The present disclosure is directed to a method for transmission of information over a transmission channel and, in particular to, a transmission over a power line and related devices.
Power line communications (PLC) technology is aimed at transmitting digital data by utilizing the existing infrastructure of the electrical grid. It allows, in particular, remote reading of electric meters, exchanges between electric vehicles and the recharging terminals and also management and control of energy networks (smart grid). PLC technology incorporates, in particular, narrow band power line communication (N-PLC) which is generally defined as a communication over an electrical line operating at transmission frequencies of up to 500 KHz. N-PLC communication thus generally uses the frequency bands defined in particular by the European committee for electrotechnical standardization (CENELEC) or by the Federal Communications Commission (FCC). Thus, if the CENELEC A frequency band (3-95 kHz) is considered, the transmission frequencies are situated between 35.9375 and 90.625 KHz for the PLC-G3 standard.
The overall performance of a receiver greatly depends on the quality of its channel estimation, in other words of the estimation of the transfer function of this channel. It is known that a transmission channel can vary over time, in frequency, in phase, and in amplitude. Furthermore, the signals conveyed via PLC and received by the receiver result from a combination of several signals having taken several propagation paths within the transmission channel (i.e., the electrical power line), each having its own time delay and its own attenuation (the transmission channel is a multi-path transmission channel). This can then result in a significant attenuation of certain frequencies.
Furthermore, the properties and characteristics of electrical power grids are not a priori known and are variable over time. Thus, the impedances of certain objects connected by the user vary with the voltage. This is the case, for example, for halogen lamps or for objects comprising voltage rectifiers. When a user connects such objects, this then results in a variation of the transfer function of the transmission channel. The channel is then considered as linear and cyclostationary or linear and varying over time in a periodic manner, i.e. Linear Periodically Time Varying (LPTV). In other situations, the channel may be considered as linear and time invariant (LTI).
Generally speaking, a method is for processing a channel analog signal from a transmission channel. The method may include converting the channel analog signal into a channel digital signal, and detecting a state of the transmission channel based on the channel digital signal to detect whether the transmission channel is, over an interval of time, at least one of LTI and linear and cyclostationary.
The current receivers, compatible with the PLC-G3 standard, may not be adapted for carrying out channel estimations when the latter is variable over time. Indeed, the PLC-G3 standard may only include two orthogonal frequency-division multiplexing (OFDM) symbols as pilot symbols for totally estimating the transfer function of the channel. Also, when the channel varies, and notably when in the presence of a cyclostationary channel, the estimation of the channel is erroneous, or even impossible, and consequently leads to errors in decoding the symbols.
Various embodiments of the present disclosure and their implementation are compatible, notably but not exclusively, with the PLC-G3 standard such as defined, for example, in the version of April 2015 of the document from the G3-PLC Alliance entitled “Narrowband OFDM PLC specifications for G3-PLC networks” to which those skilled in the art will be able to refer at their convenience. Generally, the idea is to limit the risk of decoding errors due notably to erroneous channel estimation. Relatively simple channel estimation is provided, even in the case where the channel is considered as cyclostationary, and compatible with the PLC-G3 standard.
According to an aspect, a method is provided for processing a channel analog signal coming from a transmission channel, for example, an electrical power line, the signal then being conveyed via PLC. The method comprises at least one phase for detecting the state of the transmission channel based on a digital signal coming from an analog/digital conversion of the channel analog signal in such a manner as to detect whether the transmission channel is considered, over an interval of time, as being LTI or else as being LPTV.
The channel analog signal, which will undergo the analog/digital conversion, may, for example, be the analog signal coming directly from the channel or else, as is generally the case, the analog signal delivered by an analog input stage (notably comprising band pass filters, low-pass filters and an amplifier) connected to the transmission channel. The method may differ from the prior art by the fact that, prior to a channel estimation followed by a decoding of the symbols, the state of the transmission channel is detected, this being in real time from the received signal, which allows the risk of decoding errors due to an erroneous channel estimation to be limited.
