute reference delay. The inversion of the most significant bit makes it possible to center this delay in the excursion of the multiplexer.
[0424] It may be noted that the calculation of the amplitude differential between the measured channel and the reference signal can also be carried out by calculating the square root of the ratio of their respective power densities. This has the advantage of not being biased in amplitude but, on the other hand, has the disadvantage of requiring an additional dedicated computing resource to estimate the PSD of the signal of the measured channel.
[0425] By reformulating what has just been described in a similar manner to what could have been done previously for the training method, this section relates to a method of calibrating the training system comprising:
[0426] 1) the exploitation of the first part of the architecture previously described (see previous sections) to operate:
[0427] - a 2M point spectral analysis distributed over two computing channels bearing respect tively the even frequencies and the odd frequencies of this analysis, and
[0428] - a multiplicative function in the spectral domain,
[0429] 2) a replication of the calibration signal generator on each of the circuits allowing sub-beams to be formed from a subgroup of reception channels, this generation being synchronized with the spectral analysis of point 1,
[0430] 3) the spectral analysis of this calibration signal by a device identical to that of the point 1.
[0431] 4) on each of the reception channels, the calculation of the power correlation between the spectrum of a receiving channel and that of the calibration signal and the PSD of the calibration signal, and
[0432] 5) the calculation of the correction to be made on the track as a ratio between the PSD of the signal calibration and power correlation, or, alternative solution, the exploitation of the phase of this ratio alone and the ratio of the DSPs of the reference signal and the signal of the measured channel.
[0433] As the replication of each calibration signal is synchronized with the analysis spectral of point 1 for each reception channel, by construction, the calibration process ensures that the reception channels are aligned with each other.
[0434] 4 - TRANSPOSITION OF TRAINING METHODS TO BROADCASTING
[0435] Beamforming on transmission, which could more accurately be called forming transmission paths in a multi-beam context, is generally based on the same principles as those set out for reception.
[0436] A certain number of groups of 2K signals sampled at fECH / 2K and multiplexed are available, the 2K sampling / multiplexing rank being specific to each group and each signal being intended for a transmission beam produced by a set of antennas. These different signals are complex and are included, in principle, in the sub-sampled band (i.e. [0 - fEcH / 2K ]); their transmission therefore requires that they be interpolated to be brought back to fECH on each transmission channel.
[0437] On the other hand, due to the potential sub-sampling by 2K, an ambiguity of a multiple of fECn / 2K, which is an ambiguity of principle, remains as to the real position of these signals in the final sampling band [0 - fEcu]- Furthermore, it is not to be excluded to take advantage of the beam formation operation to operate a frequency shift specific to each group of beams.
[0438] In a dual manner to the formation of beams at reception, the formation of the various beams consists of applying certain filtering to the signals:
[0439] - an interpolation adapted to the different signals, then
[0440] - the application of delays corresponding to the different beam-track pairs emission, accompanied by antenna weighting and followed by beam transposition, then
[0441] - additive groupings intended to form the emission channels, then
[0442] - the application of equalization corrections to the transmission channels.
[0443] All of these operations correspond to the reversal of the operations carried out in beamforming at reception.
[0444] In this context, it is the FFT which is modified to adapt to the multiplexing rate of the input data; this modification, which is the dual of the modification made to the IFFT in [Fig. 15], is the subject of [Fig.24].
[0445] The proposed multi-beam emission path forming system is schematically shown in [Fig.25].
[0446] This system includes in particular:
[0447] • the partial formation of the channels for the multiplexed beams corresponding to the block called "dual of the block of [Fig. 17]" which is described in [Fig.26]: this operation which can be interpreted as a transposition consists of integrating all the multiplexed beams which have been previously weighted and delayed. This operation is accompanied by sampling and resetting at the end of each slice (a slice of 2K samples corresponding to a frequency of the 2K multiplexed beams) and a hold for the duration of a frame.
[0448] • the interpolation (x2K) to fECH of the partial channels thus formed is described in [Fig.27]: this step consists of inserting zeros for all frequencies outside the useful band. Using the same notations as in the part dedicated to beamforming at reception, if n is the current index, the useful frequencies correspond to nM1N < bit_reverse(n,2N) < nMAX with nM1N = m + k 2X K and nMAX = m - 1 + (k+1) 2N K.
