High-frequency receiver with multiple-channel digital processing

Translation of description of equivalent WO 2008142278 A2

p0001High frequency digital multi-channel receiver processing

p0002The invention relates to broadband high frequency receivers.

p0003In such receivers, which are used in various fields, the receive chain typically begins with a head amplifier (front end) low noise, coupled to the antenna, usually through an antenna filter. Then came one or more frequency changes, leading to a useful signal of lower frequency, thus easier to treat. It is common to digitize this useful signal for further processing.

p0004A current trend is to place the analog to digital conversion as early as possible in the reception chain. Different public elements show.

p0005This trend is, among others, used in the field of software radio. The article by J. Mitola "Software Radio Survey, Critical Assessment and Future Directions", IEEE Telesystem National Conference, Washington, DC, May 19-20, 1992 describes the general principles of software radio.

p0006We know the article by B. Denby and. al "Towards a software-enabled radio broadcast media navigator" EURASIP 4th Conference on Video / Image Processing and Multimedia Communications, Juiy 2-5, 2003 Zagreb, Croatia. This article proposes to digitize a signal from the AM band with a specialized card computer controlled and try to recover the demodulated signal to distinguish musical and vocal components.

p0007This is accomplished using band-pass filters centered on the chosen modulation frequencies, and decimating the resulting signal. The result is only partially satisfactory, as one station is actually decoded, and must use a PC to do this, limiting the autonomy of the whole. These architectures of digital processing radio frequency receivers are not satisfactory in all cases. This is especially true where it should simultaneously receive multiple channels or channels from different carriers. The present invention improves the situation.

p0008To this end, there is provided a radio frequency receiver, comprising a reception chain including a low noise amplifier followed by treatment stages, with a digital analog conversion.

p0009The ADC works on the output of the low noise amplifier, at a chosen rate (F e), which corresponds to a sampling bandwidth and processing stages include a dedicated circuit with

p0010* An input memory arranged to contain N successive digital samples, renewed cadence chosen for every M samples,

p0011* A function of low pass filtering complex digital, selected cutoff, operating on the input memory to supply N filtered digital samples,

p0012* A function of M-periodic summing the N filtered digital samples, supplying M filtered and summed digital samples,

p0013* A stage discrete Fourier transform M x M, operating on these M filtered digital samples and summed.

p0014Digital signals on M Fourier transform outputs represent M separate channels of width defined by the cutoff frequency lowpass supra.

p0015The invention also relates to a radar device comprising a receiver as described above.

p0016Other features and advantages of the invention will appear on examining the detailed description below and the attached drawings, in which: - Figure 1 shows a very general diagram of a radar,

p0017- Figure 2 shows the general block diagram of a radar receiver,

p0018- Figure 3 is a block diagram of a radar receiver according to Figure 2, wherein the receive chain is scanned after frequency change,

p0019- Figure 4 shows the general block diagram of one embodiment of the proposed receiver,

p0020- Figure 5 is a more detailed diagram of part of the receiver of FIG 4,

p0021- Figure 6 is a more detailed diagram of another part of the receiver of Figure 4, taking into account the nature of two components I and Q of the digital signal,

p0022- Figure 7 is a flow diagram of a portion of the processing performed in the receiver of FIG 4,

p0023- Figure 8 is a diagram with XY diagrams illustrating the output processing of 1 channel,

p0024- Figure 9 is a diagram with XY diagrams illustrating the output processing on M channels,

p0025- Figure 10 is a general block diagram of the demodulation of the I and Q components on the M output channels, and

p0026- Figure 11 is a block diagram of a passive radar processing which can be applied to output signals of the receiver of Figure 4.

p0027The drawings and the description below contain, essentially, elements of a certain character. They may therefore not only serve to better understand the present invention, but also contribute to its definition, if appropriate. In the radar system of Figure 1, a transmitter 1 feeds an antenna 10 via a radio frequency signal, modulated in principle, which irradiates a target 2. The backscattered radiation from the target 2 is received by the antenna 30 of a receiver 3, which also receives the transmitted signal 19, or an electronic representation 19 of the transmitted signal, which defines the sufficient, at least in its temporal characteristics.

