Activation method, retransmission method, receiver device and relay node capable of implementing these methods

Abstract

The invention relates to a method for activating, by means of a receiver device, a retransmission of redundancy versions of information messages from source devices not decoded by the receiver device, comprising: - selecting (E90) a first set comprising at least one undecoded source device and a second set comprising relay nodes knowing the information messages from the source devices of the first set, the sets verifying an evaluated performance criterion taking into account the variations of a precoding matrix applied collectively by the relay nodes to the redundancy versions of the information messages from the source devices of the first set, the matrix being derived from a dictionary known to the receiver device and the relay nodes and varying within this dictionary; and - activating (E100) the relay nodes of the second set so that they retransmit the redundancy versions from the source devices of the first set.

Description

Description Title of the invention: Activation method, retransmission method, receiving device and relaying node capable of implementing these methods. Technical Field

[0001] The invention falls within the general field of telecommunications.

[0002] It relates more specifically to a cooperative communication technique involving a plurality of communicating devices connected to a communication network. Such a technique aims to make communicating devices cooperate with each other to transmit messages to a recipient device rather than competing for the resources of the transmission channel as is the case in traditional communication networks.

[0003] The invention has a preferred, but not limiting, application in the context of 5G or 6G communication networks, as defined by the 3GPP standard. In such a context, the receiving device is, for example, a network base station equipped with several receiving antennas, and the communicating devices cooperating with each other are user equipment (UEs). However, the invention may also be applicable in other contexts. Previous technique

[0004] Figure 1 illustrates an example of a SYS system implementing such a cooperative communication technique. The SYS system is an OMAMRC (Orthogonal Multiple Access Multiple Relay Channel) type system. It comprises M>2 source devices si, s2,..., sM, and one destination device d, and relies on cooperative relaying of messages sent by the source devices si, s2,..., sM. This relaying occurs via relay nodes, which can be source devices and / or dedicated intermediate nodes rl, r2,..., rL, L>0, that know the messages to be relayed. Transmissions within this system use an orthogonal multiple access channel scheme, such as time division multiplexing (TDM).

[0005] The document by S. Cerovic et al. entitled “Efficient Cooperative HARQ for Multi-Source Multi-Relay Wireless Networks”, August 2018, 14 thThe International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob), pages 61-68, proposes an OMAMRC system using a cooperative HARQ (Hybrid Automatic Repeat reQuest) protocol based on single-user incremental or IR (Incremental Redundancy) coding using LDPC (Low Parity Check Codes) or turbo codes such as those used in 3GPP standards. This IR-type HARQ protocol relies on the incremental retransmission of an information message by a source device (called the "source message") that is not correctly decoded by the receiving device, rather than on an identical retransmission of this source message.

[0006] More specifically, in the context of a 5G network, for example, as defined by the 3GPP standard, an information message from a source device is typically encoded with a coding efficiency R, before being mapped to the symbols of a constellation (q bits per symbol) and then transmitted on NI resource elements, or REs (for "Resource Element"). For an OFDMA (Orthogonal Frequency Division Multiple Access) multiplexing technique, such as that used in a 5G network, an RE corresponds to an OFDM symbol in the time domain and a subcarrier in the frequency domain. The size L of the encoded information message is therefore equal to L = Nl.

[0007] In the case of the IR-type HARQ protocol, the information message associated with a source device is first encoded with a low yield R0< R (typically 1 / 3 or 1 / 5), leading to a coded source message size Lc=L / R0 strictly greater than L. This coded information message is then stored in a circular buffer. At each retransmission, the coded bits to be sent are read from fixed positions in the buffer. For example, during the first transmission, the source device reads the N1.q coded bits from position POS1; then during the first retransmission, N2.q coded bits are read from position POS2; during the second retransmission, N2.q coded bits are read from position POS3; during the third retransmission, N2.q coded bits are read from position POS1, and so on. The coded bits sent at each retransmission are called "redundancy versions".

[0008] In an OMAMRC system using the IR-type HARQ protocol, (re-)transmissions are organized according to three phases: an initialization phase, during which the receiving device d determines the modulation and coding scheme to be used by each source device, based on the quality of the transmission channel separating it from each source device; a transmission phase, during which the source devices M si,..., SM successively transmit their respective coded information messages using the modulation and coding schemes determined by the receiving device d during the initialization phase. During this transmission phase, the number NI of resource elements (and incidentally of uses of the transmission channel) is fixed and identical for all source devices; and A retransmission phase occurs during which informational messages that the receiving device d failed to decode are cooperatively retransmitted over a number of time intervals by relay nodes selected by the receiving device d. The relay nodes are selected at each time interval from among source devices and / or intermediate nodes that are aware of the informational messages that the receiving device d could not decode, either because they are their own informational messages or because they themselves successfully decoded them. At each time interval, an informational message associated with a source device that was not decoded by the receiving device d is retransmitted by one or more relay nodes. The maximum number of possible retransmissions is Tmax; in other words, the retransmission phase lasts at most Tmax time intervals.

