Patent Application
20050084052
The present application claims priority from French Application for Patent No. 03 10388 filed Sep. 2, 2003, the disclosure of which is hereby incorporated by reference.
1. Technical Field of the Invention
The present invention relates to the decoding of radio transmission channels carrying coded digital information. More precisely, the invention relates to the acquisition of the carrier frequency of a signal modulated with a quadrature single-carrier continuous modulation, for example, a quadrature amplitude modulation known as QAM.
2. Description of Related Art
Solutions currently used to carry out the acquisition and monitoring of the carrier frequency are based on phase-locked loops.
More precisely, a first category of solution uses a phase-locked loop having a wide frequency locking range for the acquisition of the carrier frequency, then a narrower frequency locking range when monitoring the carrier frequency. However, such a solution has a disadvantage in that the acquisition may fail with an input signal of poor quality.
A second category of solution uses a phase-locked loop having a narrow frequency locking range in order to be able to carry out the acquisition of a poor quality signal and in parallel uses a frequency scanning algorithm in order to acquire a signal exhibiting a significant frequency shift from the carrier frequency. However, this second solution has a disadvantage of requiring considerable locking time.
A need accordingly exists in the art to address the foregoing disadvantages.
More specifically, a need exists to be able to estimate and correct significant frequency shifts, for example of the order of hundreds of kHz even with a poor quality input signal, for example exhibiting a signal-to-noise ratio of the order of 20 dB for a 64 QAM modulation, and doing so for a short time, for example of the order of a hundred milliseconds.
In accordance with one embodiment of the invention, a method of processing the frequency shift of the carrier frequency of a signal modulated with a quadrature continuous single-carrier modulation comprises making a first estimate of the shift with a frequency estimator in order to obtain a first estimate value. Then, the estimate of the frequency shift is refined and the carrier frequency is corrected in consequence, with a phase-locked loop initialized with the first estimate value and with the locking frequency range being narrower than the locking frequency range of the frequency estimator.
In other words, according to an embodiment of the invention, the acquisition and monitoring of the carrier frequency is based on the use of a frequency estimator in combination with a phase-locked loop. At the time of acquisition, the frequency estimator makes an estimate of the frequency shift in an imprecise manner while the phase-locked loop is deactivated. Then, when the result of this first estimate has been obtained, it is used to initialize the phase-locked loop exhibiting a narrow frequency locking range.
Thus, the phase-locked loop will refine the result of the estimate delivered by the frequency estimator and supply an estimate with greater accuracy.
According to one mode of implementation of the invention, the oscillator of the phase-locked loop is controlled by the output of the loop filter augmented by the first estimate value.
The signal received by the frequency estimator comprises samples belonging to a constellation of dots. This constellation may be square (as is the case for example with 16 QAM, 64 QAM, 256 QAM or 1024 QAM modulations) or not square (as is the case for example with 32 QAM, 128 QAM or 512 QAM modulations). For the so-called square constellations, and, according to one embodiment of the invention, used to make the frequency estimate in a relatively simple and rapid manner even with a modulation of the QAM type exhibiting a large number of states, the first estimate is made using only received samples situated in the corner zones of the square.
The so-called non-square constellations nevertheless possess a square internal zone of dots. For these so-called non-square constellations, and, according to one mode of implementation of the invention, used to make the frequency estimate in a relatively simple and rapid manner even with a modulation of the QAM type exhibiting a large number of states, the first estimate is made using only received samples situated in the corner zones of the internal square.
Advantageously the first estimate is made on a predetermined number of samples of the signal received by the frequency estimator, for example on some 100,000 samples.
According to one mode of implementation of the invention, after activation of the phase-locked loop, the estimate of the frequency shift is continued with the frequency estimator, and this new estimate is used to track the evolution of the value of the carrier frequency.
This, for example, is to verify that the new values estimated for the frequency shift do not vary too much. More precisely, the variance of these estimates can be compared with a predetermined threshold.
An embodiment of the invention also proposes a device for processing the frequency shift of the carrier frequency of an incident signal modulated with a quadrature continuous single-carrier modulation.
According to one general feature of the invention, the device comprises a signal input, a frequency estimator connected to the input suitable for delivering a first estimate value of the frequency shift, a phase-locked loop connected to the input and whose locking frequency range is narrower than the locking frequency range of the frequency estimator, control means suitable for activating the phase-locked loop only on delivery of the first estimate value, and initialization means suitable for initializing the phase-locked loop with the first estimate value.
