WO2008152322A2

WO2008152322A2 - Method and electronic device for frequency shifting an analogue signal, in particular for mobile telephony

Info

Publication number: WO2008152322A2
Application number: PCT/FR2008/050973
Authority: WO
Inventors: François RIVET; Didier Belot; Yann Deval; Jean-Baptiste Begueret; Dominique Dallet
Original assignee: Centre National De La Recherche Scientifique (Cnrs)
Priority date: 2007-06-04
Filing date: 2008-06-03
Publication date: 2008-12-18
Also published as: WO2008152322A3; FR2916916A1; FR2916916B1

Abstract

Electronic device for frequency shifting, comprising an input for receiving an analogue input signal having a first frequency, sampling means (MECH) able to deliver analogue samples of the input signal, first processing means (MT1) able to perform a discrete Fourier transform processing on the samples of the signal and to deliver first intermediate samples, second processing means (MT2) able to modify the spectral distribution of the first intermediate samples and to deliver second intermediate samples, and third processing means (MT3) able to perform an inverse discrete Fourier transform processing on the second intermediate samples and to deliver analogue samples of an output signal having a second frequency lower than the first frequency.

Description

Method and electronic device for frequency shifting of an analog signal, in particular for mobile telephony.

The present invention relates to signal processing, and more particularly to the processing of a radio frequency signal to lower its frequency.

The invention applies advantageously but not exclusively to mobile telephony devices.

The development of a device capable of processing the information contained in a radio frequency signal in a totally digital manner is a subject of current interest in the telecommunications sector. The goal is to minimize the number of components and thus bring together the digital processing means, usually a digital signal processing processor (DSP), of the radio frequency antenna to minimize costs and consumption. However, carrier demodulation and signal processing operations, such as signal additions and multiplications, can quickly saturate the processing capabilities of current digital signal processing processors when considering a radio frequency signal and a signal. a significant amount of data to process.

According to one aspect, there is provided an electronic frequency shift device, comprising an input for receiving an analog input signal having a first frequency, sampling means capable of delivering analog samples of the input signal, first means processors capable of performing a discrete Fourier transform processing on the samples of the signal and of delivering first intermediate samples, second processing means able to effect a modification of the spectral distribution of the first intermediate samples and to deliver second intermediate samples , and third processing means capable of performing inverse discrete Fourier transform processing on the second intermediate samples and outputting analog samples of an output signal having a second frequency lower than the first frequency.

In other words, according to this aspect, an analog signal pretreatment is carried out which makes it possible to lower the working frequency in order to be able to return to favorable conditions for a numerical approach by the use of a digital processing processor. signal.

Such a solution confers a speed and power consumption savings, particularly interesting in mobile phone applications.

This preprocessing example is based here on samples that are not bits but analog samples such as voltage samples, which makes it possible to implement a discrete Fourier transform. The second processing means may be able to perform for example a frequency translation on the first intermediate samples, for example by an operation on the indices of these samples.

The discrete and inverse discrete Fourier transform processes may comprise butterfly-type elementary treatments corresponding to several stages of a generally butterfly-shaped calculation graph, and the first and third processing means may have a multi-stage pipelined architecture. elementals respectively corresponding to the stages of the graph of calculation.

The use of a pipelined Fourier transform uses fewer resources for its implementation than the use of a traditional Fourier transform and with an identical result.

According to one embodiment, the first and third processing means are able to process successive blocks of N samples, and said elementary stages are respectively able to carry out Fourier transform processes on successively reduced sample size blocks. from one elementary stage to the next, the number N and the number of stages elementals being connected to each other via the radix of the Fourier transform.