[0009] To implement the retransmission phase, information is exchanged before the start of each retransmission between the receiving device d and the other nodes of the system (source devices and / or intermediate nodes) via feedback links or channels, which are often limited. This information exchange aims to allow the receiving device d, firstly, to identify the nodes that have knowledge of the information messages it was unable to decode and that can be used as relay nodes for these undecoded information messages, and secondly, to inform the relay node(s) selected for each time interval to retransmit the same information message associated with a source device not decoded by the receiving device.At each retransmission of an information message associated with a source, a new redundancy version resulting from the encoding of the information message in question is used by the selected relay node(s), the different selected nodes if necessary transmitting the same redundancy version.

[0010] Document WO 2023 / 242295 proposes an OMAMRC system based on the principles described above. In this system, the receiving device selects a source information message to retransmit (or equivalently, a source device whose information message is to be retransmitted) in order to maximize the overall signal-to-noise ratio (SNR) received on the receiving NR antennas of the receiving device. Thus, when several nodes know the information message in question and are able to simultaneously retransmit the same redundant version resulting from the encoding of this information message during the retransmission phase, such a system allows for spatial diversity in transmission.

[0011] The choice of the source device to be retransmitted therefore depends on the number of relay nodes capable of retransmitting a redundant version of the information message associated with that source device, as well as the quality of their transmission links with the receiving device. Once the source device is selected, the receiving device informs the system nodes, and the nodes that have correctly decoded the information message associated with that source device simultaneously retransmit the same redundant version resulting from the encoding of that information message. To allow the receiving device to coherently combine the redundant versions received on the receiving NR antennas of the receiving device, the relay nodes implement a transmission technique that maximizes the signal-to-noise ratio (MRT) in reception.This is achieved via the application in transmission of a precoding vector, each coefficient of which is applied by one of the relaying nodes involved in the retransmission phase.

[0012] Thus, the implementation of such an OMAMRC system requires, at each time interval of the retransmission phase, the exchange of a certain amount of information between the receiving device and the other nodes of the system: modulation and coding schemes to be used by the source devices, decoded or undecoded source devices, the source device selected by the receiving device, the precoding vector to be applied to the redundancy version to be transmitted, etc. There is therefore a need to reduce this amount of information. Description of the invention

[0013] The invention addresses this need in particular by proposing an activation mechanism for relay nodes operating in open loop, i.e. without feedback regarding the transmission channel or precoding vectors to be applied to the selected relay nodes (or equivalently the precoding matrix to be applied collectively by the selected relay nodes).

[0014] More specifically, the invention relates to a method for activating, by a multi-antenna device receiving a plurality of information messages from a plurality of source devices, a retransmission using a plurality of resource elements of at least one redundancy version resulting from the encoding of at least one information message from a source device not decoded by the receiving device. This method comprises, for at least one retransmission interval: - a step of selecting a first set comprising at least one source device not decoded by the receiving device and a second set comprising a plurality of relay nodes knowing each information message of each undecoded source device in the first set, said first and second selected sets satisfying a performance criterion evaluated by taking into account variations on said resource elements of a precoding matrix applied collectively by said relay nodes on the same redundancy version of each information message of each source device in the first set, said precoding matrix being derived from a dictionary known to the receiving device and the relay nodes and varying within this dictionary according to a variation pattern known to the receiving device and the relay nodes; and - a step of activating the relay nodes of the second selected set so that they retransmit, during said retransmission interval, said same redundancy version of each information message from each source device of said first selected set.

[0015] Correspondingly, the invention also relates to a multi-antenna device receiving a plurality of information messages from a plurality of source devices, said receiving device being configured to activate a retransmission using a plurality of resource elements of at least one redundancy version resulting from an encoding of at least one information message from a source device not decoded by the receiving device. This receiving device comprises modules activated for at least one retransmission interval and comprising: - a selection module, configured to select a first set comprising at least one source device not decoded by the receiving device and a second set comprising a plurality of relay nodes knowing each information message of each undecoded source device in the first set, said first and second selected sets satisfying a performance criterion evaluated by taking into account variations on said resource elements of a precoding matrix applied collectively by said relay nodes on the same redundancy version of each information message of each source device in the first set, said precoding matrix being derived from a dictionary known to the receiving device and the relay nodes and varying within this dictionary according to a variation pattern known to the receiving device and the relay nodes; and - an activation module, configured to activate the relay nodes of the second selected set so that they retransmit, during said retransmission interval, said same redundancy version of each information message from each source device of said first selected set.