According to one embodiment of the invention, the initialization means comprise a memory means, such as a register, suitable for storing the first estimate value and an adder possessing a first input connected to the output of the loop filter, a second input connected to the memory means and an output connected to the control input of the loop oscillator.
The signal received by the frequency estimator comprises samples belonging to a constellation of dots, some of which belong to a square (this square is external for square modulations and internal for non-square modulations). And, according to one embodiment of the invention, the frequency estimator makes the first estimate using only received samples situated in the corner zones of the square.
According to one embodiment of the invention, the control means disconnect the frequency estimator from the phase-locked loop after the activation of the phase-locked loop, and the estimator then continues the estimate of the frequency shift. The device also comprises tracking means connected to the output of the estimator and suitable for using any new estimate to track the evolution of the value of the carrier frequency.
The device according to the invention is advantageously produced in integrated manner.
An embodiment of the invention also proposes a receiver of signals modulated with a quadrature continuous single-carrier modulation, comprising a processing device as defined hereabove.
A more complete understanding of the method and apparatus of the present invention may be acquired by reference to the following Detailed Description when taken in conjunction with the accompanying Drawings wherein:
In
The receiver RDS comprises at the head a tuner or tuning device linked to the antenna ANT. The tuner TZ is used to transpose the signal received by the antenna into a signal with a predetermined frequency, which may be the zero frequency (the signal is then in baseband) if the tuner exhibits an architecture of the zero intermediate frequency (ZIF) type or a non-zero intermediate frequency, for example 36 or 44 MHz.
The signal that has come from the tuner TZ and has been transposed onto two phase quadrature channels I and Q if the tuner is at zero intermediate frequency or onto a single channel if the signal is centered on a non-zero frequency, is then sampled in an analog/digital conversion CAN stage.
At the output of this analog/digital conversion CAN stage, a processing digital block BNM is connected comprising in particular at the head means of correction DRT (Derotator) suitable for correcting the frequency shifts of the frequency synthesizers.
Upstream of the correction means DRT, the signal is centered on the predetermined frequency (zero in the case of a ZIF architecture, or non-zero) to within one Δf frequency shift.
At the output of the correction means DRT, the intermediate frequency is removed to within Δf and the signal is then said to be pseudo-baseband and is centered on the Δf frequency.
Such correction (derotation) means are conventional and known per se. They may for example be those described in European patent application No. 0,481,543, the disclosure of which is hereby incorporated by reference.
The correction means DRT are followed by a synchronization loop BS, equally conventional and known per se, the function of which is to synchronize the rate of the data received with the rate of the data transmitted.
The synchronization loop BS is followed by a device DIS according to the invention, used to estimate the frequency shift Δf and in consequence correct the carrier frequency of the incident signal received.
The device DIS is followed by an equalizer EGL, of conventional structure, in turn followed by an error correction stage carrying out the conventional error correction processes well known to those skilled in the art, known as FEC (Forward Error Correction). The error correction stage then delivers a flow Fx which will be decoded in a source decoding block in order to restore the transmitted data. It should be noted that the equalizer EGL may be integrated into the phase-locked loop present in the device DIS.
A frequency estimation device EST is linked to the ESG signal input.
As will be seen in greater detail hereafter, this frequency estimator EST will deliver a first estimate value of the frequency shift Δf which will be stored in a memory means RG such as a register.
In addition to this frequency estimator, the device DIS comprises a phase-locked loop PLL comprising a complex mixer MX linked to the signal input ESG via a switch controlled by a signal SC1.
The phase-locked loop PLL also comprises in conventional manner a phase detector PD followed by a bandpass loop filter FB.
Between this loop filter FB and the controlled loop oscillator, which here is a numerically controlled oscillator NCO, there is an adder ADD with a first input connected to the output of the loop filter and another input connected to the register RG.
The output of the adder ADD is linked to the control input of the oscillator NCO.
The output of the oscillator NCO is looped back to the other input of the mixer MX.
The output of the mixer MX, when the phase loop is locked after having refined the estimate of the frequency shift Δf, will be a signal on the two channels I and Q, centered on the zero frequency. This signal is delivered to the equalizer EGL.
Furthermore, control means MCM deliver the control signals SC1, SC2 and SC3 controlling respectively the corresponding switches I1, I2, I3.
