Method for searching for a useful signal in a multiplexing band

Abstract

A method of searching for the presence of a useful signal of predefined spectral width ΔB in a multiplexing band having a spectral width greater than ΔB, includes calculating a frequency signal representative of a power spectral density in the multiplexing band, calculating a non-useful signal by filtering the frequency signal by means of a filter capable of suppressing all or part of signals having a spectral width equal to or smaller than ΔB, calculating a power ratio signal representative of the ratio of the frequency signal to the non-useful signal, and comparing the power ratio signal with a predefined threshold value. The method may be employed using a computer program product, a receiver unit, and/or a station of a digital telecommunications system comprising such a receiver unit.

Claims

1. A method of searching for a useful signal in a general signal received by a receiver station, said useful signal corresponding to a radioelectric signal having a predefined spectral width ΔB transmitted by a terminal in a multiplexing band having a spectral width greater than ΔB, said general signal corresponding to all the radioelectric signals received in the multiplexing band, the method comprising: generating, from the general signal received by the receiver station, a frequency signal representative of a power spectral density of the general signal sampled at elementary frequencies of the multiplexing band; generating a non-useful signal, the generating the non-useful signal including filtering of the frequency signal using a filter to suppress from said frequency signal all or part of each signal having a spectral width equal to or smaller than ΔB; generating a power ratio signal, representative of a ratio, at each elementary frequency, of the frequency signal to the non-useful signal; comparing the power ratio signal with a defined threshold value; and identifying an elementary frequency for which said power ratio signal has a local maximum value greater than said threshold value as a central frequency of a useful signal present in the general signal.

2. The method of claim 1 wherein the elementary frequencies are separated by a sampling interval Δb smaller than ΔB, and said method comprises: generating an averaged signal by averaging the frequency signal with a sliding window on the elementary frequencies, the sliding window having a width substantially equal to ΔB; filtering of the averaged signal to generate the non-useful signal; and the generating the power ratio signal includes determining, at each elementary frequency, a ratio of the averaged signal to the non-useful signal.

3. The method of claim 2 wherein the sliding window is a rectangular window.

4. The method of claim 1, comprising selecting the defined threshold value to correspond to a value of a power ratio signal above which a probability to decode a useful signal with errors is lower than 5%.

5. The method of claim 1 wherein the elementary frequencies are separated by a sampling interval Δb smaller than ΔB/4.

6. A non-transitory computer-readable memory medium whose contents contain instructions which when executed by at least one processor of a receiver station cause the at least one processor of the receiver station to perform a method, the method comprising: generating, from a general signal received by the receiver station, a frequency signal representative of a power spectral density of the general signal sampled at elementary frequencies of a multiplexing band, the multiplexing band having a bandwidth greater than a spectral width ΔB of signals transmitted in the multiplexing band; generating a non-useful signal, the generating the non-useful signal including filtering of the frequency signal to suppress from said frequency signal all or part of each signal having a spectral width equal to or smaller than ΔB; generating a power ratio signal, representative of a ratio, at each elementary frequency, of the frequency signal to the non-useful signal; comparing the power ratio signal with a defined threshold value; and identifying an elementary frequency for which said power ratio signal has a local maximum value greater than said threshold value as a central frequency of a useful signal present in the general signal.

7. A receiver, comprising: one or more memories; one or more signal processors, wherein the one or more signal processors of the receiver, in operation: generate, from a general signal received by the receiver, a frequency signal representative of a power spectral density of a general signal sampled at elementary frequencies of a multiplexing band, the multiplexing band having a bandwidth greater than a spectral width ΔB of signals transmitted in the multiplexing band; generate a non-useful signal, the generating of the non-useful signal including filtering of the frequency signal to suppress from said frequency signal all or part of each signal having a spectral width equal to or smaller than ΔB; generate a power ratio signal, representative of a ratio, at each elementary frequency, of the frequency signal to the non-useful signal; compare the power ratio signal with a defined threshold value; and identify an elementary frequency for which said power ratio signal has a local maximum value greater than said threshold value as a central frequency of a useful signal present in the general signal.