[0449] Taking into account all the elements described up to now, still reformulating, a first embodiment of a corresponding method is a method for forming P multi-beam transmission channels from Q beams partially multiplexed on Q' groups of beams indexed by q' (between 1 and Q') for multiplexing rates equal to with Q = V ^Kc<'
[0450] The method comprises, for each group of beams, the steps of:
[0451] - generation of two channels by sum and difference of the signal and of this same signal delayed by M periods,
[0452] - realization of a frequency shift of mr / (M / 2^ j by multiplicative application of a signal e[-jmr / (M / 2^)], n being an integer ranging from 0 to M / p^-l, M / 2^' being an integer strictly greater than 2,- implementation of operations on each of the calculation channels, the operations comprising:
[0453] - the application of a discrete Fourier transform to multiplexed points by 2^v to obtain points in the spectrum, each point of the spectrum corresponding to even indices of a 2* point spectral analysis of the signal from the formed signal and being identified bijectively by an index which is an even number between 0 and 2* M / 2À^-1,
[0454] - applying a band-limiting function to points in the spectrum to get limited points,
[0455] - the formation by duplication of as many replicas as there are emission channels to be formed (if this number is large, this replication can extend over several different computing cards),
[0456] for each of the replicas formed on each of the calculation channels and for each multiplexed beam:
[0457] - applying the delay to help form the emission path associated with the replica,
[0458] - the application of amplitude weighting,
[0459] - the partial formation of the emission path by integration of the 2^« mul- points triplexed and maintained over 9^ periods ([Fig.26]),
[0460] - the application of the interpolation process may be accompanied by a translation in frequency by resampling and insertion of 0 ([Fig.27]), and
[0461] At this stage, there are Q' partially formed channels available in the spectral domain for each channel to be formed.
[0462] The method then comprises the summation of the signals from these Q' partial channels to form a transmission channel available on a first calculation channel for the even frequencies of the spectrum and on a second calculation channel for the odd frequencies of the spectrum.
[0463] For each of the calculation paths of each formed path: - application of frequency correction for equalization of the transmission channels, - application of an IFFT to M points, - on the second calculation path, carrying out a linear transposition of the spectrum,
[0464] The method then comprises a final operation of additive grouping of the two calculation channels to obtain the time signal of the formed channel.
[0465] Conclusion
[0466] The methods described make it possible to carry out, by means of a single device, all of the operations associated with beam formation at reception and multi-beam channel formation at transmission. To do this, they use specific FIR filter architectures which are controlled in the spectrum.
[0467] Thanks to the passage into the frequency domain, these architectures reduce the realization of any delay to a simple multiplication by a complex exponential with linear phase; they make it possible to jointly ensure:
[0468] - the filtering associated with the delay function,
[0469] - at reception, Hilbert filtering intended for I / Q demodulation if the signal input is real and / or the filtering of the useful band associated with the formed beam, at the emission, the interpolation filtering in the case of narrowband beams, and
[0470] - filtering associated with the equalization of the reception channels or the transmission channels.
[0471] In this architecture, a passage is made in the spectrum and all of the filtering operations mentioned are carried out in this domain; this passage in the spectrum is carried out: - at reception, at the level of each reception channel, the return in time being carried out at the level of each beam formed, - on transmission, at the level of each beam, the return in time being carried out at the level of each channel formed.
[0472] When the useful band is smaller than the sampling band, it is possible, thanks to the specific nature of the FFT and IFFT architectures which consists of a decomposition position in successive stages of radix-2, to multiplex several beams occupying the same useful band; this is done naturally via a simple adaptation of this architecture.
[0473] On reception, the output of such a beamformer therefore makes it possible to have either a fECH / 2 or fECH band beam depending on whether the input signal is real or complex (i.e. resulting from an I / Q demodulation process with decimation by 2) or 2K multiplexed fECH / 2K band beams.