According to one embodiment, each elementary stage comprises elementary input means, elementary output means, calculation means able to carry out an elementary processing of a Fourier transform of the radix butterfly type, storage means capable of storing a number of samples dependent on N and the rank of the elementary stage in the pipelined architecture, first controllable switching means connected between on the one hand the elementary input means and on the other hand the storage means and the calculation means, second controllable switching means connected between on the one hand the storage means and the calculation means and on the other hand the elementary output means, and control means adapted during a first phase to be controlled the first and second switching means so as to connect the storage means between the elementary input means and the means of elementary elements, and during a second phase to connect the storage means to the calculation means and the calculation means between the elementary input and output means so as to allow the calculation of the samples, the delivery of part of these output samples and storage of another part of these samples calculated in the storage means.

The sampling means are advantageously capable of delivering samples representing the level of the input analog signal at each sampling instant.

According to one embodiment, the sampling means comprise controllable interruption means and the storage means comprise a delay line including a chain of controllable capacitors and switches able to store and redeliver successive samples.

In another aspect, there is provided an apparatus, for example a cellular mobile telephone, having a reception chain including an analog part and a digital part having a signal processing processor, the analog part incorporating a frequency shift device such as defined above.

In another aspect, there is provided a frequency offset method of an analog signal comprising receiving an analog input signal having a first frequency, sampling the input analog signal to obtain analog samples of the input signal. input signal, a transform processing of

Discrete Fourier on the signal samples so as to deliver first intermediate samples, a modification of the spectral distribution of the first intermediate samples so as to deliver second intermediate samples, and an inverse discrete Fourier transform processing on the second intermediate samples of for outputting analog samples of an output signal having a second frequency lower than the first frequency.

Said modification of the spectral distribution of the first intermediate samples may comprise a frequency translation on the first intermediate samples.

Other objects, features and advantages of the invention will appear on reading the following description, given solely by way of nonlimiting example and with reference to the appended drawings in which:

FIG. 1 illustrates an example of a radio frequency apparatus incorporating a frequency shift device; FIG. 2 the main steps of a mode of implementation of a frequency shift method

FIG. 3 illustrates a radix-type butterfly structure. FIG. 4 shows an example of a three-stage Cooley-Tukey and radix 2 calculation graph. FIG. 5 shows the main elements of an embodiment of a frequency shift device;

FIG. 6 shows the main elements of an embodiment of an elementary calculation stage; FIGS. 7 and 8 illustrate an operation of an elementary calculation stage;

- Figures 9 to 12 show an example of frequency shift on a particular signal.

In Figure 1, we can see an example of an apparatus, for example a cellular mobile phone TP, adapted to receive on its antenna 1 a radio frequency signal.

The telephone comprises a reception chain having an analog part connected to a digital part comprising for example a processor 1 1 signal processing by an ADC digital analog converter.

The analog part of the reception chain comprises in this example conventionally a bandpass filter BPF followed by a low noise amplifier LNA.

Between the low noise amplifier LNA and the ADC digital analog converter is connected a frequency shift device DCF.

The radio frequency signal delivered to the input BE of the DCF device is processed by the DCF device in order to reduce the amount of operations to be performed by the digital signal processing means 11.

FIG. 2 represents the main steps of an implementation mode of a frequency shift method of an analog signal. The process begins with receiving an analog input signal at the BE input (step 30). In step 31, a direct Fourier transform conventionally referred to as FFT is performed on the analog input signal. In step 32, a change in the spectral distribution of the samples from the FFT, for example an offset on the indices of the samples so as to carry out a frequency translation is carried out. At step 33, a transform of Fourier inverse conventionally called IFFT is carried out in order to obtain an analog output signal. The frequency of the output signal due to the expected spectral distribution change is less than the frequency of the input analog signal. In step 34, the output analog signal is output.

Many methods can be used to perform the direct or inverse Fourier transformations. Most of these methods use a variant of the Cooley-Tukey algorithm, well known to those skilled in the art, which makes it possible to reduce the number of arithmetic operations necessary for calculating the Fourier transform. Thus, the calculation of a fast Fourier transform of initial size r ^p where r represents the "radix", according to a denomination commonly used by those skilled in the art, can be reduced to that of r Fourier transformants of size r ^{p -1} and additional complex additions and multiplications. By repeating this reduction, we arrive at the computation of Fourier transforms of size r which are easily realizable, in particular if r is chosen equal to 2 or 4.

