DATA TRANSMITTER WITH VARIABLE PUNCTURE DEVICE

Description

Field of invention

[0001] The present invention relates to the field of telecommunications. Within this field, the invention relates more particularly to methods for transmitting data from a first telecommunications device to a second device with puncturing of the data after coding and before transmission by the antenna. It applies in particular to portable telecommunications devices which establish communication via a base station or an access point. Prior art

[0002] A radio access network generally consists of several base stations or access points which allow a user equipment (UE: User Equipment according to English terminology), also called a terminal or UE, to have access to a telecommunications network and to establish communication to exchange data.

[0003] The communications transmission medium is commonly called the transmission or propagation channel, originally in reference to an aerial channel and by extension in reference to any channel. Wireless systems have a so-called RF transmission interface when it is a telecommunications system with aerial transmission of a signal belonging to a radio band (for example, 5G NR, 4G, GSM, UMTS, IEEE 802.11x, IEEE 802.16e, etc.). Among these systems, we distinguish between so-called mobile cellular access systems, more specifically specified by 3GPP, and originally non-mobile systems, which include those based on a WiFi standard specified by the IEEE.

[0004] The transmitted data may be subject to disruption due to interference introduced by the transmission channel and / or due to noise sources.

[0005] To combat interference, it is known to protect the transmitted data with a certain level of protection. Thus, a widely known and used technique is to add redundancy to the data using an encoder, sometimes called a channel encoder.

[0006] However, adding redundancy goes against the useful throughput that can be transmitted between two devices. In addition, the coding efficiency of the channel encoder may have to meet certain constraints, for example, be fixed in a telecommunications standard published by a group, for example 3GPP, IEEE. In some cases, puncturing must be implemented after coding to adapt the efficiency.

[0007] Patent applications US 2017 / 244588 A1, US 2008 / 010582 A1, US 2005 / 172202 A1 and US 2003 / 145273 A1 disclose technological background on the subject of bit puncturing within encoding. Statement of the invention

[0008] The invention proposes a communication method aimed at improving the protection of transmitted data.

[0009] The subject of the invention is a method for transmitting data implemented by a first telecommunications device to a second telecommunications device.

[0010] The process is such that it includes: encoding input data to add redundancy using an encoder, puncturing the data after encoding, outputting the data after puncturing, and is such that the punching is according to at least two different protection levels and is such that it comprises at least one punching step comprising a 1 era punching matrix whose output feeds a demultiplexer having a 1 era and a 2 e exits, the 2 e output feeding a 2 nde punching matrix to define the at least two different protection levels obtained respectively with the 1 era output of the demultiplexing and the output of the 2 nde punching die.

[0011] The invention further relates to telecommunications equipment intended to communicate with a second equipment. The equipment comprises: an encoder for encoding input data, a punch for punching the data according to at least two levels of protection after encoding, the punch comprising: an elementary structure comprising a 1 era punching matrix whose output feeds a demultiplexer having a 1 era and a 2 e exits, the 2 e output feeding a 2 nde punching matrix to define the at least two different protection levels obtained respectively with the 1 era output of the demultiplexer and the output of the 2 nde punching die.

[0012] The input data is encoded by a single encoder to add redundancy and generate an undifferentiated bitstream.

[0013] The data punching according to the invention is carried out, after coding, by implementing at least two punching matrices separated by demultiplexing. The demultiplexing makes it possible to distribute the output data, i.e. associated with different protection levels, to obtain a determined bit rate on each of these outputs. Such a method makes it easy to determine two different coding rates associated with the two outputs. This method is much less complex than known techniques with several parallel codings to deliver high bit rates.

[0014] According to one embodiment, the data transmission method is such that any punching step called the previous step is followed by a new punching step whose 1st matrix is ​​common with the 2nd matrix of the previous step, to define an additional different level of protection obtained with the output of the 2nd punching matrix of the new punching step.

[0015] According to one embodiment, the puncher is such that any elementary structure called the previous structure is followed by a new elementary structure whose 1st matrix is ​​common with the 2nd matrix of the previous elementary structure, to define an additional different level of protection obtained with the output of the 2nd punching matrix of the new elementary structure.

[0016] According to these previous embodiments of the method and equipment, each addition of a punching step / elementary structure makes it possible to obtain a new coding yield different from the previous ones and associated with the new level of protection. This mode thus makes it possible to increase the punching order with great simplicity.

[0017] According to one embodiment, a puncturing ratio is associated with each puncturing matrix and the ratios are determined for an identical bit rate on each of the outputs of the 2nd puncturing matrices of the puncturing steps / elementary structures, for a determined modulation order and for a determined coding efficiency.

[0018] According to one embodiment, a punching ratio is associated with each punching matrix and a change in data protection level is achieved by modifying the ratio of at least one of the punching matrices.

[0019] According to one embodiment, the data transmission method further comprises: mapping the data after puncturing onto a symbol among M symbols of a constellation of order q, M=2 q< q>=2, each symbol of the constellation comprising at least two bits of different weight among q bits and the method is such that, for at least one of the M symbols, the two different protection levels correspond to a different puncturing of the data depending on the weight of the bits in the symbol onto which these data are mapped and such that the transmission of the data to the second equipment occurs after mapping. According to one embodiment, the telecommunications equipment further comprises: a modulator for mapping the data after puncturing onto a symbol among M symbols of a constellation of order M, M=2 q< q>=2, each symbol of the constellation comprising at least two bits of different weights, the puncturing device being such that the puncturing of the data is different depending on the weight of the bits in a symbol onto which the data are mapped.

[0020] According to these previous embodiments of the method and equipment, post-coding puncturing distinguishes between coded data according to which bits of a symbol of the constellation they are mapped to. According to a representation of the constellation along two perpendicular axes that define four quadrants, the most significant bit of a set of bits mapped to a symbol is by convention the one located to the left of the set. However, an x ​​2 < QAM modulation can be constructed by combining two in-phase and quadrature x-order amplitude modulations, one carried by the I axis and the other by the Q axis in the baseband representation of a digital modulation. In these cases, there is a most significant bit for each set of bits mapped to a symbol for each x-AM modulation even if the two sets form a binary code globally mapped to the same symbol of the x 2 < QAM modulation.The puncturing applied to the data after encoding is such that it allows to distinguish at least two different levels of protection associated respectively with the data mapped onto bits of different weights of a symbol of the constellation. Thus, unlike conventional techniques, the different levels make it possible to obtain non-uniform protection (UEP UnEqual Protection) within the same symbol of the constellation, that is to say that the protection is different between certain bits of the same symbol of the constellation.

[0021] The invention makes it possible in particular to provide better protection to data mapped onto low-weight bits compared to higher-weight bits, whether intermediate or high-weight.

[0022] Thus, by exploiting the fact that between bits of different weights those of higher weights are more robust than those of lower weights to a decision error during decoding to distinguish two different levels of protection between these bits of different weights, the method advantageously makes it possible to maintain the same data transmission bit rate while guaranteeing better protection for the transmitted data.

[0023] According to one embodiment, each symbol of the constellation comprises at least one high-order bit and one low-order bit and the puncturing of the data is different between data mapped to the high-order bit and data mapped to the low-order bit of the same symbol.

[0024] According to one embodiment, the difference in punching between data mapped to the same symbol occurs for all symbols in the constellation.

[0025] According to one embodiment, the difference in punching between data mapped to the same symbol occurs only for some of the symbols in the constellation,

[0026] According to one embodiment, the puncturing of the data is further a function of the position of the symbol in the constellation onto which this data is mapped after puncturing.