8. A station of a digital telecommunications system, comprising: an antenna system; and signal processing circuitry of the station coupled to the antenna system, wherein the signal processing circuitry of the station, in operation: generates, from a general signal received by the station, a frequency signal representative of a power spectral density of the general signal sampled at elementary frequencies of a multiplexing band, the multiplexing band having a bandwidth greater than a spectral width ΔB of signals transmitted in the multiplexing band; generates a non-useful signal, the generating of the non-useful signal including filtering of the frequency signal to suppress from said frequency signal all or part of each signal having a spectral width equal to or smaller than ΔB; generates a power ratio signal, representative of a ratio, at each elementary frequency, of the frequency signal to the non-useful signal; compares the power ratio signal with a defined threshold value; and identifies an elementary frequency for which said power ratio signal has a local maximum value greater than said threshold value as a central frequency of a useful signal present in the general signal.

Description

DESCRIPTION OF THE DRAWINGS

(1) The disclosure will be better understood on reading of the following description, given as a non-limiting example, in relation with the drawings, which show:

(2) FIG. 1: a simplified representation of a digital telecommunications system,

(3) FIG. 2: a diagram illustrating an embodiment of a useful signal search method,

(4) FIGS. 3a to 3d: examples of frequency signals obtained from the search method of FIG. 2.

(5) In these drawings, identical references from one drawing to another designate identical or similar elements. For clarity, the shown elements are not to scale, unless otherwise mentioned.

DETAILED DISCUSSION OF EMBODIMENTS

(6) FIG. 1 very schematically shows a telecommunications system 10 comprising a plurality of terminals 20 and a station 30.

(7) Station generally means any receiver device capable of receiving radioelectric signals. Station 30 is for example any of terminals 20, or a specific device such as an access point of a wired or wireless telecommunications network, centralizing the data transmitted by each of said terminals 20.

(8) Radioelectric signal means an electromagnetic wave propagating via wireless means, having frequencies in the conventional spectrum of radioelectric waves (from a few hertz to several hundreds of gigahertz) or in neighboring frequency bands.

(9) It should be noted that the case of a data transmission from terminals 20 to station 30 is mainly considered. The possible transmission of data from station 30 to terminals 20 is not comprised within discussed herein.

(10) Terminals 20 transmit useful signals to station 30 in a shared frequency band, called “multiplexing band”, having a spectral width ΔM.

(11) The useful signals transmitted by terminals 20 have a predefined spectral width ΔB smaller than ΔM. In an embodiment, spectral width ΔM of the multiplexing band is significantly greater than spectral width ΔB of the useful signals, to facilitate multiplexing a large number of terminals 20. For example, in an embodiment spectral width ΔM is at least one hundred times greater than spectral width ΔB.

(12) Spectral widths ΔB and ΔM are both known beforehand by station 30. However, the central frequency around which a terminal 20 transmits a useful signal is not necessarily known beforehand by station 30. This is for example true, as previously indicated, when the frequency drift of the useful signals is greater than spectral width ΔB of said useful signals (see international application WO 2011/54466).

(13) The following description considers, without this being a limitation, a digital telecommunications system such as described in international application WO 2011/54466, where the useful signals have a narrow band (spectral width ΔB in the range from a few hertz to a few hundreds of hertz) and the frequency drift of said useful signals is greater than spectral width ΔB.

(14) It should further be noted that spectral width ΔB of the useful signals corresponds to the instantaneous spectral width of said useful signals. It should indeed be understood that the frequencies successively taken over time by a useful signal should correspond to a frequency band having a width greater than ΔB, particularly due to the frequency drift of said useful signal.

(15) Accordingly, station 30 should search for the presence of useful signal in the entire multiplexing band, jointly with a search for the central frequencies of the useful signals present in said multiplexing band.

(16) For this purpose, station 30 particularly comprises a receiver unit capable of receiving a general signal corresponding to all the radioelectric signals received in the multiplexing band.

(17) The receiver unit comprises an analog receiver module and a digital receiver module.

(18) The analog receiver module comprises circuitry and/or devices, considered as known by those skilled in the art (antennas, analog filters, amplifiers, local oscillators, mixers, etc.), capable of shifting the frequency of the general signal.