[0016] The selection and activation steps can be repeated for at least one more retransmission interval until a predefined stopping criterion is met, and by considering new redundancy versions at each new retransmission interval. Each retransmission interval allows the receiving device to benefit from additional redundant versions of the information messages from the source devices that it was unable to decode, thus improving its chances of decoding these information messages from the multiple redundancy versions available to it (these information messages are therefore better protected).

[0017] The invention therefore advantageously proposes to select a first set of source device(s) not decoded by the receiving device and a second set of relaying nodes intended to help the receiving device to decode these undecoded source devices, by optimizing a performance criterion evaluated under certain assumptions taking into account the absence of a return path between the receiving device and the relaying nodes in question to transmit information concerning the propagation conditions seen by the receiving device or, equivalently, the precoding matrix to be applied collectively by the relaying nodes to be applied.

[0018] More specifically, the performance criterion is evaluated by the receiving device to select the undecoded source device(s) and the relay nodes that will help the receiving device decode these undecoded source device(s) by traversing a codebook known to the receiving device and the relay nodes. These relay nodes are selected according to a variation pattern known to both the receiving device and the relay nodes. Such a variation pattern might include the frequency of variation of the precoding matrix, such as at each resource element, and / or a way of traversing the codebook at each variation, for example, cyclically. These are, of course, only illustrative examples and are not exhaustive; other variation patterns could be considered as alternatives.

[0019] The dictionary known to the receiving device and the relay nodes contains all the precoding matrices that can be applied by the relay nodes when retransmitting redundant versions of the undecoded source device(s) information messages. By knowing the variation pattern of the precoding matrix within this dictionary based on resource elements, the receiving device can accurately evaluate a relevant performance criterion, allowing it to optimize the selection (e.g., mutual information) of the first and second sets to help the receiving device decode the information messages it was unable to decode, and to activate the relay nodes of the second set without needing to specify the precoding matrix they should apply.Once activated for a retransmission interval, these relay nodes only have to apply the precoding matrices of the dictionary shared between them and with the receiving device, varying them according to the variation pattern considered by the receiving device when evaluating the performance criterion that led to their selection.

[0020] In this way, we can limit the amount of information sent by the receiving device to the relay nodes when they activate them, and apply open-loop precoding, that is, without transmitting the precoding matrix applied globally by the relay nodes (or equivalent information) or an estimate of the channel allowing it to be determined (or equivalent information).

[0021] Thus, for example, the activation step may include sending retransmission instructions comprising (only) the first and second sets and designating, for each relay node of the second set, a line of the precoding matrix to be applied by that relay node (without however transmitting the line of the precoding matrix).

[0022] It should be noted that the invention applies regardless of the number of undecoded source devices selected for the retransmission interval. This number can be equal to 1, in which case using a plurality of relay nodes allows for spatial diversity, or strictly greater than 1, in which case the invention can offer the possibility of spatial multiplexing.

[0023] In a particular embodiment, the second set is provided in the retransmission instructions sent by the receiving device as an ordered sequence of indices identifying the relay nodes of the second set, the order of an index designating the row of the precoding matrix to be applied by the relay node identified by that index.

[0024] This implementation method allows for even greater limitation of the information sent back by the receiving device.

[0025] It is therefore clear that the invention relies on the receiving device and the activation process implemented by the latter, but also on the relaying nodes selected by the receiving device for each retransmission interval.

[0026] According to a second aspect, the invention also relates to a method of retransmission by a relay node selected by a multi-antenna device receiving a plurality of information messages from a plurality of source devices, this method comprising: - a receiving step, from the receiving device, of a retransmission instruction comprising a first and a second set selected by the receiving device for a retransmission interval, the first set comprising at least one so-called undecoded source device, associated with an information message not decoded by the receiving device, and the second set comprising a plurality of relay nodes, including the selected relay node, knowing each information message from each undecoded source device in the first set; and - a retransmission step during said retransmission interval, using the same resource elements as the other relay node(s) of the second set, for each information message associated with each undecoded source device of the first set, of the same redundancy version from the encoding of said information message, by applying a line designated by said retransmission instruction of a precoding matrix applied collectively by the relay nodes of the second set and from a dictionary known to the receiving device and the relay nodes of the second set, the precoding matrix varying within this dictionary according to said resource elements according to a variation pattern known to the receiving device and the relay nodes.