This indeed allows a decision to be taken with regard to the subsequent processing operations to be applied, such as, for example, to temporarily suspend the decoding or else to use a channel estimation appropriate to the various states of the channel and thus to improve the overall performance of the receiver. The channel analog signal conforms, for example, to the PLC-G3 standard. When the channel analog signal conveys symbol frames, the detection phase can be carried out during the reception of at least one frame, the detected state of the channel being considered as valid for the whole duration of the at least one frame. In other words, the channel detected as being cyclostationary (periodically time varying) is considered as being cyclostationary for the whole duration of the at least one frame, and a channel detected as being time invariant is considered as being time invariant for the whole duration of the at least one frame.
However, it is preferable for the detection phase to be carried out during the reception of each frame since the state of the channel may potentially change from one frame to another. So, here again, it is considered that the state of the channel detected during the reception of a current frame is considered as valid for the whole duration of this current frame.
In an embodiment, the detection phase comprises a plurality of channel estimations respectively carried out using a plurality of reference symbols of the received frame in such a manner as to obtain a plurality of transfer functions for the channel, and a determination of at least one phase difference between at least one pair of transfer functions. The detection of the state comprises a comparison of the absolute value of the at least one phase difference with a threshold, the channel being considered as linear and time invariant if the absolute value of the at least one phase difference is less than or equal to the threshold and as linear and cyclostationary in the opposite case. On the transmission side, the channel analog signal comes from a digital/analog conversion of an initial digital signal, and when the sampling frequency of the digital signal (at reception) is different from the sampling frequency of the initial digital signal (at transmission), it is preferable to take into account this sampling frequency offset for correcting the transfer functions.
Thus, according to an embodiment, the plurality of channel estimations comprises estimations of base transfer functions respectively carried out using the plurality of reference symbols of the received frame, and a processing operation for correction of these base transfer functions with a phase shift corresponding to this shift in sampling frequency in such a manner as to obtain the transfer functions. In order to more easily detect the time variations of the channel, it is preferable to choose a set of reference symbols of the received frame which extend in time over a total duration equal to or slightly longer than the cyclostationary period, where it exists, of the channel, and to choose the two transfer functions of each pair, from which a phase difference is determined, in such a manner that they are respectively associated with two reference symbols separated in time by a period longer than a limiting period of time corresponding to a minimum number of reference symbols.
The total duration of the reference symbols is slightly longer than the cyclostationary period when the latter is not an integer multiple of the duration of a reference symbol. Thus, in this case, this period is, for example, included between two consecutive integer multiples of the duration of a reference symbol. Thus, by way of example, when the signal conforms to the PLC-G3 standard, each received frame comprises a preamble sequence followed by a header followed by a payload. The reference symbols then advantageously comprise the symbols of the header and two symbols of the payload corresponding to two known symbols of the transmitted frame.
The number of reference symbols is then equal to 15 and together extend over a duration equal to 10.42 ms (since each reference symbol has a duration equal to 0.695 ms). This total duration of 10.42 ms is slightly longer than the cyclostationary period, where relevant, of the channel (10 ms) which is equal to half of the period of the alternating current that will be flowing in the electrical power line (i.e., 20 ms for a frequency of 50 Hz). In the same way, in countries where the frequency of the mains is 60 Hz, the cyclostationary period of the channel is equal to 8.33 ms, and the number of reference symbols can be taken equal to 12.
The limiting number of reference symbols separating the two reference symbols of a pair is, for example, equal to 6, and the threshold allowing the time invariant or cyclostationary nature of the channel to be detected depends on the implementation chosen and may, for example, be equal to 0.75 radians. In such a manner as to further improve the detection of the state of the channel, the detection phase can comprise a determination of several phase differences respectively between several pairs of transfer functions, a processing operation for averaging the absolute values of these phase differences in such a manner as to obtain a mean phase difference, and a comparison of the mean phase difference with the threshold. The channel is considered linear and time invariant if the mean phase difference is less than or equal to the threshold and linear and cyclostationary in the opposite case.