p0028From the signal or "echo" picked up by antenna 30 and 19 of the signal, the receiver performs treatments that provide access to the speed and / or distance of the target 2. In principle, the distance treatments are related to propagation times, and the speed of processing for frequency offsets by Doppler effect.

p0029The general architecture of a radar receiver is illustrated in Figure 2. After the antenna 30 and its possible radio frequency band filter 31, typically found a high frequency low noise amplifier 33, followed by at least one stage of frequency change 35 (or IF for "intermediate frequency"). Following the reception chain includes a demodulation function 37, followed by a stage 8 of radar processing, which provides treatments that provide access to the speed and / or distance of the target 2, as already in indicated.

p0030More in detail :

p0031- The antenna 30 is adapted to the band used, as well as its bandpass filter 31 for selecting a frequency band in which the radar signal is present,

p0032- The low noise amplifier 33 corrects the received signal level without the taint of noise,

p0033- The frequency change stage 35 is a frequency converter. It performs a transposition of the received signal (from the high frequency antenna generally) to a so-called intermediate frequency (IF), which is lower, even zero. This stage 35 requires the presence of a local oscillator (not shown), which must generally be synchronized with the transmitter, in that the difference of frequency and phase between the local oscillator and the transmitter oscillator must be known or determinable; - A demodulator for baseband recovery of the transmitted signal;

p0034- A processing unit for determining the speed of information and distance of the target.

p0035Modern radar receivers are digitized, such as shown in Figure 3. After stage 35 intermediate frequency (IF), it operates a digital analog conversion 39, which is called "complex" because it has two components, like complex numbers. Most often, there are the I component (in-phase) and Q component (quadrature). These two components are used to accurately reflect the amplitude and phase of a sinusoidal signal.

p0036In a digital radar receiver, the demodulation function is not as clearly distinct. This is why we tend to consider that the analog to digital conversion 39 is part of the treatment 8. The actual processing 81 then operate on these digital signals. And they can be made, for example, at least in part, in a programmable circuit, which is the processor type of digital signal processing ( "DSP" for Digital Signal Processor, microprocessor, or "FPGA" for Field Programmable Gate Array).

p0037When the radar transmitter and receiver are close (typically, they use the same antenna), it is called monostatic radar. One speaks of bistatic radar when the transmission and reception locations are distinct.

p0038On another level, conventional radars that emit themselves, are called "active". We talk about radars "passive" or opportunistic radars, for those who use existing programs for other purposes, such as:

p0039- Analog signal broadcasting, FM and TV,

p0040- Digital signal broadcasting, DAB, DVB-S and DVB-T,

p0041- The GSM signals. For cons, the satellite signals can not be used now because their power is too low to be workable.

p0042Passive radar is an example where it is desired to receive multiple programs simultaneously. Hence the need to use multiple receiver chains. The present invention is proposing a more effective approach.

p00434 is the general block diagram of a proposed receiver here.

p0044As before, there is the antenna 30 adapted to the band used (by its band-pass filter, not shown), and the low noise amplifier 33. The analog-digital converter 40 operates on the output of the amplifier low noise, without changing pre frequency. Further processing may be performed in a programmable circuit 5, as will be described hereinafter. The outputs can go to a communication interface 61 and a microcomputer 63 for post-processing.

p0045In the example described here, the antenna 30 and the low noise amplifier 33 are integrated in a ONEFORALL-SV9510 antenna. The digital-analog converter 40 is an AD9433 circuit manufactured by Analog Devices, and 5 programmable circuit is a Stratix FPGA EP2S 180 manufactured by Altera, programmed adequately to contain computer code that performs the above functions. It would also be possible to use another specialized circuit of ASIC type instead of an FPGA. Other elements and circuits that the art will recognize can also be used.