[0027] Correspondingly, the invention also relates to a relay node capable of being selected by a multi-antenna device receiving a plurality of information messages from a plurality of source devices, this relay node comprising: - a receiving module, configured to receive from the receiving device, a retransmission instruction comprising a first and a second set selected by the receiving device for a retransmission interval, the first set comprising at least one source device, said to be undecoded, associated with an information message undecoded by the receiving device, and said second set comprising a plurality of relay nodes, including said selected relay node, knowing each information message from each undecoded source device of the first set; and - a retransmission module, configured to retransmit during said retransmission interval, using the same resource elements as the other relay node(s) of the second set, for each information message associated with each undecoded source device of the first set, of the same redundancy version from the encoding of said information message, by applying a line designated by said retransmission instruction of a precoding matrix applied collectively by the relay nodes of the second set and from a dictionary known to the receiving device and the relay nodes of the second set, the precoding matrix varying within this dictionary according to said resource elements according to a variation pattern known to the receiving device and the relay nodes.

[0028] The retransmission process and the relaying node benefit from the same advantages mentioned above as the activation process and the receiving device according to the invention.

[0029] As mentioned previously, in a particular embodiment of the activation process and / or the retransmission process, the precoding matrix varies every x resource elements, x denoting an integer such that 1 < x < N2, where N2 denotes the number of resource elements used during the retransmission interval.

[0030] The invention can therefore be applied in different contexts, in particular depending on the variations of the propagation channel between the source devices, the relaying nodes and the destination device.

[0031] In a particular embodiment, the dictionary has dimension Kv, where K denotes an integer such that: | / V| K = I— I + N mod v where N denotes the number of relay nodes in the second set and v the number of undecoded source devices in the first set.

[0032] In this embodiment, when N2 > Kvx, the precoding matrix can be obtained by cyclically traversing the dictionary.

[0033] These embodiments allow for different dimension configurations for the dictionary and the sets selected for retransmissions. It is thus possible to store, at the level of the receiving device and the devices that can be selected as relay nodes during a retransmission interval (whether they are source devices or third-party devices with knowledge of the undecoded information messages from the source devices), a plurality of dictionaries for a plurality of possible values ​​of N and v, each dictionary being designed for a particular pair of values ​​(N,v).

[0034] No limitations are attached to the precoding matrices that can be considered in the context of the invention, provided that these precoding matrices belong to a dictionary known to the receiving device and the devices sending the redundancy versions (source devices and relay nodes).

[0035] However, in a preferred embodiment, each precoding matrix is ​​constructed from an N xv dimension matrix W, where N denotes the number of relay nodes in the second set and v the number of undecoded source devices in the first set, and in which each column of the matrix W is a linear combination of distinct vectors from a basis of N orthogonal vectors.

[0036] This embodiment offers a very simple construction technique for dictionary precoding matrices, applicable regardless of the N,v) configuration considered during the retransmission interval.

[0037] Furthermore, this construction of the W matrix from linear combinations of distinct orthogonal vectors forming a basis advantageously allows scanning the space in different directions, thus compensating for the lack of knowledge of the transmitting channel. When the first set comprises a plurality of undecoded source devices, each spatial layer associated with a selected undecoded source device corresponds to a beam formed in a particular direction by the column of the precoding matrix applied to it. Consequently, if a channel corresponding to a given spatial layer (i.e., to an undecoded source device) suffers from significant attenuation, the channels corresponding to other spatial layers (i.e., to other undecoded source devices) can offer better signal quality.This creates diversity in the broadcast, which further enhances the reliability of the transmission.

[0038] This effect is amplified by the variation of the precoding matrix according to the channel usage, i.e., the resource elements. Typically, when the precoding matrix varies for each resource element, the proposed construction allows averaging the phase of the propagation channel coefficients applied to each undecoded source device across the different channel usages, thus providing greater diversity.

[0039] In a particular embodiment, the N orthogonal vectors, denoted w n n = 0,..., N - 1, are defined by: 1 r j2nn j2n2n j2n(N w n = 1 / √N [1 e^(j2πn / N) e^(j2π2n / N) ... e^(j2π(N-1)n / N)]^T

[0040] Such vectors are commonly called "IDFT vectors" (for "Inverse Discrete Fourier Transform"). However, other orthogonal vector bases can be used in the context of the invention. For example, a Hadamard basis, known to those skilled in the art, can be used.