The reference symbols of the received frame from which the various transfer functions will be determined, are, for example, symbols of the received frame corresponding to known symbols of the transmitted frame and/or are symbols that can be decoded without knowing the transfer function of the transmission channel. Thus, in the case of the PLC-G3 standard, as indicated hereinbefore, the reference symbols may be the thirteen symbols of the header of a frame, which can be decoded without knowing the transmission channel transfer function since they are coded in a differential manner, together with the two symbols of the payload of the frame, which correspond to the two known symbols S1, S2 of the payload of the transmitted frame.
It will accordingly be noted here that, although the PLC-G3 standard only provides as pilot symbols for the estimation of the transfer function of the channel two known symbols, namely the symbols S1 and S2, it is advantageously provided here for this number of pilot symbols to be increased by using the symbols of the header of a frame which may be decoded without needing to know the transfer function of the channel. Any potential time variations of the transmission channel can then readily be detected. However, it would also have been possible to also use as reference symbols all or part of the known symbols of the preamble sequence of a frame in order to improve the channel estimation or to increase the number of estimations if the duration of the header and of the symbols S1 and S2 is shorter than the duration of a half-period of the alternating signal conveyed by the channel (half-period of the mains supply).
Additionally, each channel estimation carried out using a received decodable symbol (e.g., a symbol of the header) comprises a decoding of the received symbol, a re-encoding of this decoded symbol in such a manner as to obtain a re-encoded symbol and a determination of the transfer function of the channel using the re-encoded symbol and the received decodable symbol. When the state of the transmission channel is detected, several possibilities are offered.
It is possible, in some applications, that the receiver having detected a transmission channel of the LPTV type is not adapted for correctly estimating the transfer function of such a channel. As a consequence, for performing the decoding of the symbols without too great a risk of error, the receiver decides to wait for the channel to, potentially, return to being of the LTI type in order to carry out the decoding of the symbols of the corresponding frame in the conventional manner according to the PLC-G3 standard, using only the symbols S1 and S2 of the frame to carry out the channel estimation then the decoding.
However, according to an embodiment, when the transmission channel is considered as linear and cyclostationary for the whole duration of the frame, it is possible to carry out an estimation of the transfer function of such a cyclostationary channel followed by a decoding. Accordingly, the generation may then be provided of a sequence of N transfer functions of the transmission channel, respectively associated with N successive time slices, the total duration of the N time slices being equal to almost the whole of the cyclostationary period of the channel (when this cyclostationary period is not an integer multiple of the duration of a time slice or of a symbol) or to the entirety of the period, then a decoding of at least some of the symbols of a frame by using for each of these symbols the transfer function associated with the time slice containing the symbol. A total duration of the N time slices equal to almost the whole of the cyclostationary period of the channel means or the channel, for example, that the cyclostationary period is in the range between N times and N+1 times the duration of a time slice.
In order to generate these N transfer functions, an initial sequence of M transfer functions respectively associated with M successive reference symbols of each frame may, for example, be generated, the M reference symbols covering a period of time equal to or slightly longer (when this cyclostationary period is not an integer multiple of the duration of a symbol) than the whole of the cyclostationary period, the N transfer functions being generated starting from these M transfer functions. The M transfer functions are advantageously those obtained during the phase for detecting the state of the transmission channel. Thus, in the case of the PLC-G3 standard, M is, for example, equal to 15 since fifteen reference symbols are used for generating the fifteen transfer functions.
Although it would be possible to use fourteen of these fifteen transfer functions for decoding the symbols of the payload of the frame, it turns out to be helpful, in order to obtain a better signal-to-noise ratio, to only use seven transfer functions (N=7), these N transfer functions being, for example, the averages taken two by two of fourteen successive transfer functions of the initial sequence of the M transfer functions. The decoding of the symbols following the last reference symbol then advantageously uses, successively and cyclically, the transfer functions of the sequence of the N transfer functions (e.g. N=7).
When the duration of the N time slices is different from the cyclostationary period, in other words when this cyclostationary period is not an integer multiple of the duration of a symbol, the decoding of symbols following the last reference symbol also comprises, periodically, a shift of transfer function in the successive and cyclical use of the transfer functions. Thus, for example, at a given moment in time and periodically, the same transfer function could be used twice in succession before returning to the normal cycle of the transfer functions.