p0046The processing performed by the programmable circuit 5 will now be described with reference to an example to facilitate understanding. The example is that of the radio band frequency modulation, called FM band, which ranges from approximately 88 to 108 MHz. spaced there are transmitters of about 200 to 400 kHz. And each emits with limited modulation excursion around 100 kHz (200 kHz bandwidth).

p0047In the example, the analog-digital converter 40 may have a sampling rate F e of 43.9 MHz, and work on 14 bits. The sampling period Te is a little less than 23 nanoseconds. Processing then comprises the four steps Sl to S4 described below with reference to Figure 7.

p0048Sl - Filtering

p0049This step is performed by a block 51, wherein N successive digital samples from the analog digital converter 40 (label 700 in Figure 7) sequentially perform a memory 510 of the FIFO type N-stage (operation 702 in Figure 7). The memory 510 is rated "M memory." In the example, the size of the memory Ml of N x 14 bits. It was further N = K * M, for reasons that will be understood hereinafter.

p0050We also stored N coefficients, in 511, here with the same precision as the samples from the analog digital converter 40. These N coefficients correspond to coefficients of a type of low-pass filter bandwidth F band Butterworth_lp. This filter is preferably of the finite impulse response (FIR Finite Impulse Response), although other filters may be used, at least in some cases, where stability is not critical.

p0051In the example, the accuracy of each coefficient is 14 bits, and the cutoff frequency F_lp is 100 kHz, that is to say the modulation bandwidth of an FM transmitter.

p0052When the memory 510 is filled, the N acquired samples it contains are weighted by coefficients stored in N 511, respectively (operation 704). Ie that each filter coefficient weights sample acquired through a multiplier 512.

p0053The results of product samples by the filter coefficients are stored in a memory size of 515 N. To reflect the I and Q components of the signal, the filter coefficients are complex. Therefore, we must see the memory products 515 as elements to both I and Q components Using real filter coefficients, it is of course possible to achieve a real and not complex filtering. 52 - Addition and Folding

p0054The next processing occurs when the memory 515 is completed and is implemented by a block 53. It then considers the split in the sampling order, K blocks of M elements each.

p0055These K blocks are added among them by means of an adder 531 to produce a single final block of M values ​​533 (operation 706). The first final block element is the addition of all the first elements of the K blocks. The second element of the final block corresponds to the addition of all second blocks of K elements, and so on. The last element of the final block corresponds to the addition of all the final elements of the K blocks.

p005653 - Discrete Fourier Transform

p0057When the step of adding and folding is finished, the block of M values ​​is input a block of 55 digital Fourier transform discrete ( "FTD") on M points (operation 708).

p0058The outputs y (j) of this block match the frequency decomposition of the input signal on two channels, I and Q I corresponds to the real parts and Q for the imaginary parts of the frequency decomposition of the signal. The output 559 of this block corresponds to two I and Q signals having M values. The first values ​​of I and Q, QO and 10 correspond to the zero hertz components of the input signal. The second values, II and Q, correspond to the components F / M Hertz input signal. The third values, 12 and Q2 correspond to the components 2 * F / M Hz of the input signal, etc. The last values, IM-I and QM-I correspond to the components (M-1) * F / M Hz of the input signal.

p005954 - Installation Channel (TM)

p0060When the treatment of FTD is finished, the block 57 channelization (TM) performs demultiplexing of the results temporally (operation 710). As shown in 6, the I and Q values ​​corresponding to the same frequency are grouped over time. The outputs of this block are equal to 10, OQ, II, Ql, ..., IM-I and the QM (t).

p0061When the channel formatting operation is completed, the result of M channels is forwarded to subsequent stages for demodulation. The circuit clears acquisition of new samples M input (return to operation 702). The memory Ml is then filled with these M new samples followed by (K-1) * M most recent previous samples, while the M first input samples preceding disappear. Steps Sl to S4 are then repeated.