[0041] In a particular embodiment, v - 1 columns of the matrix W can result from a linear combination of |^| distinct basis vectors and a column of the matrix W can result from a linear combination of + N mod v) distinct basis vectors.

[0042] By "linear combination" we mean here any weighted sum of basis vectors as long as at least one weighting coefficient considered in the sum and applied to a basis vector is non-zero.

[0043] This embodiment allows the space to be divided uniformly into groups comprising an identical number (when N is a multiple of v) or a similar number (when N is not a multiple of v) of vectors.

[0044] Furthermore, when all the weighting coefficients are non-zero, we ensure that the entire space is covered by the precoding matrix, since each of the basis vectors is used to construct a precoding vector (i.e., a column of the matrix).

[0045] However, a suboptimal implementation is also possible in which only one weighting coefficient per group is non-zero, for example, the weighting coefficient applied to the vector located in the middle of the group under consideration. This amounts to favoring an average direction for each group and allows for a precoding of constant magnitude.

[0046] Other configurations can be considered, such as, for example, at least two non-zero weighting coefficients per linear combination, or different numbers of vectors w n considered for each column of the JV matrix, etc. The invention offers great implementation flexibility.

[0047] In a particular embodiment, each column of the matrix W is a linear combination of distinct adjacent basis vectors.

[0048] This embodiment is particularly simple to implement since it is sufficient to apply a cyclic shift of the basis vectors taken into account to construct each column of the precoding matrix.

[0049] In a particular embodiment, each dictionary precoding matrix is ​​further constructed from a complex diagonal matrix D of dimensions vxv introducing a diversity of cyclic delays and a complex rotation matrix U of dimensions vxv.

[0050] When v >1, the precoding matrix thus constructed allows us to benefit advantageously from both spatial multiplexing, which improves the spectral efficiency of the system, and from a diversity of cyclic delays (CDD) provided by the diagonal matrix D, which offers the possibility of having a more reliable transmission.

[0051] When the first set includes a single undecoded source device (i.e., v = 1), each precoding matrix can be composed of a single column vector chosen from N orthogonal basis vectors by cyclically traversing said basis, where N denotes the number of relay nodes in the second set.

[0052] In an alternative configuration, i.e. when the first set includes a single undecoded source device (i.e. v = l), each dictionary precoding matrix can be composed of a single vector obtained from a product of a complex diagonal matrix D' of dimensions N x N introducing a diversity of small cyclic delays and a column vector { / ' dimension N whose components are all equal to 1, N denoting the number of relay nodes of the second set.

[0053] The invention thus offers the possibility of constructing a complete precoding matrix dictionary for a large number of N and v value configurations.

[0054] The invention also relates, according to a third aspect, to a communication system comprising: - a multi-antenna receiver device according to the invention, - a plurality of source devices associated with a plurality of information messages intended for the receiving device, and - a plurality of relaying nodes according to the invention, a said relaying node being able to be a said source device or a third device distinct from the source devices.

[0055] The invention has a preferred, but not limiting, application when an orthogonal multiple access multiple-relay channel (OMAMRC) scheme is used between the source devices, the relay nodes, and the destination device. There are no limitations on how orthogonality is ensured between the devices in the system (time, frequency, etc.).

[0056] Furthermore, the invention applies in a preferred but not limiting manner when the receiving device is a base station of a telecommunications network and the source devices and, where applicable, relay devices are user equipment.

[0057] It should be noted that the invention can also be applied in other contexts where retransmission of redundancy versions associated with information messages from source devices is envisaged in a system, regardless of how relay nodes acquire knowledge of the information messages.

[0058] In a particular embodiment, the activation process and / or the retransmission process is / are implemented by a computer.

[0059] The invention also relates to a computer program on a recording medium, this program being capable of being implemented in a computer or more generally in a receiving device conforming to the invention and comprising instructions adapted to the implementation of an activation method as described above.

[0060] The invention also relates to a computer program on a recording medium, this program being capable of being implemented in a computer or more generally in a relay node according to the invention and comprising instructions adapted to the implementation of a retransmission process as described above.

[0061] Each of these programs can use any programming language, and be in the form of source code, object code, or code somewhere between source code and object code, such as in a partially compiled form, or in any other desirable form.

[0062] The invention also relates to an information carrier or a recording medium readable by a computer, and comprising instructions for a computer program as mentioned above.

[0063] The information or recording medium can be any entity or device capable of storing programs. For example, the medium may include a storage means, such as a ROM, for example a CD-ROM or a microelectronic circuit ROM, or a magnetic recording means, for example a hard drive, or a flash memory.