(19) The analog receiver unit outputs an analog signal corresponding to the general signal shifted around an intermediate frequency lower than the central frequency of the multiplexing band, and which may be zero (in which case two analog signals are provided, corresponding in known fashion to the paths in phase I and in quadrature Q).

(20) The digital receiver module comprises, in known fashion, one or a plurality of analog-to-digital converters (AD) capable of sampling the analog signal(s) delivered by the analog receiver module, with a predefined sampling period, to obtain a digital signal St representative of the analog signal(s).

(21) The digital receiver module further comprises a unit for processing the digital signal at the output of the A/D converters. The processing unit particularly executes, based on the digital signal, a method 50 of searching for the presence of a useful signal in the general signal, described in further detail hereafter.

(22) The processing unit for example comprises a processor and an electronic memory having a computer program product stored therein, in the form of a set of program code instructions which, when they are executed by the processor, implement all or part of the method 50 of searching for the presence of a useful signal. In an embodiment, the processing unit comprises programmable logic circuits, of FPGA, PLD, etc., type and/or application-specific integrated circuits (ASICs), capable of implementing all or part of said search method 50.

(23) FIG. 2 shows an embodiment of a method 50 of searching for the presence of a useful signal transmitted by a terminal 20 in the multiplexing band, which comprises: 51 calculating a frequency signal Sf, 54 calculating a signal, called “non-useful signal” Snu, 56 calculating a signal, called “power ratio signal” Srp, 57 comparing power ratio signal Srp with a predefined threshold value.

(24) In the embodiment illustrated in FIG. 2, search method 50 also comprises, in addition to the above acts, optional acts which will be described hereafter.

(25) Frequency signal Sf calculated during calculation 51 is representative of a power spectral density of the general signal, said power spectral density being called “PSD” hereafter.

(26) More particularly, frequency signal Sf is formed of N samples Sf(n) (0≦n≦N−1) representative of the values taken by power spectral density PSD of the general signal at elementary frequencies fn (0≦n≦N−1) regularly distributed in the multiplexing band with a sampling interval Δb assumed to be such that N.Math.Δb=ΔM. In other words: Sf={Sf(n), with 0≦n≦N−1}, Sf(n)˜PSD(fn), “˜” meaning “representative of”, fn=f0+n.Math.Δb, with 0≦n≦N−1.

(27) To accurately estimate the central frequency of a possible useful signal present in the general signal, sampling interval Δb is further selected to be smaller than spectral width ΔB of the useful signals. In some embodiments, sampling interval Δb is much smaller than spectral width ΔB, for example, smaller than ΔB4.

(28) FIG. 2 shows an embodiment of act 51 of calculating frequency signal Sf. In this embodiment, act 51 of calculating frequency signal Sf comprises: 510 calculating a fast Fourier transform (FFT) of digital signal St, which is a signal defined in the time domain, 511 calculating frequency signal Sf as being the modulus, at each elementary frequency, of the result of the fast Fourier transform.

(29) The power spectral density of digital signal St (representative of the general signal) is known to be obtained, at each elementary frequency, by calculating the square of the modulus of the result of the fast Fourier transform. Thus, frequency signal Sf may be linked to power spectral density PSD of the general signal by the following expression:
Sf(n)=√{square root over (PSD(fn))}

(30) FIG. 3a shows, in logarithmic scale, an example of power spectral density PSD of a general signal shifted to the baseband. In other words, FIG. 3a shows the variation of function 20.Math.log(Sf(n)) for elementary frequencies distributed in a frequency interval [−Δ/M2; ΔM/2].

(31) It should be noted that power spectral density PSD, in FIG. 3a, comprises a line centered on the zero frequency, which corresponds to an unwanted DC component, called “DC” in said FIG. 3a. It is indeed known by those skilled in the art that the zero frequency is generally very disturbed, whereby it is most often ignored in the search for the presence of useful signals due to the fact that: the unwanted DC component risks leading to a false detection (that is, to erroneously considering that a useful signal, centered on the zero frequency, is present), even though a useful signal would be present and centered on the zero frequency, the probability of decoding it with no errors is low.