The Cooley-Tukey algorithm uses a calculation graph showing a butterfly-shaped structure commonly referred to in the English language as "butterfly".

A first solution consists in producing a hardware operator capable of performing a butterfly type calculation by means of a butterfly of the graph. However, this solution is conceivable only for the implementation of small Fourier transforms. A second solution consists in producing a throttle type operator for each stage of the graph, as well as a storage element, for example delay lines or shift registers, the function of which is to present the input of the operator with the data in the right order considering the butterflies of the graph of the stage considered. Such architectures are called series or pipelines.

The Fourier transform of a signal x (t) is defined as follows continuously:

X (f) =

- e- ^j2πβ dt When the signal is discretized, for example by sampling, the Fourier transform becomes discrete (DFT):

N M = O

1 NI 1 NI so it X (k) = - ^ ^x ( ⁿ ) 'cos (2πnk) + j - ^ ^x ( ⁿ )' sin (2πnA :)

with x (n) the sampled input signal

N the number of samples taken into account

Figure 3 shows a butterfly structure of radix equal to 2. For a pair of input samples (a; b), the sample pair (A; B) at the output is: A = a + W ^Z _N * b

B = a - W ^Z _N * b where W ^Z _N are complex coefficients well known to those skilled in the art and whose values depend on a function g of k and N. More precisely, for a direct Fourier transform we have: W ^Z _N = e ^{"2j * π * g (k} ' ^N)

for an inverse Fourier transform, we have:

W ^Z _N = _e ^{+ 2j * π * g (k} ' ^N)

Figure 4 shows an example of a calculation graph according to the Cooley-Tukey algorithm. We have a pipelined structure with three radix 2 stages. In general, N = r ^v or r denotes the radix of the Fourier transform and v the number of treatment stages.

Materially as illustrated in FIG. 5, the device DCF comprises, between the input BE and the output BS, sampling means MECH, followed by the first processing means.

MT 1, second processing means MT 2 and third processing means MT 3 successively connected in series.

The sampling means (MECH) delivers analog samples of the input signal.

The first processing means (MT 1) perform discrete Fourier transform processing on the signal samples and deliver first intermediate samples. The second processing means (MT2) perform a modification of the spectral distribution of the first intermediate samples and to deliver second intermediate samples.

The third processing means (MT3) performs inverse discrete Fourier transform processing on the second intermediate samples and outputs analog samples of an output signal having a second frequency different from the first frequency.

The sampling means MECH schematically here comprise a switch controlled at the rhythm of a clock signal CLK and for delivering successive samples of voltage of the input signal.

FIG. 6 schematically illustrates an exemplary embodiment of an ETL processing elementary stage, MT 1 or MT 3 means. The structure of these means is identical. Only the values of the coefficients W ^Z N change according to whether it is necessary to carry out a direct or inverse Fourier transform processing.

It will be recalled that there is an elementary stage of treatment per stage of the graph of calculation of the Fourier transform. Each elementary stage uses only the data provided by the previous stage.

The input 6 of the basic processing stage ETL is connected to a first switching device 20 connected on the one hand to a delay line 22 by the connection 21 and on the other hand to a computing unit 27 via a connection incoming line 26 and an outgoing connection 25. The delay line 22 is connected to the second switching device 24 via the connection 23. The calculation unit 27 is also connected to the second switching device 24 via an outgoing link 29 and an incoming link 28. The second switching device 24 is connected to the output 8 of the ETL processing elementary stage. A control means 12 that can be produced, for example, from gates and / or logic circuits, receives the clock signal CLK and controls the first switching means 20, the second switching means 24, and the various storage elements of the line of 22. The different storage elements of the delay line 22 may be formed for example of a switch and a storage capacitor. A connection 16, connected to the calculation unit 27, makes it possible to supply the values of the weighting coefficients W ^Z N to the calculation unit 27.