[0064] On the other hand, the information or recording medium can be a transmissible medium such as an electrical or optical signal, which can be carried via an electrical or optical cable, by radio link, by wireless optical link or by other means.

[0065] The program according to the invention can in particular be downloaded onto an Internet-type network.

[0066] Alternatively, the information or recording medium may be an integrated circuit in which a program is incorporated, the circuit being adapted to execute or to be used in the execution of the activation method or the retransmission method according to the invention.

[0067] It can also be envisaged, in other embodiments, that the activation method, the receiving device, the retransmission method, the relaying node and the communication system according to the invention have in combination all or part of the aforementioned characteristics. Brief description of the drawings and the appendix

[0068] Other features and advantages of the present invention will become apparent from the description below, with reference to the accompanying drawings which illustrate an example of an embodiment without being limiting in any way. In the figures:

[0069] [Fig. 1] Figure 1, already described, represents a state-of-the-art cooperative communication system;

[0070] [Fig. 2] Figure 2 represents a communication system according to the invention, in a particular embodiment;

[0071] [Fig. 3] Figure 3 represents the hardware architecture of the receiving device and the relaying nodes of the communication system of Figure 2, in a particular embodiment;

[0072] [Fig. 4] Figure 4 represents the functional modules of the receiving device (Fig. 4A) and the relaying nodes (Fig. 4B) of the communication system of Figure 2, in a particular embodiment;

[0073] [Fig. 5] Figure 5 represents the construction of a matrix used by the relay nodes of the system in Figure 2 for precoding the redundancy versions of the retransmitted undecoded devices, in a particular embodiment;

[0074] [Fig. 6] Figure 6 represents the main steps of an activation process and a retransmission process as implemented respectively by the receiving device and by the relay nodes of the communication system in Figure 2; and

[0075] [Fig. 7] Figure 7 represents an incremental coding implemented in the context of the invention.

[0076] Furthermore, the Annex illustrates an advantage resulting from the construction proposed by the invention for the precoding matrix used by the system in Figure 2. Description of the invention

[0077] Figure 2 represents, in its environment, a communication system 1 according to the invention, in a particular embodiment.

[0078] Communication system 1 includes: a plurality of source devices if, S2,..., SM, M denotes an integer greater than or equal to 2, each source device being capable of sending at least one information message. It is assumed here that each source device comprises a single transmission antenna; and A DEST device, the recipient of information messages from source devices (S1, S2, ..., SM – referred to as the recipient device for simplicity), conforms to the invention. The DEST recipient device is equipped here with a number NR of receiving antennas, NR > 1.

[0079] In the embodiment described here, the communication system 1 is designed to implement a cooperative communication technique based on relay nodes configured to retransmit messages from source devices that could not be correctly decoded by the destination device DEST. These relay nodes can be selected from among the source devices si, S2,..., SM (which are then said to be cooperative) and / or, depending on the configuration of the communication system 1, from among third-party devices of the communication system 1 (distinct from the source devices si, S2,..., SM), called relays.

[0080] In a configuration using relays, these relays are denoted RM+I, ..., RM+L, where L represents an integer greater than or equal to 1. These relays RM+I, -, RM+L are equipped with a single transmission antenna and are dedicated solely to relaying messages from the source devices si, S2, SM to the destination device DEST. Thus, unlike the source devices, they do not have their own informational messages to transmit to the destination device DEST. However, it should be noted that the communication system 1 may not include any RM+I, -, RM+L relays solely dedicated to relaying, in which case the relay nodes are selected only from among the source devices si, S2, ..., SM.

[0081] In the following, for the sake of simplification, we index by j, je X={1,..., M+L} the devices of communication system 1 (then generally designated by "device ]", for je X ) with the exception of the destination device DEST: the indices j=l,..., M designate respectively the source devices si, S2,..., SM of communication system 1 and the indices j=M+l,..., M+L designate respectively the relays RM+I, -, RM+L when communication system 1 relies on such relays (i.e. for L>1).

[0082] No limitations are attached to the nature of the various devices belonging to the communication system 1. For example, the source devices si, S2,..., SM can be user equipment or UEs (for "User Equipment") such as terminals, IoT (Internet of Things) devices, etc., the receiving device DEST a base station of a 5G or 6G network, and the relays of other user equipment connected to the network. This example is, however, given only for illustrative purposes and is not limiting in itself; the invention can be applied in other contexts, such as within a proprietary network, an ad-hoc network, etc.