When the signal is conveyed via PLC, the cyclostationary period of the channel is equal to half of the period of the alternating signal (current or voltage) designed to flow over the electrical power line. In such an application, a reference signal is delivered at each passage/crossing through “0” of the alternating signal (this is a signal known by those skilled in the art using the term “zero-crossing signal”). In this case, the value of the cyclostationary period of the channel can advantageously be reset based on the occurrences of the reference signal.
When the transmission channel is considered as LTI for the whole duration of a frame, it is particularly advantageous to perform a processing operation for averaging the plurality of transfer functions for the channel obtained during the detection phase (e.g., the fifteen transfer functions respectively obtained from the fifteen reference symbols (13 symbols of the header and the two symbols corresponding to the two known transmitted symbols S1 and S2)) in such a manner as to obtain a mean transfer function. Furthermore, in this case, the decoding of the symbols of the frame following the reference symbols is carried out using this mean transfer function. This allows the signal-to-noise ratio of the mean transfer function to be significantly increased, which will ultimately lead to an increase in the decoding performance by several dB.
According to another aspect, a receiver is comprises an input stage designed to be connected to a transmission channel and configured for delivering a channel analog signal coming from the transmission channel, a stage for analog/digital conversion of the channel analog signal so as to deliver a digital signal, and means for processing or a processor the digital signal. The processing means may comprise detection means or a detector configured for carrying out at least one phase for detecting the state of the transmission channel based on the digital signal in such a manner as to detect whether the transmission channel is considered, over an interval of time, as being linear and time invariant or else as being linear and cyclostationary.
Also, the channel analog signal conveys symbol frames, and the detection means are configured for performing the detection phase during the reception of at least one frame, the detected state of the channel being considered as valid for the whole duration of the at least one frame. However, the detection means can be configured for carrying out the detection phase during the reception of each frame, the state of the channel detected during the reception of a current frame being considered as valid for the whole duration of this current frame.
According to an embodiment, the detection means may comprise estimation means or an estimator configured for respectively carrying out a plurality of channel estimations using a plurality of reference symbols of the received frame in such a manner as to obtain a plurality of transfer functions of the channel, calculation means or a calculator configured for determining at least one phase difference between at least one pair of transfer functions, and comparison means or a comparator configured for performing a comparison of the absolute value of the at least one phase difference with a threshold. The channel is considered as LTI if the absolute value of the at least one phase difference is less than or equal to the threshold and as linear and cyclostationary in the opposite case.
According to an embodiment, the channel analog signal comes from a digital/analog conversion of an initial digital signal, and when the sampling frequency of the digital signal is different from the sampling frequency of the initial digital signal, the estimation means are configured for respectively carrying out estimations of base transfer functions based on the plurality of reference symbols of the received frame, and a processing operation for correction of these base transfer functions with a phase shift corresponding to this shift in sampling frequency so as to deliver the transfer functions. Moreover, the set of the reference symbols of the received frame extends in time over a total duration equal to or slightly longer than the cyclostationary period, where relevant, of the channel and the two transfer functions of each pair are respectively associated with two reference symbols separated in time by a period longer than a limiting period of time corresponding to a minimum number of reference symbols.
According to one embodiment, the calculation means is configured for respectively determining several phase differences between several pairs of transfer functions and for determining the average of the absolute values of these phase differences in such a manner as to obtain a mean phase difference. The comparison means is configured for carrying out a comparison of the mean phase difference with the threshold, the channel being considered as LTI if the mean phase difference is less than or equal to the threshold and as linear and cyclostationary in the opposite case. The reference symbols of the received frame are, for example, symbols of the received frame corresponding to known symbols of the transmitted frame and/or are symbols that can be decoded without knowing the transfer function of the transmission channel. The estimation means is then advantageously configured for carrying out a decoding of a received decodable symbol, a re-encoding of this decoded symbol in such a manner as to obtain a re-encoded symbol, and a determination of the transfer function associated with this received decodable symbol, using the re-encoded symbol and the received decodable symbol.