The operation of the device DIS will now be described in detail also with reference to
During a first phase of reception of the samples ECHS, the loop PLL is deactivated (step 40,
Although different structures of estimator may be used, the estimator EST described here as an indication is similar to an estimator used for a 4 QAM or QPSK modulation and this is so irrespective of the size of the modulation. It estimates (step 41) the frequency shift Δf by first computing the phases of the different samples retained and then computing the time derivatives of these computed phases (after removal of the modulation). This is done on a predetermined number of samples, for example on around 100 000 samples. The average of the values thus obtained is proportional to the estimate of the frequency shift and is used to supply a first estimate value Δfb of the value of the frequency shift.
One solution for computing the phase of a complex number consists in building into the estimation means a specific algorithm known to those skilled in the art as the CORDIC (Coordinate Rotation Digital Computer) published in the review IRE Trans.Electrons.Compute. EC-8, pages 330-334 (J. Volder-1959), the disclosure of which is hereby incorporated by reference.
This being so, in the presence of a QAM type modulation for example, with a large number of states, 64 for example, the removal of the modulation becomes very complex (the operation of removing the modulation is also routinely called ambiguity resolution).
Also, advantageously the choice will be made to take only “easy” dots to process, that is to say, in this particular instance, dots situated in corner zones of the constellation.
More precisely,
The constellation dots situated in the corners of the square are those for which I and Q are (in the absence of frequency shift) equal to 7 in absolute value.
In consequence, the corner zones Z1-Z4 will be defined as the zones for which the values I and Q of the samples received are (in the absence of frequency shift) in absolute value higher than a predetermined threshold, in this particular instance 6.
Similarly,
The internal square of the constellation corresponds to the constellation dots for which I and Q are (in the absence of frequency shift) equal to 3 in absolute value.
Consequently, the corner zones Z1-Z4 will be defined as the zones for which the values I and Q of the samples received are (in the absence of frequency shift) in absolute value higher than a first predetermined threshold, in this particular instance 2, and lower than a second predetermined threshold, in this particular instance 4.
Thus, for each sample received, the estimation means EST retain and process only those situated in the corner zones Z1-Z4.
This therefore boils down, to an extent, to the processing of a QPSK type modulation (that is to say a 4-QAM modulation) much less complex to process with regard to removal of the modulation.
On the other hand, the frequency estimate thus obtained loses accuracy but will in any case be refined by the phase-locked loop.
In this respect, the locking frequency range of the estimator, that is to say the frequency range outside which it is not possible to estimate the frequency shift, depends on the speed of data transmission and on the characteristics of the estimation algorithm. Those skilled in the art will be able to adjust the algorithm characteristics in order to obtain a sufficiently large estimator locking loop, for example of the order of a few hundred kHz. One of the characteristics of the algorithm that can be used to adjust the locking frequency range of the estimator is the number of samples not situated in the corner zones that are tolerated between two samples situated in the corner zones.
As an indication, for a number equal to four, a locking frequency range is obtained of the order of a few hundred kHz for modulations having speeds of the order of 5 Mbaud.
After this estimation phase 41 (
The phase-locked loop PLL is then activated (step 43; switch I1 closed; switch I2 open) and switch I3 is opened.
The phase-locked loop PLL is therefore initialized with the value Δfb originating from the register RG and will refine the estimate of frequency shift until it delivers at the output of the mixer MX a signal centered on the zero frequency. The locking frequency range of the phase shift loop is narrow. More precisely, it is chosen to be narrower, even much narrower than the frequency locking range of the frequency estimator EST, but greater than the maximum estimation error at the output of the estimator EST. As an indication, the characteristics of the loop filter FB will be chosen in particular in order to have a locking frequency,range of a few kHz for the loop PLL.
On activation of the loop PLL (step 43), the estimation means EST advantageously continue the estimation of the frequency shift (step 44). These estimate values are then transmitted to tracking means MCTL (switch I3 closed), in order to track the evolution of the carrier frequency Fc (step 45). In this respect, the tracking means will verify that the different estimated values of the frequency shift do not vary too much. More precisely, a way of verifying these consists in computing the variance of the estimated values and comparing this variance with a predetermined threshold. According to one mode of implementation of the invention, if this variance is greater than the predetermined threshold, the system may be reinitialized (return to step 40).
Although preferred embodiments of the method and apparatus of the present invention have been illustrated in the accompanying Drawings and described in the foregoing Detailed Description, it will be understood that the invention is not limited to the embodiments disclosed, but is capable of numerous rearrangements, modifications and substitutions without departing from the spirit of the invention as set forth and defined by the following claims.
| Number | Date | Country | Kind |
|---|---|---|---|
| 03 10388 | Sep 2003 | FR | national |