[0027] In this mode, the variable protection also takes into account the position of the symbol in the constellation. In the presence of phase noise, the transmitted symbols undergo phase rotation.

[0028] And a phase error that exceeds a threshold δθ induces a change of decision region for the symbols at the periphery of the constellation during the decoding of these symbols at reception. Thus, the method makes it possible in particular to increase the protection of complex symbols at the periphery of the constellation and consequently to specifically combat particular noise sources such as phase noise type which more particularly affect these symbols at the periphery.

[0029] The invention thus makes it possible to limit interference due to these noise sources.

[0030] According to one embodiment, the position of the symbol is evaluated by calculating a metric for evaluating the distance of the symbol from the center of the constellation.

[0031] In this mode, calculating the symbol's distance from the center of the constellation gives the possible positions of the symbol on a circle with a radius equal to the calculated distance. The symbols for which the calculated distance is the greatest correspond to peripheral symbols. These symbols are the most sensitive to phase rotation.

[0032] The invention further relates to a method of communication between a first device and a second telecommunications device with reception of a data signal by the second device, the data being mapped before transmission onto symbols of a constellation of order M, M=2 q< q>=2, each symbol of the constellation comprising at least two bits of different weights. The communication method is such that it comprises: symbol demodulation with data demapping, data depuncturing after data demapping, data puncturing at transmission being different depending on the weight of the bits in a symbol onto which the data is mapped, data decoding after depuncturing.

[0033] According to one embodiment, the de-punching of the data takes into account a punching on transmission depending on the position of the symbol in the constellation on which this data is mapped.

[0034] The invention further relates to telecommunications equipment intended to communicate with a first equipment to receive a data signal, the data being mapped before transmission onto symbols of a constellation of order M, M=2 q< q>=2, each symbol of the constellation comprising at least two bits of different weights. The equipment is such that it comprises: a demodulator for demodulating symbols with data demapping, a de-punching device for de-punching data after data demapping, the data punching on transmission being different depending on the weight of the bits in a symbol onto which the data is mapped, a decoder for decoding the data after de-punching.

[0035] The invention further relates to a computer program on an information medium, said program comprising program instructions adapted to the implementation of a method according to the invention when said program is loaded and executed in telecommunications equipment.

[0036] The invention further relates to an information medium comprising program instructions adapted to the implementation of a method according to the invention, when said program is loaded and executed in telecommunications equipment.

[0037] The invention further relates to a digital signal comprising data transmitted by a first device to a second device, the transmitted data being mapped onto symbols of a constellation, each symbol of the constellation comprising at least two bits of different weights, the data having been punched differently before mapping according to the weight of the bits in a symbol onto which this data is mapped. List of figures

[0038] Other characteristics and advantages of the invention will appear more clearly on reading the following description of embodiments, given as simple illustrative and non-limiting examples, and the appended drawings, among which: [ Fig 1 ] There Figure 1 is a diagram of an embodiment of a transmission chain according to the invention, [ Fig 2 ] There Figure 2gives a pure binary coding and a Gray coding carried out on k=4, k=3 and k=2 bits, [ Fig 3 ] There Figure 3 is a representation of a 16-QAM modulation in an (I,Q) frame which respects a Gray coding with illustration of decision regions, [ Fig 4 ] There Figure 4 is a diagram of an embodiment of a reception chain according to the invention, [ Fig 5 ] There Figure 5 is a representation of a 64-QAM modulation in an (I,Q) frame which respects a Gray coding with illustration of decision regions, [ Fig 6 ] There Figure 6 is a representation of a 16-QAM modulation in an (I,Q) frame which respects a Gray coding with identification of the positions furthest from the center and indication of an angle θ, [ Fig 7 ] There Figure 7is a representation of a 16-QAM modulation in an (I,Q) frame with illustration of decision regions and with indication of the impact of a rotation of an angle θ on the position of certain symbols, [ Fig 8 ] There figure 8 is a diagram of an embodiment of the elementary structure of the puncher according to the invention connected to the output of the encoder to obtain two levels of protection, [ Fig 9 ] There figure 9 is a diagram of an embodiment of a cascade assembly of the elementary structure of the puncher according to the invention to obtain three levels of protection, [ Fig 10 ] There Figure 10 is a diagram of an embodiment of the elementary structure of the puncher according to the invention connected to the output of the encoder with a diagram of an equivalent representation of this structure with the encoder, [ Fig 11 ] There Figure 11is a representation of a 64 QAM modulation with the distinction of three zones Z1, Z2, Z3 of symbols according to their distance from the center of the constellation. Description of particular embodiments

[0039] The general principle of the invention is based on a puncturing of the data to be transmitted with the application to the data of at least two successive puncturing matrices separated by a demultiplexing which makes it possible to distinguish at least two different levels of protection for these data. The puncturing device which implements this puncturing comprises a basic elementary structure comprising the 1 st puncturing matrix followed by a demultiplexer with two outputs itself followed on its 2 nd output by the 2 nd puncturing matrix. The at least two different levels of protection are obtained respectively with the 1 st output of the demultiplexer and the output of the 2 nd puncturing matrix. The proposed multi-level puncturing structure allows a reduction in the complexity of a multi-level channel coding structure at transmission while allowing downstream binary signal coding.

[0040] The two different protection levels can be associated respectively with the data mapped onto different weight bits of the symbols of a constellation associated with a digital modulation. In other words, within the same symbol of the constellation, depending on whether a data item is mapped onto a high-order bit or onto a low-order bit, then it does not benefit from the same puncturing ratio. The difference in protection level between data mapped onto the same symbol can be implemented according to the invention for all the symbols or for only one symbol or certain symbols of the constellation.

[0041] Alternatively or in addition, two different protection levels can be associated respectively with the data mapped onto symbols of a constellation located at different positions. In other words, depending on whether a data item is mapped onto a symbol positioned at a certain point of the constellation or onto a symbol positioned at another point of the constellation, then it does not benefit from the same puncturing ratio. The difference in protection level between data mapped onto different symbols of the constellation can be implemented according to the invention for all the symbols or for only certain symbols of the same constellation associated with a digital modulation.

[0042] There Figure 1is a diagram of an embodiment of a transmission chain for implementing a method according to the invention. This transmission chain is part of a telecommunications device which can be either a base station SB or a terminal Tal such as a smartphone. The chain comprises at least a COD channel coder, a POIN punch and a MAP modulator.

[0043] The method 10 of transmitting data is implemented by the device SB / Tal. The method comprises at least the coding 11 of the input data, the puncturing 12 of the data after coding, the modulation 13 with mapping of the data after puncturing onto a symbol among M symbols of a constellation of order M, = 2 q< , q ≥ 2, each symbol of the constellation comprising at least two bits of different weights. M, q integers.

[0044] The method is such that, for at least one of the M symbols, the puncturing of the data is different depending on the weight of the bits in the symbol on which this data is mapped.

[0045] The transmission 14 of data after mapping is to another telecommunications equipment which can be either a terminal or a base station.

[0046] The COD channel encoder performs the coding 11 of the input binary data coming from an information source which can be a microphone of a mobile terminal as well as a local or remote application such as a short message application (SMS), a multimedia content transmission application. The encoder introduces a binary redundancy to the input data with a coding efficiency Ri to output binary data with a certain bit rate. The POIN puncher performs the punching 12 of the data after channel coding. The punching of the data after coding is carried out by implementing several punching matrices with a demultiplexer between two matrices. Punching makes it possible to increase the useful bit rate of the information for a fixed bit rate.

[0047] The MAP modulator modulates the data after puncturing to generate complex modulated symbols associated with modulation as output.