In some embodiments, the signal is modulated according to an OFDM modulation. The transmission channel may be an electrical power line, and the signal may be designed to be conveyed via PLC and the cyclostationary period, where relevant, of the channel is half of the period of the alternating current designed to flow over the electrical power line. According to an embodiment, the signal conforms to the PLC-G3 standard.
Each received frame may comprise a preamble sequence followed by a header followed by a payload, and the reference symbols comprise the symbols of the header and two symbols of the payload corresponding to two known symbols of the transmitted frame. The limiting number of reference symbols may be equal to 6. The threshold may be equal to 0.75 radians.
According to an embodiment, when the transmission channel is considered as linear and cyclostationary for the whole duration of the at least one frame, the processing means comprises generation means or a generator configured for generating a sequence of N transfer functions of the transmission channel respectively associated with N successive time slices. The total duration of the N time slices may be equal to almost the whole of the cyclostationary period of the channel and/or to the whole of the period. The processing means may include decoding means or a decoder configured for decoding at least some of the symbols of a frame using, for each of these symbols, the transfer function associated with the time slice containing the symbol.
The generation means may be configured for generating an initial sequence of M transfer functions respectively associated with M successive reference symbols of each frame, the M reference symbols covering a period of time equal to or slightly longer than the whole of the period, and for determining the N transfer functions from these M transfer functions. The M transfer functions may be those delivered by the estimation means incorporated within the detection means.
According to some embodiments, M is equal to 15, N is equal to 7, and the generation means is configured for calculating the averages taken two by two of fourteen successive transfer functions of the initial sequence in such a manner as to obtain the N transfer functions. The decoding means may be configured for using, successively and cyclically, the transfer functions of the sequence of the N transfer functions in order to decode the symbols following the last reference symbol.
Additionally, when the total duration of the N time slices is different from the period of cyclostationary, the decoding means may be configured for periodically applying a shift of transfer function in the successive and cyclical use of the transfer functions in order to decode the symbols following the last reference symbol also comprises. The signal may be designed to be conveyed via PLC, and the cyclostationary period of the channel may be half of the period of the alternating current designed to flow on the electrical power line. The input stage is configured for furthermore delivering a reference signal at each passage through zero of the alternating current, and the processing means are configured for resetting the value of the cyclostationary period of the channel based on the occurrences of the reference signal. When the transmission channel is considered as LTI for the whole duration of the at least one frame, the processing means may comprise an averaging module or an averaging circuit configured for carrying out a processing operation for averaging the plurality of transfer functions of the channel delivered by the detection means in such a manner as to obtain a mean transfer function, and decoding means configured for decoding symbols of the frame following the reference symbols using this mean transfer function.
The embodiments and their implementation will now be described in the framework of a transmission of information by PLC according to the PLC-G3 standard, although the present disclosure is not limited to this type of application. Throughout the following, each time that the PLC-G3 standard is cited, it will be assumed that the CENELEC A band of frequencies (3-95 kHz) is considered.
Reference is now made to
Each symbol contains modulation coefficients associated with carriers that will be modulated accordingly. The symbols are delivered at the input of processing means MTFI designed to perform an inverse fast Fourier transform (IFFT) operation. It will be noted here, referring more particularly to
Thus, in the PLC-G3 standard, the size of the inverse Fourier transform is equal to 256, whereas the modulated carriers of the subset SNS are included between the ranks 23 and 58, which corresponds to a frequency band F1-F2 in the range between 35.9375 and 90.625 KHz. Here, the sampling frequency is equal to 400 KHz resulting in a separation between the carriers equal to 1.5625 KHz, which thus renders the frequencies orthogonal (OFDM modulation). The modulation coefficients associated with the unused carriers are equal to 0.
The OFDM signal in the time domain is generated at the output of the processing means MTFI, and means MCP adds a cyclical prefix to each OFDM symbol in the time domain, which is a header copy of the OFDM symbol of a certain number of samples situated at the end of this symbol. By way of example, in the PLC-G3 standard, the length of the cyclical prefix is 30 samples for a sampling frequency of 400 KHz.