(32) In the embodiment illustrated in FIG. 2, search method 50 comprises calculating at act 52 a signal, called “averaged signal” Sm. Averaged signal Sm calculated during act 52 is obtained by averaging of frequency signal Sf with a sliding window on elementary frequencies fn (0≦n≦N−1), said sliding window having a width substantially equal to spectral width ΔB of the useful signals.

(33) This averaging of frequency signal Sf aims at concentrating the power, in each band substantially having a spectral width ΔB, around the elementary frequency substantially located at the center of the band substantially of frequency width ΔB. Thereby, in the presence of a useful signal, averaged signal Sm should theoretically comprise a local maximum value at the level of the elementary frequency closest to the real central frequency of said useful signal, while this is not necessarily true in frequency signal Sf.

(34) Accordingly, averaged signal Sm is representative, at each elementary frequency, of the entire power received in a band of frequency width ΔB centered on this elementary frequency, which corresponds to the power of the non-useful signal in this band, possibly cumulated with the power of a useful signal present in this band.

(35) In an embodiment, act 52 of calculating averaged signal Sm for example comprises calculating the following expression:

(36) $\mathrm{Sm}\left(n\right)=\underset{m=-M/2}{\overset{M/2}{.Math.}}h\left(m\right).Math.\mathrm{Sf}\left(n+m\right)$

(37) where: h={h(m), with −M/2≦m≦M/2} is the sliding window used for the averaging, M being an even integer, Sf(n+m)=Sf(n+m) if 0≦n+m≦N−1 Sf(n+m)=0 if n+m<0 or if n+m>N−1.

(38) To have an averaging window of frequency width substantially equal to spectral width ΔB, M is for example selected to be such that (M−2).Math.Δb≦ΔB≦(M+2)Δb.

(39) In a specific embodiment, the sliding window used is a simple rectangular window. In other words, h(m)=1/(M+1) with M/2≦m≦M/2. Nothing precludes, according to other examples, considering other types of sliding windows, possibly representative of weighted averages.

(40) In the specific embodiment illustrated in FIG. 2, search method 50 comprises act 53 during which averaged signal Sm is converted to the logarithmic scale, by calculating a signal S log:
S log(n)=20.Math.log(Sm(n)), with 0≦n≦N−1

(41) Passing to a logarithmic scale is advantageous to enable to more simply calculate power ratio signal Srp.

(42) FIG. 3b shows a signal S log obtained from the signal illustrated in FIG. 3a. As illustrated in FIG. 3b, the averaging shows local maximum values R1, R2, R3 capable of corresponding to useful signals, by a concentration of the power of the useful signals around their central frequency, but also by a decrease in the fluctuations of a power spectral density of noise present in frequency signal Sf.

(43) In a specific embodiment illustrated in FIG. 2, signal S log originating from averaged signal Sm and representative of power spectral density DSP of the general signal is then delivered to two different processing branches: a first branch where non-useful signal Snu is calculated from signal S log (calculation act 54) a second branch where a signal, called “shifted signal” Sd, is calculated 55 from signal S log.

(44) In the first branch, non-useful signal Snu is calculated by filtering of signal S log using a filter capable of suppressing all or part of signals having a spectral width equal to or smaller than ΔB.

(45) Thus, act 54 of calculating non-useful signal Snu performs a low-pass filtering of the frequency variations of signal S log in the multiplexing band. Indeed, the “fast” variations of signal S log, which correspond to signals which occupy a narrow frequency band, will be more attenuated than “slow” frequency variations of signal S log, which correspond to signals which occupy a wide frequency band.

(46) The filter used during act 54 of calculating non-useful signal Snu is capable of strongly attenuating signals which occupy a frequency band equal to or smaller than ΔB, as compared with the signals occupying a frequency band greater than ΔB. The design of such a filter is considered as being within the abilities of those skilled in the art.

(47) Due to the use of a filter having the above characteristics, it should be understood that, in the obtained non-useful signal Snu: the useful signals possibly present in the general signal (and thus in signal S log) will be strongly attenuated, the noise, of substantially constant spectral density in the entire multiplexing band, will be lightly attenuated.