[0048] The transmitter EM transmits the data after modulation via a transmit antenna ANT_E in the form of a signal Yes. The signal transmitted by the ANT_E antenna generally comes from the modulation of a so-called RF (Radio Frequency) carrier by the modulating signal which carries the data. Other processing operations can of course take place in the chain before or after each of the processing operations illustrated by the Figure 1 .

[0049] The encoded data is arranged into blocks of binary data and the modulation by the MAP modulator maps this binary data onto the symbols of a constellation to construct the signal that is involved in the modulation of the RF carrier. The data mapping can respect a so-called Gray coding.

[0050] Thus, the modulating signal of the RF carrier described in baseband carries the information to be transmitted and is represented in the form of complex symbols distributed in the (I,Q) plane onto which the data has been mapped.

[0051] Each symbol in the constellation includes at least one high-order bit and one low-order bit. The MAP modulator maps the data onto one of M symbols in a constellation of order M, M≥ 4 respecting the constraint that data mapped to the most significant bit and data mapped to the least significant bit of the same symbol have been differently punctured. The mapping thus covers an operation called binary-to-symbol coding which can be described as a transformation of a binary set {b k-1 , b k-2 ,..., b 0} into a symbol S c of the constellation. The transformation makes it possible to construct M-ary symbols with a binary representation of k=log2(M) bits, {b k-1 , b k-2 ,..., b 0}, corresponding to a binary word with M = 2 k< . According to one implementation, the M-ary symbols of the constellation take their value in the space of relative integers Z consisting of M elements taken from the alphabet A={±1, ±3, ...± (2p+1), ...,± (M-1)}, according to an NRZ coding which has the advantage of being more favorable for transmission. k an integer.

[0052] The transformation generally follows a Gray coding and a specific combination of bits is assigned to each relative integer of the alphabet A. A Gray coding is deduced from a so-called pure binary coding. A pure binary coding is based on the operations of addition and multiplication in the Galois field formed by the integers {0,1} where the addition corresponds to the logical operation 'Exclusive OR' and the multiplication to the logical operation 'AND'. The increment in the M-ary alphabet of a relative number (2p+1) to 2(p+1)+1 is achieved by adding (addition Exclusive OR (⊕ )) a low-order bit equal to 1 on the current binary codeword to generate the new binary codeword associated with the symbol 2(p+1)+1. p an integer.

[0053] The most significant bit (MSB or Most Significant Bit b k-1 ) is the bit, in a given binary representation, having the greatest weight or the greatest position (the one on the left in the usual positional notation), it is the bit most robust to state transitions from one symbol to another. The least significant bit (LSB or Least Significant Bit) is the bit, in a given binary representation, having the least weight or the smallest position (the one on the right in the usual positional notation). It corresponds to the elementary unit of variation (of state) of a symbol. This notion of bit weight in a symbol relates to the mode of construction of a pure binary coding where the robustness to state transitions increases with the weight of the bit in each code word, that is to say with its position in the code word.

[0054] Gray coding is a specific coding derived from pure binary coding that changes two successive binary codewords by only one bit (a codeword is the binary representation of the M-ary symbol). This Gray code minimizes transition errors from one codeword to another when the state index (index of the M-ary symbol) is incremented by 1. The transition from pure binary coding to Gray coding is done by performing an Exclusive OR (⊕ ) operation on the bits of a pure binary codeword.

[0055] By designating by bn (b 0 = LSB, least significant bit) any bit in pure binary code and by G n the bit sought in Gray code in position n, for a given code word, the bit G n following a Gray coding, is obtained as follows: G n = b n ⊕ b n + 1 ; G MSB = b MSB where bn and b n+1 are two bits in the same codeword of a pure binary encoding at positions n and n+1 and of weight n and n+1 respectively. The most significant bit is the leftmost bit in the representation of a pure binary encoding.

[0056] At reception, Gray decoding to regenerate a pure binary code is carried out by performing the following operation: b n = G n ⊕ b n + 1

[0057] The weight of bits in a symbol is ranked according to its robustness to a binary-to-symbol decoding error.

[0058] [Table 1] and [Table 2] in the Appendix give an illustration of pure binary coding and Gray coding for M=16 (and k=4) and M=8 (and k=3) respectively. The most significant bits remain unchanged in pure binary coding and Gray coding for the symbol indices varying from 0 to 7 and from 8 to 15, which gives them greater robustness during a transition from one state to another.

[0059] The pure binary coding and Gray coding operations performed on k=4, k=3 and k=2 bits can be deduced from each other by their mode of construction because the weights of the bits are decreasing from left to right as illustrated by the Figure 2 In other words, the hierarchy and nesting of the coding according to the number of bits per symbol appears explicitly in the construction of the code words according to the weight of the bit and its position in the code word.

[0060] A so-called MAQ-M=2 2k< =N 2< modulation can be constructed by combining two amplitude modulations of order N=2 k< in phase and in quadrature, one carried by the I axis and the other by the Q axis in the baseband representation of a modulation. For each of the two modulations, the abscissas S n,i and orderly S n,q of each symbol S n take their values ​​from the alphabet A={±1, ±3, ...± (2p+1), ...,± (N-1)}. S n = S n , i + j S n , q

[0061] The combination of the two modulations generates the M-state QAM modulation to which corresponds a constellation of M symbols S m . The complex symbols of the signal modulating the RF carrier are each formed of 2xlog2(N) bits (b 2k-1 , b 2k-2 ,... bk , b k-1 ,..., b 0} where {b 2k-1 , b 2k-2 ,... bk} describes the binary coding of the in-phase amplitude modulation and {b k-1 , b k-2 ,..., b 0} describes the binary coding of the quadrature amplitude modulation as given in [Table 3] in the Appendix. The symbols are distributed in the I, Q plane in such a way that adjacent symbols differ by only one bit in accordance with the Gray coding at a given position relative to the symbol with index 0 as illustrated by the Figure 3 for 16-QAM modulation. According to this construction, the I and Q channels are modulated and demodulated independently. Concept of decision regions at reception

[0062] There Figure 4is a diagram of an embodiment of a reception chain for implementing a communication method according to the invention. This reception chain is part of a telecommunications device which can be a Tal terminal such as a smartphone or a base station SB. This chain performs at least the inverse functions of those for transmission illustrated by the Figure 1 This reception chain includes at least a DEMAP demodulator, a DEPOIN depuncturer and a DECOD decoder.

[0063] The data signal Sr received 21 by a receiver RE via the receiving antenna ANT_R corresponds to the signal Se emitted by the transmitting antenna ANT_E after transmission by the propagation channel: the data carried by the transmitted signal having been mapped before transmission onto symbols of a constellation of order M, M=2 q< q>=2, each symbol of the constellation comprising at least two bits of different weights, and the puncturing of the data before mapping being different depending on the weight of the bits in the symbol onto which the data are mapped.

[0064] The communication method 20 comprises at least the demodulation 22 of the symbols with demapping of the data, the de-punching 23 of the data after demapping of the data, the decoding 24 of the data after de-punching by the Tal / SB device.

[0065] The DEMAP demodulator performs a demapping function 22 inverse to that implemented by the MAP modulator.

[0066] The DEPOIN puncher performs a punching function 23 which is the opposite of that implemented by the POIN puncher in the sense that it allows the punched bits to be restored.

[0067] The DECOD decoder performs a 24-bit inverse decoding function of the encoding function implemented by the COD encoder.

[0068] The demodulation 22 implemented by the DEMAP demodulator aims to determine from a received point the most probable transmitted symbol. The received points are affected by thermal noise but also by noise of various origins such as phase noise, which generate a variation in the position of the complex symbols in the constellation by phase rotation. During this demodulation operation 22, the method performs for example a maximum likelihood detection. According to this mode of implementation, the method determines the symbol closest to the observation (received symbol) according to a criterion of Euclidean distance between the symbols of the constellation and the received symbol.