The signal is subsequently converted in a digital/analog converter CNA then processed in a stage ETA, commonly denoted by those skilled in the art under the term “Analog Front End”, where it notably undergoes a power amplification, prior to being transmitted over the electrical power line LE. In reception, it can be seen, referring more particularly to
This analog input stage ET1 conventionally comprises a band pass filter BPF, a low-pass filter LPF, together with amplification means or an amplifier AMP. The output of the stage ET1 is connected to an analog/digital conversion stage CAN whose output is connected to the input of a processing means or stage ET2. The processing stage ET2 here comprises automatic gain control means or an automatic gain control circuit AGC allowing the value of the gain of the amplification means AMP of the stage ET1 to be controlled.
The signal SAC delivered at the output of the analog stage ET1 and at the input of the analog/digital conversion stage CAN represents a channel analog signal coming from the transmission channel (electrical power line) LE. The processing stage ET2 also comprises a low-pass filter LPF2 followed, although this is not indispensable, by under-sampling means or an under-sampling circuit MSCH. The sampling frequency of the signal upstream of the means MSCH is denoted Fs, whereas the sampling frequency of the signal at the output of the means MSCH is denoted Fss.
The signal SNC at the output of the means MSCH then here denotes a channel digital signal which comes from the analog/digital conversion of the channel analog signal SAC and to which notably a synchronization processing, channel estimations and decoding of symbols will be applied, as will be seen in more detail hereinafter. The channel estimations are carried out once the synchronization has been acquired.
The frequency Fc denotes the processing frequency at which the various processing operations will be carried out. In the PLC-G3 standard, for example, the sampling frequency Fs specified is 400 KHz for a size of FFT of 256. Although it would have been possible to perform all of these various processing operations at a processing frequency Fc equal to the sampling frequency Fs of 400 KHz, under-sampling the signal at a frequency Fss lower than Fs and performing all the operations at the processing frequency Fc equal to Fss allows the complexity of implementation of the processing stage to be reduced and also allows a processing operation for direct fast Fourier transform (FFIT) to be carried out which has a reduced size with respect to the specified size of 256.
Before returning in more detail to the various means/circuitry that are incorporated within the processing stage ET2, reference is now more particularly made to
The preamble sequence PRM of the frame TRM allows the receiver to synchronize itself, in other words to obtain an indication IND1 allowing the structure of the frame to be recovered in order to be able to identify the start of the header HD. If reference is now again made to
Amongst these various means/circuitry are conventionally found synchronization means or a synchronizer MSYNC allowing the receiver to synchronize itself, in other words to obtain the indication IND1 allowing the structure of the frame to be recovered in order to be able to identify the start of the header HD. These synchronization means may have a conventional structure known to the skilled person or else, as a variant, those incorporating the filtering means or a filter described in the French Patent application No. 1552588. Amongst the other means/circuitry incorporated into the sub-stage 20 are found detection means MDET configured for carrying out a detection of the state of the transmission channel using the digital signal SNC in such a manner as to detect whether the transmission channel is considered as being LTI or else as being linear and cyclostationary.
When the transmission channel is considered as linear and cyclostationary for the whole duration of a frame, for example, the processing means also comprise generation means MLB configured for generating, as will be seen in more detail hereinafter, a sequence of several transfer functions for the transmission channel respectively associated with N successive time slices. Decoding means MDCD, with a conventional structure known to the skilled person, are then configured for decoding at least some of the symbols of a frame, notably the symbols of the payload PLD of the frame, by using for each of these symbols the transfer function associated with the time slice containing the symbol. Here again, this decoding will be considered in more detail hereinafter.
On the other hand, when the transmission channel is considered as LTI for the whole duration of a frame, the processing means advantageously comprise an averaging module MMY configured for carrying out a processing operation for averaging the plurality of transfer functions for the channel delivered by the detection means MDET in such a manner as to obtain a mean transfer function. The decoding means then decode the symbols of the payload PLD of the frame advantageously using this mean transfer function. Here again, this functionality will be considered in more detail hereinafter.
If the reference is now more particularly made to
The thirteen symbols FCHi and the two symbols S1 and S2 here form M reference symbols SYMRi (M is equal to 15 in this example). These reference symbols will be used for detecting the state of the transmission channel. The transmission channel is a linear channel, in other words it behaves as a linear filter. However, the channel may be considered as LTI or cyclostationary (i.e. LPTV). The LTI or cyclostationary (LPTV) state of the channel can vary from one frame to another, but this state is considered as unchanged during one frame.