Of course, the different connections are in fact "double" to convey in parallel the real and imaginary parts of the samples.

The computing unit receives samples X _1n and y _in as input and outputs samples

W ^Z N and y _out = x _in -y _m *

W ^Z _N.

This structure can be achieved using complex adders and multipliers.

The depth or size of the delay line depends on the number N and the rank of the elementary stage of processing in the graph. This size decreases from one floor to another. Thus, in the example of the Fourier transform of radix 2 illustrated in FIG. 6, the delay line of the first stage stores N / 2 samples, that of the second stage N / 4 and that of the third stage N / 8 samples. The coefficient W ^Z N at the same periodicity as the displacement interval (N / 2, N / 4, ..).

The first stage will process successive groups of N samples, the second stage successive groups of N / 2 samples and the third stage successive groups of N / 4 samples.

Two phases are to be distinguished during a Tcaicui calculation period dealing with N samples:

1) the receipt and storage of the N / 2 samples input during _calculation T / 2, 2) the calculation with the N / 2 samples following and the N / 2 samples stored in the delay line during T _ca icui / 2

Figure 7 shows the first phase of operation of the device. The control means 12 controls the switching means 20 and 24 so that they respectively connect the connections 6 and 21, and 23 and 8. During this phase, N / 2 samples are stored in the delay line 22. Simultaneously, the delay line 22 emits by the elementary output 8, N / 2 samples calculated during the second phase previous and which had been stored in this delay line. The flow of data is symbolized by the dashed arrows.

Figure 8 shows the second phase of operation of the device. The control means 12 controls the switching means 20 and 24 in order to modify the order of circulation of the data. The first switching means 20 connects the elementary input 6 with the incoming connection 26 and the outgoing connection 25 with the delay line 22 via the connection 21. Likewise, the switching means 24 connects the delay line 22 via the connection 23 at the second incoming connection 28 and the second outgoing connection 29 at the elementary output 8. The N / 2 samples stored during the first phase in the delay line 22 are processed by the calculation unit 27 via the connections 23, 28 and 29.

Simultaneously, the N / 2 remaining samples of the block of N samples are admitted by the elementary input 6 and processed by the calculation unit 27.

N / 2 calculated samples are directly output 8 and N / 2 calculated samples are stored in the delay line 22 to be output 8 in the next first phase. Thus, at the output of means MT 1 or MT 3, the N output samples are delivered according to a distribution in reverse order with respect to the input order, this inverted distribution being well known to those skilled in the art.

The above description was made for a calculation based on radix 2 butterfly type Fourier transforms. Another size of radix as well as another number of N samples could have been used.

Thus for an application to a radiofrequency signal, and to obtain a frequency resolution of one megahertz, it is possible to use a Fourier transform which processes N = 4096 samples. This is possible by switching to a higher radix. The number of samples is always N = radix ^v , v being the number of stages. For a radix 8, we would then always have four floors processing 4096 samples.

Figures 9 to 12 show an example of frequency shift for a sinusoidal signal over three periods (Figure 9) with sixteen samples (N = 16). After direct Fourier transformation, the sixteen samples X (O) ... X (15) are transformed into the first six intermediate samples Xl (O) ... Xl (15) according to the inverse order distribution (0 8 4 12 2 10 6 14 1 9 5 13 3 1 1 7 15) (Figure 10). A frequency shift operation is performed by performing an operation on the indices i of the first intermediate samples.

Thus, for example, one translates the information by shifting the indices of two. The frequency distribution of FIG. 11 is then obtained. The line at the third index in FIG. 10 is then shifted to the first index in FIG. 1. After inverse Fourier transformation, the output signal is obtained. Said signal (Figure 12) has the same amplitude as the input signal but a reduced frequency (a single period instead of three).