[0069] Decision making on the closest symbol can be illustrated with the notion of decision regions. Such decision regions are shown schematically in the Figure 3in the case of a distribution of symbols in the constellation which respects the Gray coding (two adjacent symbols differ by only one bit). The decision region associated with a complex symbol as a whole is the intersection of the decision regions attached to each bit forming the symbol knowing that this symbol is positioned in the constellation according to its index specified during the binary coding operation.

[0070] According to the illustration of the Figure 3, rectangle A0 represents the decision region associated with the most significant bits for the quadrature 4-QAM amplitude modulation i.e. along the Q axis. That is to say that in this region the most significant bit of each of the symbols of this modulation is at one while all the symbols outside this region have their most significant bit at zero. Rectangle B0 represents the decision region associated with the most significant bits for the in-phase 4-QAM amplitude modulation i.e. along the I axis. The intersection of the two rectangles A0 and B0, i.e. the region (AB0), provides the decision region for the most significant bits of the 16-QAM modulation in which each symbol has two most significant bits relating to each of the 4-QAM modulations. This decision region AB0 is the one where the two most significant bits of a 16-QAM symbol are at one according to this construction with two 4-QAM modulations.

[0071] Rectangle A1 represents the decision region associated with the least significant bit for MA-4 amplitude modulation along the Q axis.

[0072] Rectangle B1 represents the decision region associated with the least significant bit for MA-4 amplitude modulation along the I axis.

[0073] The intersection of the two rectangles A1 and B1, i.e., region AB1, provides the decision region for the low-order bits of the 16-QAM modulation according to this construction with two 4-AM modulations.

[0074] During digital demodulation, the estimated symbol meets the maximum likelihood criterion such that the decision regions considered for each symbol correspond to the intersection of the decision regions of the bits forming the symbol. These intersections correspond to those of the least significant bits.

[0075] The demodulation 22 according to the invention which performs binary symbol decoding makes it possible to weight the estimation error of the transmitted bits associated with the variation of the decision regions according to the weight and possibly the position of the bit in the constellation.

[0076] [Table 4] in the Appendix and the Figure 5 are related to a 64-QAM. According to the illustrated construction, 64-QAM modulation is generated by combining in phase and quadrature two eight-state amplitude modulations MA-8 carried by the I and Q channels respectively. Each MA-8 amplitude modulation is characterized by three bits of different weight (or levels) associated with their positions during binary coding. This results in three different types of zones for the delimitation of the decision regions of the bits transmitted during the binary-to-symbol decoding operation.

[0077] Each 64-AQM symbol consists of 6 bits {b k-1 , b k-2 ,..., b 0} with k=6. The first three bits describe the in-phase 8-AQM and the last three bits describe the quadrature 8-AQM.

[0078] On the Figure 5 the decision regions AB 0 associated with the most significant bits MSB of position {G 5 ,G 2} result from the intersection of the regions of type A 0 and B 0 .

[0079] The dotted square areas AB 1 represent the delimitation of the decision region for the intermediate weight bits corresponding to the bits in position [G 4 ,G 1 ]. This is the intersection of the two decision regions A 1 and B 1 associated with the intermediate weight bits for the MA-8 in-phase and quadrature modulations.

[0080] The intersection of the decision regions of type A 2 and B 2 , i.e., the decision region AB 2 , provides the decision region for the low-order bits LSB of the 64-QAM modulation. These regions of type AB 2 coincide with the decision regions associated with the complex symbol S m of the 64-QAM modulation.

[0081] THE figures 6 And 7 represent a 16-QAM. Periphery symbols are shown surrounded by a square on the Figure 6 , they are those whose abscissas and ordinates are the largest in absolute value. In the presence of phase noise, the transmitted symbols undergo a phase rotation. A phase error that exceeds a certain threshold induces a change in the decision region for the symbols on the periphery of the constellation when decoding these symbols at reception. This change therefore induces an error when making a decision on the transmitted symbol which occurs during digital demodulation.

[0082] Let us consider a phase rotation δθ corresponding to the angle between on the one hand the median D 1 passing through the center O of the constellation and connecting two symbols of the constellation whose distances are equal on the in-phase and quadrature components of the digital modulation and on the other hand the straight line D 2 starting from the center O of the constellation and passing through the nearest adjacent symbol and corresponding to an adjacent decision region.

[0083] The symbols surrounded by a square on the Figure 6 are the symbols most sensitive to phase rotation δθ, they change decision region after rotation. These symbols lie on a circle of radius R 2: R 2 = 3 2 a with 2 a the width of a decision region.

[0084] The same phase rotation δθ results in indeterminacy on the decision region for symbols surrounded by a circle and located at an intermediate distance R1 from the center O of the constellation. Indeed, the rotation, with a precision of the order of 9%, positions one of these symbols received at the intersection of the lines delimiting four decision regions (points I and I' on the Figure 7 ). This positioning can therefore generate an indeterminacy on reception of the emitted symbol. The radius R1 is given by: R 1 = 10 a

[0085] This same rotation δθ on the symbols closest to the center of the constellation and located at a distance R0 from the center of the constellation ( R 0 = 2 a ), does not cause a change in decision region as illustrated in the Figure 7 . Elementary structure of the puncher

[0086] An embodiment of a puncher is shown in figure 8 .

[0087] According to this mode, the POIN puncher comprises an elementary structure. The elementary structure delimited by the broken line has one input Ei and two outputs Si and Si+1. It is made up of a first punching module Pi with one input Ei and one output E'i, followed by a demultiplexer Mi [1:2] (one input E'i to two outputs Fi and Fi+1) whose output branch Fi+1 is connected to a second punching module Pi+1. The input of the second punching module Pi+1 is Fi+1 and its output is denoted F'i+1. The demultiplexer Mi is intended to distribute on its two outputs Fi and Fi+1 the data intended for the branches Si and Si+1 in such a way that the output rates are controlled on each of the branches. Each punching module Pi, Pi+1 implements a punching matrix, denoted Pi, Pi+1 as the corresponding punching module, the size of which depends on the punching ratio of the module.

[0088] The POIN punch then includes two outputs Si and Si+1 having protection levels that can be different from each other depending on the setting of the punching matrix Pi+1, i.e. at most two different protection levels. If this punching matrix Pi+1 contains at least one zero then the two outputs Si and Si+1 have protection levels that are different from each other.

[0089] According to a determined choice, the data on the output Si which corresponds to the output Fi of the demultiplexer Mi are mapped onto the low-order bits. According to this same choice, the data on the output Si+1 which corresponds to the output F'i+1 of the second puncturing module Pi+1 are mapped onto the high-order bits. Thus, according to this embodiment of the invention, the input data Ei are punctured differently depending on whether they are mapped onto low-order bits or onto high-order bits of a symbol of the constellation.

[0090] The Pi matrix leads to uniform punching on both outputs Si and Si+1 when the Pi+1 matrix is ​​inactive i.e. is formed only of “ones”.

[0091] According to one implementation mode, the previous puncher with two outputs, therefore with at least one elementary structure, can make it possible to obtain more than two levels of protection. In particular, a first parameterization of the punching matrix Pi+1 makes it possible to obtain two levels of protection on the two outputs Si and Si+1 and a second parameterization of the punching matrix Pi+1 makes it possible to obtain two other levels of protection on the two outputs Si and Si+1.

[0092] Thus, by only modifying the setting of the punching matrix Pi+1 it is possible to obtain different pairs of protection levels on the two outputs Si and Si+1, whether the matrix Pi is active or not.