When the transmission channel becomes cyclostationary during the reception of one or more frames, owing, for example, to the connection of a halogen lamp or of an object comprising voltage rectifiers, onto the electrical power line, the channel is periodic in amplitude and in phase in a synchronized manner with the alternating current or voltage present on the electrical power line. In other words, for a cyclostationary channel, the characteristics of its transfer function vary in the course of the cyclostationary period of the channel but are subsequently repeated periodically.
For this reason, the set of the 15 reference symbols SYMR1-SYMR15 extend in time over a total duration equal to 10.42 ms, which is therefore here slightly longer than the cyclostationary period PC. In other words, in the present case, the cyclostationary period PC is not an integer multiple of the duration of a reference symbol. The phase for detecting the state of the channel will now be described in more detail referring more particularly to
As illustrated in
More precisely, if the mean phase difference DPHM is less than the threshold TH, then the channel is considered as being of the LTI type during the frame in question, whereas if the mean phase difference DPHM is greater than threshold TH, the channel is considered as cyclostationary (LPTV) during this frame. In the case of a receiver according to PLC-G3, the threshold TH could, for example, be of the order of 0.75 radians.
Reference is now more particularly made to
More precisely, this transfer function Hi is equal to the product of the reference symbol received SYMRi with the complex conjugate of the corresponding symbol transmitted by the transmitter over the transmission channel. In the present case, conforming to the PLC-G3 standard, this transfer function Hi is in fact a complex vector having 36 complex components respectively corresponding to the 36 tones of the symbol. At the end of the step 80, 15 transfer functions H1-H15 are therefore obtained respectively corresponding to the 15 reference symbols SYMR1-SYMR15.
It would then be possible to only determine at this stage a single phase difference between two of these transfer functions. However, it is helpful to determine such a phase difference between two transfer functions spaced out over time by a minimum number of reference symbols, for example, 6 reference symbols, so as to better determine this phase difference.
However, it is even more advantageous to determine several phase differences based on several different pairs of transfer functions and to calculate the average so as to obtain the mean phase difference DPHM. This will now be described in more detail. The step 81 determines the phase difference between two transfer functions Hi and Hk by simply taking the product of Hi with the complex conjugate of Hk.
Since a minimum number of symbols equal to 6 has been chosen, 9 doublets can be formed: H1, H7; H2, H8; H3, H9; H4, H10; H5, H11; H6, H12; H7, H13; H8, H14 and H9, H15. Of course, it would be perfectly possible not to use these nine doublets in order to obtain nine phase differences. It is nevertheless assumed, in this embodiment, that the nine doublets are used in such a manner as to obtain nine phase differences DPH1-DPH9.
In the step 82, the average of the absolute values of these phase differences is calculated in such a manner as to obtain the mean phase difference DPHM. In the embodiment that has just been described, it has been assumed that the sampling frequency for the transmission was identical to the sampling frequency of the channel digital signal SNC. However, as indicated hereinbefore, the sampling frequency of the digital signal SNC may be different from the initial digital signal generated in the transmitter.
In this case, a shift in the sampling frequency results, known by those skilled in the art as “sampling frequency offset”, that needs to taken into account in the estimation of the transfer functions Hi. This is illustrated schematically in
Then, the sampling frequency offset is estimated using, for example, two base transfer functions that are relatively widely spaced out in time, for example, the transfer functions HB1 and HB13. The phase shift resulting from this is then obtained by taking the product of the transfer function HBi with the complex conjugate of the transfer function HB13, dividing it all by the number of symbols. A phase shift correction DPHC is then obtained, which is applied in a step 801 for correcting the transfer functions HBi and obtaining the transfer functions Hi. The estimation of a transfer function based on each of the symbols S1 and S2 can be carried out without difficulty since the transmitted symbols corresponding to the received symbols S1 and S2 are known.