p0062The memory blocks such as 510, 511 and 515 described above are useful for understanding the treatment. Of course, in practice, these blocks have not be distinct, and may be portions of the same memory. Similarly, the described treatments can be performed at least partly sequentially or serialized.

p0063Functionally, the above processing can be described as follows, with reference to the non-limiting example above, with:

p0064N = 8192

p0065K = 16

p0066M = 512

p0067Sampling is done F_e = 43.9 MHz, about 44 MHz. It produces a spectral folding on the FM band from 88 to 108 MHz approximately expressed by the following table:

p0068<img id="imgf000011_0001" he="26" wi="87" file="imgf000011_0001.tif" img-format="tif" img-content="table" orientation="portrait" inline="no" />

p0069The low-pass filtering with cut-off at 100 kHz brings all its low frequency variations. There is a strong over-sampling, since the sampling rate is 43.9 MHz. It is therefore possible to make sample means. These averages are comparable to one another, provided one follows the same periodicity in constructing every average. The formula Eq. 1 given as an example meets this condition. • £ (Eq. 1) 1^{= 0}

p0070Thus, low frequency was condensed the information on the FM band, with insufficient sampling to represent the FM carrier frequency, but more than enough to regain (s) modulation (s).

p0071Two successive average can be regarded as temporally offset from each other by a sampling period Te. Therefore, the working frequency of the discrete Fourier transform is F_fWith, in Example F_f = 51.2 MHz.

p0072The Fourier transform performed on M points analysis, stepwise in time, on the M channels, the desired modulation. In the example, with F_f = 51, 2 MHz and M = 512 is obtained at the output 512 channels of 100 kHz.

p0073As apparent in view of the above, the sampling frequency and the FTD of the working frequency are slightly different. Alternatively, the samples from the memory 510 can be interpolated before or during step S2 it to take account of this difference in frequency.

p0074The skilled person will understand that this is a sampling bandwidth, rather than a sampling deNyquist, where one would seek to accurately reflect the amplitude and phase of the carrier.

p0075Those skilled in the art will understand that the WOLA algorithm works on the basis of F oversampling₆ compared to F_lp. When taking into account the nature of FIFO memory 510, Eq. 1 can be seen as a circular convolution of each sample x (j) by the low-pass filter h (j), with a sampling frequency of Ε / M. Therefore, at the entrance of the FTD, there is a set of M samples, game which renews the ΕflΛ rate. By cons, the "frequency sensitivity", or frequency characteristic of these samples remains bound to the original sampling rate, namely F_e.

p0076FTD on all r (j) discriminates then K aliasing spectrum relative to the sampling frequency F_e. The resulting collection is (K) / K, since 'each cycle Sl to S4, only new M samples on K * M are renewed. That's what keeps a good time coherence.

p0077The signal is then processed to recover the signal on each channel as defined above. As shown in Figures 8 and 9, this block makes it possible to demodulate the signal on a particular channel (Figure 8), or on all the channels (Figure 9).

p0078The demodulation can be performed by software processing, and depends on the type of modulation used when transmitting.

p0079Thus, for a frequency modulation (Figure 8), arc tangent type of processing is performed to restore the information. For an amplitude modulation (Figure 9), an envelope detector comprising a low pass filter is used.

p0080An embodiment of the demodulation means is given in Figure 10. The demodulation set 59 includes a series of lookup tables LUT (i) each associated with a channel at the output of block 57. In general, the LUTs (i) can be implemented in the form of 512 separate tables and a comprehensive table containing all these tables with an identifier of each table. Alternatively, it may also be possible to replace the tables by corresponding mathematical formulas.

p0081The output of each LUT (i), each pair of signals (I (i); Q (i)) corresponds to a numerical value designating an angle. The resulting signal is then sent to a differentiator which provides the frequency of the output signal of the LUT (i), and the demodulated signal on each channel. The foregoing description has been carried out iteratively, in relation to the processing of a given series of samples. However, the receiver may operate at least partially in parallel, for example by channeling and demodulating signals from samples already processed by the folding block and addition, while the following samples are being treated by the same block.