[0093] According to one embodiment, the telecommunications device SB / Tal intended to communicate with a second device Tal / SB comprises a POIN puncher with the elementary structure described above. The method 10 for transmitting data implemented by this embodiment of the device SB / Tal is such that the punching 3 of the data is according to at least two different protection levels respectively for data mapped onto bits of different weights of the same symbol. And more particularly, the method is such that the punching 3 comprises at least one punching step comprising a 1st punching matrix whose output feeds a demultiplexing having a 1st and a 2nd output, the 2nd output feeding a 2nd punching matrix to define the at least two different protection levels obtained respectively with the 1st output of the demultiplexing and the output of the 2nd punching matrix.

[0094] According to another embodiment illustrated by the figure 9, the POIN puncher comprises a first elementary structure followed by at least one second elementary structure cascaded in a nested manner called forward-backward cascade casc-av-arr. According to this mode, the first punching module of the second structure, i.e. the matrix Pi+1, is common with the second punching module of the first structure. The POIN puncher then comprises three outputs Si, Si+1, Si+2 having protection levels which may be different from each other, i.e. at most three different protection levels. If the punching matrix Pi+2, i.e. the distinctive added matrix, contains at least one zero then the two outputs Si+1 and Si+2 have protection levels which are different from each other. If the punching matrix Pi+1 contains at least one zero then the output Si has a protection level which is different on the one hand from the protection level of the output Si+1 and on the other hand from the protection level of the output Si+2.

[0095] The addition of a new elementary structure delimited by a broken line on the figure 9 allows to obtain an additional level of protection which can be distinct from the other levels of protection depending on the setting of the non-common punching matrix, Pi+2, of the added structure.

[0096] The forward-backward cascade architecture with the resumption of the previous punching module allows the punching order (i.e. the number of punching levels) to be increased with great simplicity and flexibility. An additional punching level is obtained by adding a single stage of the elementary structure and configuring the puncher according to the protection considered. The elementary structure mounted in forward-backward cascade simplifies the adaptation of the punching according to the order M of the modulation.

[0097] According to one embodiment, the SB / Tal telecommunications device intended to communicate with another Tal / SB device comprises a POIN puncher with at least two elementary structures according to a forward-backward cascade assembly described above. Each elementary structure is associated with a punching step of the transmission method.

[0098] The method 10 for transmitting data implemented by this embodiment of the SB / Tal device is such that it comprises a first punching step comprising a 1st punching matrix whose output feeds a demultiplexer having a 1st and a 2nd output, the 2nd output feeding a 2nd punching matrix to define the at least two different protection levels obtained respectively with the 1st output of the demultiplexer and the output of the 2nd punching matrix. And the method is such that any punching step called the previous step, therefore in particular the first punching step, is followed by a new punching step whose 1st matrix is ​​common with the 2nd matrix of the previous step, to define an additional different protection level obtained with the output of the 2nd punching matrix of the new punching step.

[0099] Thus, when the puncher comprises two elementary structures in forward-backward cascade, the transmission method implemented comprises two punching steps. Each additional elementary structure of the puncher adds a punching step to the method implemented.

[0100] A puncher has four outputs, therefore with at least three elementary structures in forward-backward cascade, it can be used to assign distinct protection to data depending on the position of the symbol on which these punched data are mapped. This protection depending on the position of the symbol is added to a distinct protection between data mapped on the most significant bit and data mapped on the least significant bit of the same symbol. In particular, two levels of protection can be reserved for data mapped on a periphery symbol of the constellation, therefore depending on the position of the symbol, with a distinction between these two levels depending on whether the data is mapped on a most significant bit or a least significant bit of this symbol.

[0101] Thus, the invention makes it possible to jointly differentiate the protection within a symbol by taking into account the position of the bit in the code word and according to the position of the symbol in the constellation.

[0102] To assign distinct protection to punctured data based on the position of the symbol to which that data is mapped, the parameterization of the forward-backward cascade structure of the puncturer can take into account a metric for evaluating the distance of the complex symbol from the center of the constellation.

[0103] According to one embodiment of the invention, the parameterization of the puncher determines the number of elementary structures in front-back cascade to distinguish protection levels between the symbols according to their position in the constellation evaluated by the metric. Determination of punching dies

[0104] The COD encoder performs, for example, convolutional or LDPC channel coding with efficiency R i , the code is called the yield mother code R i .

[0105] Considering the elementary structure of the punch illustrated by the figure 8 , the punching ratio R pi greater than or equal to 1, designates the ratio between the number of bits at the input of the punching module Pi and the number of bits at the output of this module Pi. This ratio is deduced from the punching matrix Pi which specifies the number and position of the bits to be punched and therefore not transmitted by the punching module Pi. By convention, by virtue of the decoding downstream of the unpunching, the punched bits are designated by the integer '0' in the matrix Pi and the transmitted bits are designated by the integer '1' in this matrix Pi. The punching ratio R pi is written: R pi = N in , i / N out , i ≥ 1 Or N in,i denotes the number of bits at the input of the Pi punching module and N out,i denotes the number of bits at the output of this Pi module. According to a simple implementation, N in,i represents the total number of elements of the punch matrix Pi and N out,i represents the number of '1' elements in the punch matrix Pi.

[0106] More generally, the punch matrix Pi has a number of rows that is a multiple of the inverse of the yield R i of the mother code. The number of columns is a multiple of the numerator of the desired code output R if after punching such as: R si = R i × N in , i / N out , i R if is therefore the channel coding efficiency on branch i, output Si, after puncturing by the puncturing module Pi.

[0107] Determining the channel coding efficiency on branch i thus makes it possible to determine at least one puncturing matrix which makes it possible to obtain the ratio R pi .

[0108] The demultiplexing performed by the demultiplexer Mi consists of adjusting the number of bits on each of the output branches Si and Si+1 taking into account the punching matrices Pi and Pi+1. This adjustment is determined step by step, following the optimization constraints of the input and output rates of the puncher.

[0109] According to one embodiment of the invention, the bit rate is set identically on each of the branches at the output of the elementary structure: D Yes = D If +1 = D b / nb, with nb the number of branches at the punch output, nb = 2 according to the example detailed below.

[0110] The demultiplexer Mi distributes the bits on the branches Fi and Fi+1 taking into account the ratio R pi +1 of the punching matrix Pi+1. The bit rates at different points of the elementary structure are then calculated as follows: D Si = D Si + 1 = D Fi = D F ′ i + 1 = D b / 2 D Fi + 1 = D F ′ i + 1 × N in , i + 1 / N out , i + 1 = D F ′ i + 1 × R pi + 1 = R pi + 1 × D b / 2 D E ′ i = D Fi + 1 + D Fi = 1 + R pi + 1 × D b / 2

[0111] Bit rate D Ei at the output of the channel encoder is expressed as a function of the bit rate at the output of the elementary structure as follows: D Ei = D E ′ i × R pi = R pi × 1 + R pi + 1 × D b / 2

[0112] Since the Pi punching matrix punches the bits of the input stream uniformly, changing its parameter setting alone allows the couple of yields on the two outputs to be modified. According to a first parameter setting, the Pi punching matrix can be formed only of '1' so as not to perform any punching on the bits coded at the output of the channel encoder. These bits can for example correspond to the information bits from a systematic encoder, information bits which are generally not punched. According to another parameter setting, the Pi punching matrix can be used for example to reduce the redundancy associated with certain symbols depending on their position in the constellation. Thus, a first parameter setting of the Pi punching matrix, for example R pi = 1, gives a pair of protection levels for all peripheral symbols. And a second setting, for example R pi> 1, gives another pair of protection levels for all symbols except the peripheral ones. Thus, the puncturing module Pi which uniformly punctures the bits can allow additional puncturing of the data mapped to the symbols closest to the center O of the constellation.