On the other hand, this is not the case for the reference symbols of the header, in other words the symbols FCHi. Nevertheless, as indicated hereinbefore, these symbols FCHi have been coded at transmission in a differential and particularly robust manner. Their decoding does not therefore require the knowing the transfer function of the channel. The
The decoding means also comprise “de-mapping” means supplying, for each carrier, a value of the corresponding modulation coefficient (bin). These de-mapping means are followed by a module configured for determining, for each modulation coefficient, a confidence indicator (or “soft decision”) for the value. This module is conventional and known per se and uses, for example, an algorithm of the LogMAP type.
The decoding means also comprise de-interlacing means or a de-interlacer followed by a decoder, for example, a decoder of the Viterbi type, followed by means/circuitry designed to carry out a parity check. The output of these means/circuitry is connected to the output terminal BS of the sub-stage ET20 which is connected to the means/circuitry forming the MAC layer of the receiver. Since the various symbols FCHi are referenced with respect to the preceding symbols in the frame, it is necessary to decode all the symbols FCHi of the header with the aforementioned decoding means. Then, after verifying that the parity check is correct, the various decoded symbols FCHDi can be obtained. Then, in the step 101, a re-encoding of each of these symbols FCHDi is carried out using a convolutional encoder, an interlacer and a mapping means analogous to the corresponding means/circuitry ENC, INTL, MP illustrated in
It will be noted, with regard to the above, that the operations remain in the frequency domain. Re-encoded symbols FCHECi corresponding to the transmitted symbols are then obtained. In the step 80, analogous to the step 80 in
Reference is now more particularly made to
For this purpose, the M transfer functions Hi determined during the phase for detecting the state of the channel (
Accordingly, the total duration D of all the time slices TR1-TR7 is equal to 9.73 ms, which is slightly less than the cyclostationary period PC (10 ms). In the step 110, in order to obtain the seven transfer functions HFk, fourteen transfer functions Hi are chosen from amongst the fifteen transfer functions obtained during the detection phase. The transfer functions H2-H15 will, for example, be chosen and, in this case, the transfer function HF1 is equal to the average of the transfer functions H2 and H3. The transfer function HF2 is equal to the average of the transfer functions H3 and H4, and so on, until the transfer function HF7 is obtained which is equal to the average of the transfer functions H14 and H15.
As illustrated in
In other words, once the complete cycle of the seven transfer functions HF1-HF7 have been used, the cycle is repeated starting with the use of the transfer function HF1 up to HF7 and so on. However, when the duration D of the N time slices is different from the cyclostationary period, there will be a progressive time shift between the cycle of the transfer functions HF1-HF7 and the cyclostationary period of the channel. In other words, at a given moment in time, there will be a time difference equal to a time slice between the cycle of the transfer functions and the variation cycle of the channel. In this successive and cyclical use of the transfer functions, a transfer function shift should then be applied. This is illustrated schematically in
In the present illustrative example, with a duration D equal to 9.73 ms and a cyclostationary period of 10 ms, this shift is applied after five transfer function cycles and for the execution of the sixth cycle, the transfer function HF7 will be used a second time instead of directly using the transfer function HF1. Of course, after having used the transfer function HF7 a second time, the cycle of the transfer functions HF1-HF7 is continued.
It may also happen that the value of the cyclostationary period, which is directly linked to the period of the alternating signal flowing in the electrical power line, is time shifted owing to a time shift of the period of the alternating electrical signal on the line. This may be corrected by way of a reference signal ZCS (
Reference is now more particularly made to
Other features relating to communications are disclosed in co-pending U.S. patent application Ser. No. 15/042,326, filed on Feb. 12, 2016, and incorporated herein by reference in its entirety.
Number | Date | Country | Kind |
---|---|---|---|
1556487 | Jul 2015 | FR | national |
This application is a continuation of U.S. application Ser. No. 15/042,429, filed on Feb. 12, 2016, which is based upon prior filed co-pending French Application No. 1556487 filed Jul. 9, 2015, both of which are incorporated herein by reference in its entirety.
Number | Date | Country | |
---|---|---|---|
Parent | 15042429 | Feb 2016 | US |
Child | 15797947 | US |