The second intermediate means MT2 which make it possible to carry out the frequency shift can be made from logic circuits and delay lines.

The invention is not limited to the embodiments and embodiments described above. Indeed, it is possible to use the same means to achieve an increase in the frequency between the input signal and the output signal. The second means MT2 can thus generally implement a frequency shift to obtain an output signal having a different frequency, higher or lower, the frequency of the input signal.

Such a solution thus makes it possible to increase the frequency of the signal in a transmission channel in a mobile telephone application to switch to a radio frequency frequency, and constitutes an alternative to a solution providing for the use of a mixer.

Claims

An electronic frequency shift device, comprising an input for receiving an input analog signal having a first frequency, sampling means (MECH) for outputting analog samples of the input signal, first processing means ( MT 1) capable of carrying out discrete Fourier transform processing on the samples of the signal and of delivering first intermediate samples, second processing means (MT2) capable of effecting a modification of the spectral distribution of the first intermediate samples and to be delivered second intermediate samples, and third processing means (MT3) capable of performing inverse discrete Fourier transform processing on the second intermediate samples and delivering analog samples of an output signal having a second frequency lower than the first frequency.

2. Device according to claim 1, wherein the second processing means (MT2) are able to carry out a frequency translation on the first intermediate samples.

3. Device according to one of the preceding claims, wherein the direct and inverse discrete Fourier transform processes include butterfly-type elementary processing corresponding to several stages of a generally butterfly-shaped calculation graph, and the first and second third processing means have a pipelined architecture comprising several elementary stages (ETL) respectively corresponding to the stages of the calculation graph.

4. Device according to claim 3, wherein the first and the third processing means are able to process successive blocks of N samples, and said elementary stages are respectively able to perform Fourier transform processing. on blocks of samples of successively reduced sizes from one elementary stage to the next, the number N and the number of elementary stages being connected to one another via the radix of the Fourier transform.

5. Device according to claim 4, wherein each elementary stage comprises elementary input means (6), elementary output means (8), calculation means (27) capable of performing a Fourier transform elementary processing. of the radix butterfly type, storage means (22) capable of storing a number of samples depending on N and the rank of the elementary stage in the pipelined architecture, first controllable switching means (20) connected between on the one hand the elementary input means (6) and on the other hand the storage means (22) and the calculation means (27), second controllable switching means (24) connected between on the one hand the storage means (22) and the calculation means (27) and on the other hand the elementary output means (8), and control means (12) adapted during a first phase to control the first and second switching means in order to connect the storage means (22) between the elementary input means (6) and the elementary output means (8), and during a second phase to connect the storage means (22) to the calculation means (27) and the computing means (27) between the input means (6) and output means (8) elementary so as to allow the calculation of samples, the delivery of a portion of these samples calculated output and the storage of another part of these samples in the storage means (22).

6. Device according to one of the preceding claims, wherein the sampling means (MECH) are able to deliver samples representing the level of the input analog signal at each sampling time.

7. Device according to one of the preceding claims, wherein the sampling means (MECH) comprise controllable interruption means and the storage means comprise a delay line (22) including a chain of capacitors and controllable switches able to store and redeliver successive samples.

8. Apparatus comprising a reception chain including an analog part and a digital part having a signal processing processor (1 1), characterized in that the analog part incorporates a frequency shift device (7) according to one of the Claims 1 to 7.

Frequency offset method of an analog signal, comprising receiving an input analog signal having a first frequency, sampling the input analog signal to obtain analog samples of the input signal, and discrete Fourier Transform (31) processing on the signal samples to provide first intermediate samples, modification (32) of the spectral distribution of the first intermediate samples to deliver second intermediate samples, and treatment (33). ) of inverse discrete Fourier transform on the second intermediate samples so as to output analog samples of an output signal having a second frequency lower than the first frequency.

10. The method of claim 9, wherein said modification (32) of the spectral distribution of the first intermediate samples comprises a frequency translation on the first intermediate samples.