[0113] There Figure 10 gives a diagram of two embodiments of the invention.

[0114] According to the first mode, the POIN puncher comprises an elementary structure as already described with regard to the figure 8 The puncher punches the data after encoding by a COD encoder. The output data of the puncher can be represented by two streams corresponding to the two outputs Si and Si+1 or can be represented as a single stream grouping the two outputs.

[0115] According to the second mode which is equivalent to the first mode from a channel coding point of view, the input data are first demultiplexed by a demultiplexer M'i to distinguish two outputs. The first output of the demultiplexer M'i is encoded by a first COD channel coder. The output data of the first COD channel coder are punctured by a first matrix puncturing module Pi. The second output of the demultiplexer M'i is encoded by a second COD channel coder identical to the first COD channel coder. The output data of the second COD channel coder are punctured by a second matrix puncturing module Pi. The data punctured by this second matrix puncturing module Pi are again punctured by a third matrix puncturing module Pi+1.

[0116] This second mode allows the calculation of the equivalent channel coding efficiency R eq,elemof the elementary punching structure with channel coding, between input point A and output point B after serial connection of the two puncher outputs.

[0117] The equivalent channel coding efficiency is deduced from equation (10): R eq , elem = R i × R pi × 1 + R pi + 1 2

[0118] An equivalent punching ratio can be defined for the elementary structure: R peq , elem = R eq , elem / R i = R pi × 1 + R pi + 1 2 The equivalence of the structures of the two modes allows the ratios to be adjusted R pi And R pi +1 for a given COD channel encoder yield R i in order to generate a target equivalent channel coding efficiency for the constellation symbols.

[0119] The equivalent channel coding efficiencies on branches Si and Si+1 are given by: R Si = R i × R pi et R Si + 1 = R i × R pi × R pi + 1 The values ​​of the couple { R pi , R pi +1} for target coding efficiency R eq,elem are adjusted according to the desired granularity of the flow rates.

[0120] The equivalence of the structures of the two modes illustrated on the Figure 10 illustrates the reduction in complexity provided by the elementary multi-level puncturing structure when a parallelization of the coding structures is chosen in order to limit the coding processing speed associated with each branch. Such a choice of parallelization of the coding can be implemented for very high data rate systems, typically for systems operating in millimeter band or in the THz band. The proposed multi-level puncturing structure allows a reduction in the complexity of a multi-level channel coding structure at transmission while ensuring downstream a binary coding with a multiple protection signal. Puncher has two elementary structures

[0121] For the puncher illustrated by the figure 9ie with two elementary structures, and in the case where the three outputs Si, Si+1 and Si+2 have the same bit rate, the rates at different locations can be expressed as follows: D Si = D Si + 1 = D Si + 2 = D G ′ i + 2 = D Gi + 1 = D Fi = D b n b avec n b = 3 D Gi + 2 = D G ′ i + 2 × N in , i + 2 N out , i + 2 = D G ′ i + 2 × R pi + 2 = R pi + 2 × D b 3 D F ′ i + 1 = D Gi + 2 + D Gi + 1 = 1 + R pi + 2 × D b 3 D Fi + 1 = D F ′ i + 1 × N in , i + 1 N out , i + 1 = D F ′ i + 1 × R pi + 1 = R pi + 1 × 1 + R pi + 2 × D b 3 D E ′ i = D Fi + 1 + D Fi = 1 + R pi + 1 × 1 + R pi + 2 D b 3 D Ei = R pi × D E ′ i = R pi 1 + R pi + 1 × 1 + R pi + 2 D b 3

[0122] The equivalent channel coding efficiencies on each of the three branches are therefore given by: R Si = R i × R pi ; R Si + 1 = R i × R pi × R pi + 1 ; R si + 2 = R i × R pi × R pi + 1 × R pi + 2

[0123] The overall equivalent channel coding efficiency is therefore given by: R eq = R i × R pi × 1 + R pi + 1 × 1 + R pi + 2 3

[0124] The equivalent punching ratio for the structure is expressed in the form: R peq = R pi × 1 + R pi + 1 × 1 + R pi + 2 3 Puncher for J elementary structures

[0125] A puncher that includes J elementary structures in forward-backward cascade provides J+1 outputs. If the bit rate is set the same as D b J + 1 on each of the outputs also called branches then the flow rate at the input of each non-uniform demultiplexer is calculated step by step taking into account the punching ratio of each punching module. The equivalent channel coding efficiencies on each branch are therefore given by: R Si = R i × R pi ; R Si + l = R i × ∏ m = 1 l R pi × R pi + m l > 0 with R pi + l = N in , i + l N out , i + l , l = 0 , … J + 1 .

[0126] And the equivalent channel coding efficiency is given by: R eq = 1 J + 1 × ∑ l = 0 J R Si + l = R i × R pi 1 + ∑ l = 1 J ∏ m = 1 l R pi + m J + 1

[0127] Calculating the flow rate D Ei at the punch entry can then be written in the form: D Ei = R eq R i × D b

[0128] The puncher's equivalent punching ratio is then expressed in the form: R peq = 1 J + 1 × 1 R i × ∑ l = 0 J R Si + l = R pi × 1 + ∑ l = 1 J ∏ m = 1 l R pi + m J + 1 Examples of setting the punch during implementations of the invention

[0129] For the following examples, the considered telecommunications devices are compatible with an IEEE 802.1 1ax standard. The COD encoder performs convolutional coding of rate ½ and it has a constraint length L=7. The considered target MCS (Modulation and Coding Scheme) 16-QAM ¾, 64-QAM ¾. 64-QAM 5 / 6 are specified by this standard.

[0130] According to a first use case, the punch implemented according to the invention modifies the protection of the bits within each symbol as well as the protection of the symbols furthest from the center of the constellation while keeping unchanged the code rate fixed by the MCS.

[0131] According to a first example, the MCS of index 4 is a 16-QAM ¾.

[0132] The MCS is constant for all symbols in the constellation, which means that the useful information flow rate is not modified by the invention and is identical for all points in the constellation. This therefore requires generating several pairs ( R p 1 , R p 2) as a function of the distance R from the center of the constellation.

[0133] Two distinct levels are considered depending on the distance, on the one hand for the distances R0, R1 and on the other hand for the distance R2. A puncher with an elementary structure (two branches / two outputs {S1, S2}) and two distinct settings can therefore be suitable since two protection levels are obtained for each of the two settings. The two outputs of the puncher make it possible to distinguish between data mapped to the most significant bit and data mapped to the least significant bit of the same symbol.

[0134] The two settings correspond to two pairs of values ​​( R p 1 , R p 2) respectively for the points located at distances R0 and R1 and for those located at distance R2 from the center of the constellation. The constraint of an unchanged MCS efficiency R whatever the point of the constellation imposes a set of two pairs of values ​​( R p 1 , R p 2) .

[0135] The performance of the channel encoder mother code R i = 1 / 2 and the MCS allow us to deduce from equation (11) that: R eq , elem = R MCS = 3 4 = 1 2 × R p 1 × 1 + R p 2 2 d ′ où R p 1 × 1 + R p 2 = 3 Several values ​​of R p 1 and R p 2 can be selected depending on the protection chosen based on the distance to the center of the constellation. The ratios R p 1 and R p 2 are greater than or equal to one under the puncturing operation. Channel coding yieldsR S 1 and R S 2 on each of branches S1 and S2 are less than or equal to one. For points at distance R2.