[0474] By multiplying the training operations and the output channels, it is also possible to simultaneously have on different outputs a set of beams as described but having different bands on each output n with the relation Bn x Nn = fECH / 2 or fECH depending on the type of input signal, Bn being the band of the beams carried by output n and Nn their number.
[0475] On transmission, this type of architecture is applicable in a dual manner to transmission with a few adaptations associated with the fact that beams are multiplexed on reception and demultiplexed on transmission.
[0476] As this beam and / or channel formation is very wideband, the present methods are suitable for any reception (respectively transmission) system comprising several digital reception chains, the combination of which allows improved listening in one direction (respectively several transmission chains, the combination of which allows simultaneous transmission of several beams in different directions).
Claims
1. Claims Method for forming Q multiplexed beams on Q' groups of beams indexed by q', q' being an integer varying between 1 and Q', Q' being an integer greater than 1, and the multiplexing rate being equal to 2^*, being equal to a respective integer for each value of q', the method being implemented by a computer of a beam forming system comprising P reception channels each suitable for receiving signals coming from a respective antenna, each antenna having a respective aperture, Q and P being two integers greater than or equal to 2, the method comprising: - the passage in the spectral domain of signals from each antenna, - the implementation of a set of operations in the spectral domain, called depointing operations, the set of depointing operations comprising: - for each reception channel, a generation operation by duplication of a contribution to each of the Q' groups of beams to be generated, - for each contribution, a decimation operation by 9¾ and a maintenance over 2^ periods to constitute a group of ^^ / contributions multiplexed by - the implementation for each group of contributions of an introduction operation multiplexed by 2^ of a set of 2^' delays, and of a possible set of 9^ / weighting operations of the opening of the antenna, and - an operation of summing the contributions to a beam to be generated, from all the reception channels, multiplexed for each channel on Q' groups of contributions, the summing operation being carried out for each beam to be generated, the method comprising, for each reception channel, the formation of two calculation channels, the input signal of the first calculation channel being obtained by implementing the shift of the signal received by the antenna of M points, M being an integer greater than or equal to 2, to obtain a shifted signal and the calculation of the sum of the received signal and the shifted signal, and the input signal of the second calculation channel being obtained by putting implementing the shift of the received signal by M points to obtain a shifted signal and calculating the difference between the received signal and the shifted signal.
2. The training method of claim 1, wherein the operations implemented in the spectral domain further comprise a channel equalization operation, the equalization operation being implemented before the depointing operations.
3. A training method according to claim 1 or 2, wherein the operations implemented in the spectral domain further comprise a band-limiting operation, the band-limiting operation being implemented after the depointing operations.
4. Training method according to any one of claims 1 to 3, in which the passage into the spectral domain is carried out by applying a discrete Fourier transform to M points on each of the calculation channels.
5. A training method according to any one of claims 1 to 4, wherein the method comprises a step of returning to the time domain, the step of returning to the time domain comprising the application of inverse discrete Fourier transforms at M / 2^ points multiplexed by ^1' on each of the calculation channels, the method comprising a spectral shift operation of the second calculation channel by multiplicative application of a signal e[+j floor(n / 2^' jjt / ( M / 9Aî')], e designating the exponential function and j designating the complex number such that j2=-l, n being an integer ranging from 0 to Ml and the floor function designating the integer part function, and an operation of summing the two outputs of the two calculation channels.
6. Computer, in particular programmable logic circuit, adapted to implement the operations of a training method according to any one of claims 1 to 5.
7. System for forming Q beams multiplexed on Q' groups of beams indexed by q', q' being an integer varying between 1 and Q', Q' being an integer greater than 2, Q' being an integer greater than 1, and the multiplexing rate being equal to , Kq' being equal to a respective integer for each value of q', the forming system comprising P reception channels each suitable for receiving signals coming from a respective antenna, Q and P being two integers greater than or equal to 2, and the forming system comprising a calculator according to claim 6.
8. A computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the operations of a method according to any one of claims 1 to 5.
9. A computer-readable medium comprising instructions which, when executed by a computer, cause the computer to carry out the operations of a method according to any one of claims 1 to 5.