[0083] It should be noted that the term "antenna" here encompasses a physical antenna, a radiating element of such a physical antenna, or a logical antenna. It generally refers to an antenna port (or AP for "Antenna Port") as defined by the 3GPP standard specifications, specifically section 4.4.1 of 3GPP document TS 38.211 entitled "Technical Specification Group Radio Access Network; NR; Physical channels and modulation (Release 17)" V17.7.0 (2024-03). Thus, an antenna port is a physical or logical entity such that the channel on which a symbol from the antenna port is transmitted can be deduced from the channel on which another symbol from the same antenna port is transmitted.

[0084] In the embodiment described here, the communication system 1 uses an orthogonal multiple access scheme of the OMAMRC type, based on an incremental redundancy retransmission protocol of the "IR-type HARQ" type, as previously mentioned. The (re-)transmissions between the devices of the communication system 1 are organized according to three phases: A cpO initialization phase, during which the receiving device DEST determines the modulation and coding scheme to be used by each source device si, S2,..., SM, based on the quality of the transmission channel separating it from that source device. The receiving device DEST informs the source devices si, S2,..., SM of the chosen modulation and coding schemes via a return link; a cpl transmission phase, during which the source devices si, S2,..., SM successively transmit their respective information messages, coded and modulated using the modulation and coding schemes determined by the receiving device DEST during the cpO initialization phase. During this transmission phase, the source devices si, S2,...SM successively uses an integer number NI greater than 1 of resource elements, NI being fixed and identical for all source devices (in other words, during the power line communication (PLC) transmission phase, each source device uses the transmission channel NI times to transmit its information message to the destination device DEST). In the example of the incremental redundancy retransmission protocol of the "IR-type HARQ" type considered here, during the PLC transmission phase, each source device si, S2,..., SM transmits more specifically a first redundancy version resulting from the encoding of its information message; and. a retransmission phase cp2, which lasts at most Tmax time intervals referred to herein as retransmission intervals, and which relies at each retransmission interval, according to the invention, on a plurality of relay nodes that simultaneously retransmit one or more information messages associated respectively with one or more source devices not decoded by the receiving device DEST, in the form of other redundancy versions resulting from the encoding of the information messages in question. It should be noted that an "other" redundancy version can be, for an information message associated with a given source device, a second redundancy version with respect to the first redundancy version transmitted during the cpl transmission phase or a subsequent redundancy version (third, fourth, etc.).) depending on the retransmission interval considered and the number of previous retransmissions of the information message in question during the retransmission phase cp2. For each information message considered, the same redundancy version is retransmitted simultaneously by the plurality of relay nodes using the same resource elements (N2 resource elements where N2 is an integer greater than 1). The relay nodes also collectively apply a precoding matrix to the redundancy versions of the retransmitted information messages (this matrix can be limited to a precoding vector if only one information message is retransmitted during a given retransmission interval).

[0085] In the embodiment described here, the devices (source devices, receiver and where applicable, relay) of the communication system 1 have the hardware architecture of a computer 2 as shown in Figure 3. This computer 2 includes in particular a processor PROC, a random access memory MEM, a read-only memory ROM, a non-volatile memory NVM, and COM means of communication.

[0086] The non-volatile NVM memory of the DEST receiving device constitutes a recording medium according to the invention, readable by the PROC processor and on which a PROG-DEST program according to the invention is recorded.

[0087] This PROG-DEST program includes instructions defining the main steps of an activation method according to the invention, and more specifically defines the functional modules of the DEST receiving device that rely on and / or control all or part of the PROC, MEM, ROM, NVM, and COM elements of computer 2 mentioned previously. These functional modules include, in particular, in the embodiment described here, as illustrated in Figure 4A, modules activated following the inability of the DEST receiving device to (correctly) decode at least one message received from at least one source device, referred to as undecoded.Note that for the sake of simplification in the following, we refer to information messages and undecoded source devices in the plural, even though it is possible that a single (or more) information message associated with a single (or more) source device may not have been correctly decoded by the recipient device D at the end of the cpl transmission phase.