p0082The applicant has established that the signal / noise ratio ( "S / N") obtained at the output of the 5 programmable circuit is a level enabling a radar processing on the demodulated signals. For example, rates of S / N of 60 dB were achieved in experiments, which confirmed the interest of the implementation of the receiver of the invention to the passive radars applied to the FM band.

p0083The output of this circuit is digital and is connected to the computer 63 through the interface 61 for radar processing itself.

p0084A passive bistatic radar is because the radar reception is on backscatter targets that emit one or more transmitters that exist on the ground for other reasons. The radar receiver is an opportunistic use of these transmitters, which we call here "issuers [radar] involuntary."

p0085In general, the radar receiver and transmitter are not involuntary at the same point. In addition, the radar receiver has not directly time references of the involuntary transmitter: it simply knows when he himself receives the signals transmitted and knows his position and that of the transmitter, so its distance to the transmitter, thus the travel time of the waves between them.

p0086Consider a given time difference DT between firstly the signal backscattered by a target and received by the radar antenna, and secondly the signal received also by the receiver in direct propagation from involuntary transmitter. It is known that the target is on an ellipse, set by DT, and whose foci are the involuntary transmitter and the radar receiver. The position of the target on the ellipse can be determined based on the receiving direction of the signal received by the receiver, that is to say the axis of the antenna. The movement of the target has the effect of varying the parameter DT. Furthermore, this movement also has an effect on the backscattered signal, that those skilled in the art known as the Doppler effect. Thus, it is known that a signal transmitted from a transmitter to a receiver that are driven in a relative movement undergoes a frequency shift proportional to the relative velocity of the transmitter and receiver. More specifically, this phase shift is a function of the affine projection of this axis velocity transceiver.

p0087In the case of bistatic radar as the target moves both relative to the involuntary transmitter and from the receiver, the signal is Doppler effect said bi- static which comprises two components formed by the affine projections vector velocity of the target on the radial axis target involuntary transmitter on the one hand, and the other target receiver.

p0088In practice, the parameter DT is determined by correlation of the signal emitted by the involuntary transmitter and the signal backscattered by the target. However, because of the nature of the signal emitted by the involuntary transmitter, these signals can be correlated with themselves.

p0089Therefore, we must change the signal reflected by the target to reflect the phase shift

p0090Doppler bi-static, and thus allow the correlation with the signal emitted by the involuntary transmitter to determine the parameter DT.

p0091One way to determine these parameters by attempts is described in the article by

p0092Howland et al "FMRadio basedbistatic radar", IEE Proceedings online no.20045077, IEE

p00932005.

p0094In short, we find here the treatment of a bistatic radar with a somewhat particular application due to the use of involuntary rémetteur. Passive radars have been described in various public elements.

p0095This is firstly the case of Manastash project of the University of Washington, whose description is available at the following link: http://klickitat.ee.washington.edu/Proiects/Manastash/ In this project, Professor John Sahr of the electrical engineering department of the University of Washington has developed a passive radar dedicated to environmental applications control the fluctuation of the ionospheric layer.

p0096[This is also the case of the article by Y. Zhao et al, "Adaptive Baseband Architecture for Software-Defined Radar Application", IEEE CCECE inproc of 2003, Montreal, May 2003. This article reports a receiver radar whose architecture comprises an analog radio frequency stage for receiving radar signals, a digital analog converter, analog and digital, a processor optimized for signal processing (DSP) and a communication interface with a computer. This architecture has been developed primarily for receiving radar signals supporting multiple encoding types FSK, BPSK, PSK and CHIRP. The characteristics of the transmitted signals are programmed in software on the DSP. The same architecture is used for transmission of radar signals]

p0097yet we know the article by R. Walke et al. "An FPGA based digital radar receiverfor soft radar," Signals, Systems and Computers, 2000. Conference Record of the Thirty-Fourth Asilomar Conference on Volume 1, the 29 Oct.-Nov. 2000 Page (s): 73-77 vol. l. This article describes the architecture of a radar system dedicated to beamforming by computation. The system consists of N antennas provided with an analog filter, a digital analog converter, and a digital receiver. The latter performs baseband conversion using a local oscillator and a low-pass filter. The output of the digital receiver is a signal not demodulated baseband.