(48) Accordingly, the filtering of signal S log enables to obtain a non-useful signal Snu representative, at each elementary frequency, of the power received in a band of frequency width ΔB centered on this elementary frequency, the signals of spectral width equal to or smaller than ΔB (and accordingly all the useful signals possibly present in the general signal) having been previously suppressed. Non-useful signal Snu is thus representative of noise/interference present in the multiplexing band.

(49) Accordingly, the filtering of signal S log provides a non-useful signal Snu where the useful signals have been substantially suppressed.

(50) FIG. 3c shows a non-useful signal Snu obtained from signal S log illustrated in FIG. 3b. It can be observed that, due to the filtering, local maximum values R1, R2, and R3 have been strongly attenuated, while an average level of signal S log in the multiplexing band has been lightly attenuated.

(51) In the second branch, search method 50 comprises act 55 during which signal S log, originating from the averaged signal, is frequency-shifted, to obtain a shifted signal Sd having its frequency realigned with non-useful signal Snu.

(52) Indeed, non-useful signal Snu exhibits, in known fashion, a frequency shift which is introduced by the frequency filtering and which depends on the order of the implemented frequency filter. The determination of the introduced frequency shift and its compensation may be performed in a simple manner, by operations considered as being within the abilities of those skilled in the art.

(53) Useful signal search method 50 then comprises act 56 of calculating power ratio signal Srp, said power ratio signal being representative, at each elementary frequency, of a ratio of the averaged signal Sm obtained for this elementary frequency to the non-useful signal Snu obtained for this elementary frequency.

(54) In the example illustrated in FIG. 2, the power ratio signal Srp is calculated from non-useful signal Snu and shifted signal Sd, which essentially corresponds to averaged signal Sm.

(55) Due to the fact that non-useful signal Snu and shifted signal Sd are both in logarithmic scale, power ratio signal Srp is calculated by simply subtraction, at each elementary frequency, of non-useful signal Snu from shifted signal Sd:
Srp(n)=Sd(n)−Snu(n), with 0≦n≦N−1

(56) FIG. 3d shows a power ratio signal Srp obtained from the signals illustrated in FIGS. 3b and 3c.

(57) Useful signal search method 50 then comprises act 57 of comparing power ratio signal Srp with a predefined threshold value Vmin. During this act, an elementary frequency for which said power ratio signal Srp has a local maximum value greater than said threshold value Vmin is considered as being the central frequency of a useful signal present in the general signal.

(58) In the example illustrated in FIG. 3d (where the unwanted DC component has been ignored), the considered threshold value Vmin is substantially equal to 8 decibels (dB) and it can be observed that local maximum values R1, R2, and R3 are all three greater than said threshold value Vmin. Accordingly, these three local maximum values R1, R2, and R3 are considered as corresponding to useful signals, and the elementary frequencies for which these local maximum values have been obtained are considered as being the respective central frequencies of these useful signals.

(59) In an embodiment, threshold value Vmin is previously determined as being a value of power ratio signal Srp above which the probability of decoding a detected useful signal with errors is lower than 5%, in an embodiment lower than 1%.

(60) Such a selection of threshold value Vmin may be advantageous since facilitates optimizes the use of a calculation capacity of station 30, since said calculation capacity may be mainly used to process useful signals which are sufficiently powerful to be decoded.

(61) More generally, the scope of the present disclosure is not limited to the embodiments described hereabove as non-limiting examples.

(62) In particular, nothing precludes, according to other examples, having act 53 of conversion to the logarithmic scale executed at a different location in search method 50, for example, executed on the one hand on non-useful signal Snu and, on the other hand, on shifted signal Sd, just before act 56 of calculating power ratio signal Srp.

(63) Further, search method 50, in the above-described example, comprises two processing branches which take signal S log as an input. These two processing branches could however take other signals as an input. According to a non-limiting example, the two branches may take frequency signal Sf as an input. In this case, the first branch (obtaining of non-useful signal Snu) may perform the averaging of act 52 and the filtering of act 54, and possibly a conversion to the logarithmic scale. The second branch then performs the averaging of act 52 and possibly a conversion to the logarithmic scale. In this example, each processing branch thus comprises the averaging of act 52. In the first branch, the averaging of act 52 and the filtering of act 54 may further be performed jointly with a single filter.