[0136] For example if R p 1 = 6 5 SO R p 2 = 9 6 = 3 2 .

[0137] From equation (12), the equivalent channel coding efficiencies on branches S1 and S2 are given by: R S 1 = R i × R p 1 = 1 2 × 6 5 = 3 5 R S 2 = R i × R p 1 × R p 2 = 1 2 × 6 5 × 3 2 = 9 10

[0138] Branch S1 is assigned to the low-order bits with an equivalent efficiency of 3 / 5 while branch S2 is assigned to the high-order bits with an efficiency of 9 / 10. The overall efficiency is indeed equal to ¾.

[0139] The associated punching matrices are formed of two rows and can have three columns. They can be in the form: P 1 = 1 1 1 0 1 1 et P 2 = 1 0 1 1 1 0 For points located at distances R0 and R1.

[0140] A second couple ( R p 1 , R p 2) is determined by considering a yield RS 1 higher for the low-order bits than for the points at distance R2 because the decision region of the symbols is the same as that of the low-order bits.

[0141] For example R S 1 = 2 3 > 3 5 :

[0142] SO R S 1 = 2 3 = R i × R p 1 = 1 2 × R p 1 from where R p 1 = 4 3

[0143] And as R p 1 × (1 + R p 2) = 3 then R p 2 = 5 4 = 10 8

[0144] The yields on each branch are therefore: R s 1 = 2 3 et R s 2 = R i × R p 1 × R p 2 = 1 2 × 4 3 × 5 4 = 5 6

[0145] The corresponding punching dies can have the form: P 1 = 1 1 0 1 et P 2 = 1 1 1 0 1 1 0 1 1 1

[0146] [Table 5] in the Appendix gives a summary of the values ​​for this example.

[0147] According to a second example, the modulation is a 64-QAM ¾ and the puncturing according to the invention is implemented for only certain symbols, those at the distance R2, i.e. those furthest from the center. The MCS remains constant for all the symbols. [Table 6] in the Appendix gives a summary of the values ​​for this example. The method implements two pairs of puncturing and therefore of equivalent channel coding efficiency, one pair for the data mapped onto the symbols at the distances R0 and R1, one pair for the data mapped onto the symbols at the distance R2. The table gives two values ​​(2 / 3, 5 / 6) (3 / 5, 9 / 10) of the pair of equivalent channel coding efficiency for the data mapped onto the symbols at the distance R2. This example illustrates that the same input data can be punctured differently, using the two puncturing pairs (4 / 3, 10 / 8) and (6 / 5, 6 / 4), while obtaining the same equivalent code yield, ¾ according to the example.These same data punched differently make it possible to obtain diversity in the transmitted data.

[0148] In a third example, the modulation is 64-QAM ¾. Considering that the 64-QAM modulation is built with two 32-QAM modulations, then three bits are associated with each symbol of an 8-QAM constellation. Three different protection levels can therefore be used to protect the data differently depending on which of the three bits it is mapped to. [Table 7] gives a summary of the values ​​for this example.

[0149] The chosen puncher comprises two elementary cascaded structures which make it possible to obtain three outputs with three equivalent coding efficiencies ( R S 1 , R S 2, R S 3) and three punching ratios ( R p 1 , R p 2, R p 3) .

[0150] Three zones Z1, Z2, Z3 of symbols are distinguished as illustrated by the Figure 11 depending on their distance from the center of the constellation. To protect these three areas, three different triplets of equivalent coding efficiencies are determined ( R S 1 , R S 2, R S 3) and punching ratio ( R p 1 , R p 2, R p 3) .

[0151] According to the first use case, the code rate set by the MCS is kept unchanged for all points of the constellation regardless of their position.

[0152] The performance of the channel encoder mother code R i = 1 / 2 and the MCS allow us to deduce from equation (11) that: R eq = R MCS = 3 4 And R eq = 1 2 × R p 1 × 1 + R p 2 + R p 2 × R p 3 3 From where: R p 1 × 1 + R p 2 + R p 2 × R p 3 = 9 2

[0153] For example if R p 1 = 1 then it is possible to choose Rp 2 = 6 / 4 and R p 3 = 14 / 12.

[0154] The equivalent channel coding efficiencies on each of the branches are then given by: R S 1 = R i × R p 1 = 1 2 , R S 2 = R i × R p 1 × R p 2 = 3 4 et R S 3 = R i × R p 1 × R p 2 × R p 3 = 1 2 × 6 4 × 14 12 = 7 8

[0155] Branch S3 is assigned to the high-order bits, branch S2 to the middle-order bits, and branch S1 to the low-order bits.

[0156] The punching ratio set (1, 6 / 4, 14 / 12) is considered for symbols in zone Z3.

[0157] The punching ratio set (9 / 8, 4 / 3, 10 / 8) is considered for the symbols in the Z2 zone.

[0158] The punching ratio set (4 / 3, 19 / 16, 1) is considered for symbols in zone Z1.

[0159] [Table 7] in the Appendix gives a summary of the values ​​for this previous example.

[0160] According to a second use case, the punch implemented according to the invention keeps unchanged on average over all the symbols of the constellation the yield R MCS code fixed by the MCS. According to this use case, the useful bit rate assigned to the different symbols of the constellation can be different depending on their distance from the center of the constellation while keeping constant the average bit rate on the scale of the M symbols of the modulation of order M.

[0161] For example, a 16-QAM ¾ modulation is considered with a higher useful rate for the closest symbols of the constellation, for example at distances R0 and R1 and a lower useful rate for the most distant symbols, for example at distance R2 while keeping the average rate unchanged on the M points of the constellation. This amounts to defining two equivalent channel coding efficiencies R eq ,1 and R eq,2 such that the average yield weighted by the numbers of constellation points affected by these two yields meets the following equation: R eq = N 1 R eq , 1 + N 2 R eq , 2 N 1 + N 2 = R MCS N 1 + N 2 = M N 1 is the number of points in the constellation having the equivalent yield R eq ,1 and N 2 is the number of points in the constellation having the equivalent yield R eq, 2 .

[0162] For example, N 1 = 12 and N 2 = 4. So, assuming R eq ,1 > R eq ,2 we must solve the equation 12 × R eq , 1 + 4 × R eq , 2 = 3 4 × 16 = 12 .

[0163] For example, R eq , 1 R eq , 2 = 5 6 1 2 which amounts to not punching the bits mapped to the most distant symbols in the constellation located in zone 2, i.e. at distance R2.