p0098In the context of the invention, as shown in Figure 11, the signals received directly channeled to the involuntary transmitter (or direct signal) and those received from the target (or backscattered signals) are processed by the computer 63 or any other suitable processing means for determining the time difference ti and the Doppler shift Φi on each channel.

p0099Ti differences can be obtained by correlating, channel by channel, the direct signal and the backscattered signal demodulated. The Doppler shift can be determined by comparing, channel by channel, the spectra of the direct signal and the spectrum of the backscattered signal modulated. From a theoretical point of view, all the ti must be identical, so you can settle for a single calculation. However, for best accuracy, it is possible to determine more or all ti, then to average them. It is also possible to sum the direct signals demodulated on the one hand and the demodulated signals rétrodifrïïsés other hand, and to correlate these two sums of signals for the same average.

p0100Like all reports Φi / fi (where fo is the channel modulation frequency i) are theoretically equal, the same average principle can be applied for Doppler shifts. However, in this case, it is not possible to average in a single operation: it is necessary first to calculate all Φi individually.

p0101Other examples of radar treatments can be applied to demodulated signals are described in the publications: "digital signal processing radar" by JJ. Julie R. Sapienza, Editions Hermes, 2004, "Digital processing of radar was signing detailed" by Schuster, J. et al, Joint Publications Research Service icts in East Europe Report (JPRS-EER- 860010) pl29-141 (EES N86- 11-32 20665) 1986 or "Radar HandBook" MI Skolnik, 2nd edition, McGraw-Hill Professional Publishing, 1990. other post-processing can also be applied it is not necessary to describe here .

p0102The description above has been described in the context of a radar application. However, those skilled in the art will recognize that the above can be implemented in almost any "time" application of wave propagation.

p0103Thus, the receiver described above may be applied for research clear frequencies, that is to say for unused channels for the radar emission. Other applications include coastal radars as receivers arranged on buoys and operating in the range HF / VHF / UHF for detecting naval vessels and pleasure boats.

p0104Finally, the receiver described above could also be applied to radars OFDM and in other particular applications.

p0105Those skilled in the art will appreciate that, in the context of these applications, the timing accuracy is crucial. It is therefore necessary to digitize the signal in a form known as "complex", that is ie with an in-phase component, denoted by (I) and quadrature component, denoted (Q), and applying a complex subsequent treatment.

p0106Other applications are less demanding in terms of time accuracy, and allow direct real treatment, which can be supplemented with a complex treatment as described.

p0107These applications include the concept of "Radio On Demand", that is to say the demodulation of the entire FM band for simultaneously dispensing separate channels in various places with a single receiver. An application can be found in the car, with the possibility for each passenger to listen to a particular station on the basis of a single receiver. Of course, this listening can support RDS.

p0108Another application is the concept of "cognitive radio", which aims to test different radio channels available for own issue or for simple detection. Finally, the receiver can also be used as the AIS receiver, which operates on 88 25 kHz channel in the 157-162 MHz band, to prevent collision situations.

p0109As mentioned above, those skilled in the art will recognize that these applications do not require the same temporal precision, and can be performed more simply.

p0110The invention also covers, as products, the software elements described, made available under any "medium" (support) readable by computer. The term "computer readable medium" includes the data storage media, magnetic, optical and / or electronic as well that a carrier or vehicle transmission, such as an analog or digital signal.