[0164] For zone 1, i.e. the symbols at distances R0 and R1, the efficiency is equal to 5 / 6. The method can uniformly apply the standard puncturing to all bits mapped to the symbols in zone 1. Or alternatively, the method can apply variable puncturing within each symbol in zone 1 such that: R eq , 1 = 5 6 = 1 2 × R p 1 × 1 + R p 2 2 → R p 1 × 1 + R p 2 = 10 3

[0165] Several values ​​satisfy the above equation. For example, R p 1 R p 2 = 3 2 11 9 Appendix

[0166] [Table 2] symbol index 3-bit binary coding 3-bit Gray coding b 2 b 1 b 0 G 2 G 1 G 0 0 0 0 0 0 0 0 1 0 0 1 0 0 1 2 0 1 0 0 1 1 3 0 1 1 0 1 0 4 1 0 0 1 1 0 5 1 0 1 1 1 1 6 1 1 0 1 0 1 7 1 1 1 1 0 0 [Table 3] Symbol index 16-ary symbol for 16 QAM modulation b 3 b 2 Track I b 1 b 0 Track Q 0 00 -3 00 -3 1 01 -1 01 -1 2 11 1 11 1 3 10 3 10 3 [Table 4] Symbol index 64-ary symbol for 64 QAM modulation b 5 b 4 b 3 Track I b 2 b 1 b 0 Track Q 0 000 -7 000 -7 1 001 -5 001 -5 2 011 -3 011 -3 3 010 -1 010 -1 4 110 1 110 1 5 111 3 111 3 6 101 5 101 5 7 100 7 100 7 [Table 5] Modulation 16-QAM 3 / 4 Bit mapping (b 3 ,b 1 )→ R s 2 (b 2 ,b 0 )→ R s 1 Distance ( R p 1 , R p 2 ) ( P 1, P 2) ( R S 1 , R S 2 ) R eq Pi (number of lines, number of columns) R 1 = 10 a 4 3 10 8 P 1 = (2,2) 2 3 5 6 3 / 4 R 0 = 2 a 1 punched element P 2 = (2,5) 2 punched elements R 2 = 3 2 a 6 5 6 4 P 1 = (2,3) 3 5 9 10 3 / 4 1 punched element P 2 = (2,3) 2 punched elements [Table 6] Modulation 16-QAM 3 / 4 Bit mapping (b 3 ,b 1 )→ R s 2 (b 2 ,b 0 )→ R s 1 Distance ( R p 1 , R p 2 ) ( P 1, P 2) ( R s 1 , R s 2 ) R eq Pi (number of lines, number of columns) R 1 = 10 a 6 4 1 P 1 Punching die of standard (2,3) 3 4 3 4 3 / 4 R 0 = 2 a P 2 formed from 1 (1,1) R 2 = 3 2 a 4 3 10 8 P 1 = (2,2) 2 3 5 6 3 / 4 1 punched element P 2 = (2,5) 2 punched elements R 2 = 3 2 a 6 5 6 4 P 1 = (2,3) 3 5 9 10 3 / 4 1 punched element P 2 = (2,3) 2 punched elements [Table 7] 64-QAM 3 / 4 (b 5 ,b 2 )→ R S 3 (b 4 ,b 1 )→ R S 2 (b 3 ,b 0 )→ R S 1 Distance ( R p 1 , R p 2 , R p 3 ) ( P 1 , P 2 , P 3) ( R S 1 , R S 2 , R S 3 ) Pi (number of lines, number of columns) Zone 3 1 6 4 14 12 P 1 = (2,1) 1 2 3 4 7 8 0 punched elements P 2 = (2,3) 2 elements punched with the standard die P 3 = (2,7) 2 punched elements with a die to be defined Zone 2 9 8 4 3 10 8 P 1 = (2,9) 9 16 3 4 15 16 2 punched elements P2 and P3 can be the matrices given in Example 1 Zone 1 4 3 , 19 16 , 1 2 3 , 19 24 19 24

Claims

1. Method (10), implemented by a first telecommunications equipment (SB / Tal), for transmitting data to a second telecommunications equipment (Tal / SB), comprising: - coding (11) input data to add redundancy by way of an encoder, - puncturing (12) the data after coding, - transmitting (14) the data after puncturing, and such that the puncturing is in accordance with at least two different levels of protection and such that it comprises at least one puncturing step comprising a first puncturing matrix the output of which is supplied to a demultiplexing stage having a first and a second output, characterized in that the second output is supplied to a second puncturing matrix so as to define the at least two different levels of protection obtained with the first output of the demultiplexing stage and the output of the second puncturing matrix, respectively.

2. Data transmission method (10) according to Claim 1, such that any puncturing step, referred to as previous step, is followed by a new puncturing step whose first matrix is common with the second matrix of the previous step, so as to define an additional different level of protection that is obtained with the output of the second puncturing matrix from the new puncturing step.

3. Data transmission method (1) according to either of Claims 1 and 2, such that, with a puncturing ratio being associated with each puncturing matrix, these ratios are determined for an identical bit rate on each of the outputs of the second puncturing matrices of the puncturing steps, for a determined order of the modulation and for a determined coding rate.

4. Data transmission method (1) according to one of Claims 1 to 3, such that, with a puncturing ratio being associated with each puncturing matrix, levels of protection of the data are changed by modifying the ratio of at least one of the puncturing matrices.

5. Data transmission method (10) according to one of Claims 1 to 4, furthermore comprising: - mapping (13) of the data after puncturing onto a symbol from among M symbols of a constellation of order q, M=2q q>=2, each symbol of the constellation comprising at least two bits of different weight from among q bits, the method being such that, for at least one of the M symbols, the two different levels of protection correspond to a different puncturing of the data depending on the weight of the bits in the symbol, onto which bits these data are mapped, and such that the transmission (14) of the data to the second equipment (RE) takes place after mapping.

6. Data transmission method (10) according to the preceding claim, such that each symbol of the constellation comprises at least one most significant bit and one least significant bit and such that the puncturing (12) of the data is different between data mapped onto the most significant bit and data mapped onto the least significant bit of one and the same symbol.

7. Data transmission method (1) according to one of Claims 5 to 6, such that the puncturing difference between data mapped onto one and the same symbol applies for all of the symbols of the constellation.

8. Data transmission method (1) according to one of Claims 5 to 6, such that the puncturing difference between data mapped onto one and the same symbol applies only for some of the symbols of the constellation.

9. Data transmission method (10) according to one of Claims 5 to 8, such that the puncturing (12) of the data is also dependent on the position of the symbol in the constellation onto which these data are mapped after puncturing.

10. Data transmission method (10) according to the preceding claim, such that the position of the symbol is evaluated by computing a metric for evaluating the distance from the symbol to the center of the constellation.

11. Method (20) for communication between a first equipment (SB / Tal) and a second telecommunications equipment (Tal / SB) with reception (21) of a data signal by the second equipment (Tal / SB), the data being mapped before transmission onto symbols of a constellation of order M, M=2q q>=2, each symbol of the constellation comprising at least two bits of different weight, said method comprising: - demodulating (22) the symbols with demapping of the data, - depuncturing (23) the data after demapping of the data, the puncturing of the data at transmission being different depending on the weight of the bits in a symbol, onto which bits the data are mapped, - decoding (24) the data after depuncturing.

12. Communication method (20) according to the preceding claim, such that the depuncturing (23) of the data takes into account a puncturing at transmission dependent on the position of the symbol in the constellation onto which these data are mapped.

13. Telecommunications equipment (SB / Tal) intended to communicate with a second equipment (Tal / SB), comprising: - an encoder (COD) for coding input data, - a puncturer (POIN) for puncturing the data in accordance with at least two levels of protection after coding, the puncturer comprising: - an elementary structure comprising a first puncturing matrix the output of which is supplied to a demultiplexing stage having a first and a second output, characterized in that the second output is supplied to a second puncturing matrix so as to define the at least two different levels of protection obtained with the first output of the demultiplexing stage and the output of the second puncturing matrix, respectively.

14. Telecommunications equipment (SB / Tal) according to the preceding claim, the puncturer being such that any elementary structure, referred to as previous structure, is followed by a new elementary structure whose first matrix is common with the second matrix of the previous elementary structure, so as to define an additional different level of protection that is obtained with the output of the second puncturing matrix from the new elementary structure.

15. Telecommunications equipment (SB / Tal) according to either of Claims 13 and 14, further comprising: - a modulator (MAP) for mapping the data after puncturing onto a symbol from among M symbols of a constellation of order M, M=2q q>=2, each symbol of the constellation comprising at least two bits of different weight, the puncturer (POIN) being such that the puncturing of the data is different depending on the weight of the bits in a symbol, onto which bits the data are mapped.