[0088] In the embodiment described here, the functional modules of the DEST receiving device, which are activated for at least one retransmission interval t of the retransmission phase cp2, include: - a selection module 3, configured to select a first and second set satisfying a given performance criterion (for example, maximizing mutual information, or having mutual information greater than a certain threshold). The first set is denoted here as S opt, includes, for the considered retransmission interval t, v > 1 source devices not decoded by the receiving device DEST. The second set, denoted here B opt , comprises a plurality of relaying nodes (i.e., an integer N > 1) knowing each information message from each undecoded source device in the first set S opt For the sake of simplicity, the index t is omitted to denote the sets S opt and B opt as well as their respective cardinalities, v and N. The performance criterion is evaluated by the receiving device DEST, for the retransmission interval t, taking into account the variations on the N2 resource elements used during the retransmission interval of the precoding matrix applied collectively by the relay nodes of set B opt on the (same) redundancy versions of the v information messages from the v undecoded source devices of set S optThe method by which the DEST receiving device estimates variations in the precoding matrix to evaluate the performance criterion is detailed later; - an activation module 4, configured to activate the relay nodes of assembly B opt selected so that they retransmit, during the retransmission interval t, the same redundancy version of each information message from each source device in set S opt selected by selection module 3; and - a decoding module 5, configured to decode the information messages associated with the source devices si, S2,..., SM of the communication system 1 from the redundancy versions it receives or has received, either directly from the source devices during the cpl transmission phase, or from the relay nodes selected by the selection module 3 during the cp2 retransmission phase, and which in this case correspond to the information messages associated with the source device(s) of the set S opt .

[0089] The functions of modules 3 to 5 are described further later with reference to the steps of the activation process according to the invention illustrated in Figure 6.

[0090] In the embodiment described here, the source devices si, S2,..., SM and, where applicable, the relays RM+I, RM+2,..., RM+L of the communication system 1 are all capable of being selected as relaying nodes by the receiving device DEST during the retransmission phase cp2, and thus, of being activated by the latter to retransmit during a retransmission interval of the retransmission phase cp2. Each of these devices therefore incorporates the means of a relaying node according to the invention, and more specifically includes, in its non-volatile memory NVM, a PROG-R program according to the invention.

[0091] This PROG-R program includes instructions defining the main steps of a retransmission process according to the invention, and more specifically the functional modules of a relay node according to the invention that rely on and / or control all or part of the PROC, MEM, ROM, NVM, and COM elements of computer 2 mentioned previously. In the embodiment described here, these functional modules include, in particular, as illustrated in Figure 4B: - an information module 6, configured to inform the receiving device DEST that the relay node is aware, if applicable, of all or part of the information messages from the source devices (S2, SM). As mentioned previously, a relay node's awareness at a given time of an information message from a source device may arise, for example, because the relay node is itself the source device, or because it correctly decoded the message received from the source device during the cpl transmission phase or at the end of a previous retransmission interval of the cp2 retransmission phase. The relay node may also acquire knowledge of the information message in question by other means, notably following an exchange with other devices in the communication system 1.No other limitations are attached to the way in which the information module 6 informs the receiving device DEST of this knowledge: it can send to the receiving device DEST the set of source devices that it has managed to decode at the end of the cpl transmission phase or a retransmission interval of the cp2 retransmission phase, or it can, in response to a sending from the receiving device DEST of the set of source devices that the latter has not managed to decode, indicate the source devices of this set that it has managed to decode, or send a bit equal to 0 or to 1 to indicate whether it has managed to decode all the source devices not decoded by the receiving device DEST, etc.; - a receiving module 7, configured to receive from the destination device DEST (and more particularly from its activation module 4) a retransmission instruction intended to activate the retransmission by the relaying nodes of the set B. opt selected by the receiving device DEST for a given retransmission interval t of the retransmission phase cp2, using the same N2 resource elements, other redundancy versions of the information messages from the source devices of S opt The same redundancy versions are retransmitted by the relay nodes of set B. opt to which the relay nodes collectively apply the precoding matrix varying on the N2 resource elements as taken into account by the receiving device DEST to select the S sets opt and B opt ; And - a retransmission module 8, configured to execute the retransmission instruction of the receiving device DEST when the relay node is identified in set B opt for the considered retransmission interval t, and thus retransmit jointly during this retransmission interval t, with the other relay nodes identified in set B opt , the other relevant redundancy version(s) for the set S opt duly pre-coded. More specifically, the retransmission module 8 applies redundant versions of the information messages from the source devices of the S set opt a particular line of the precoding matrix varying according to the N2 resource elements used, designated in the retransmission instruction received from the receiving device DEST.

[0092] The functions of modules 6 to 8 are described further later with reference to the steps of the retransmission process according to the invention illustrated in Figure 6.

[0093] According to the invention, the mechanism implemented by the communication system 1, based on the activation and retransmission methods according to the invention, operates in an open loop. This means that there is no transmission by the receiving device DEST, via a return channel, to the other devices of the communication system 1, and in particular to the relay nodes of each set B opt selected for a retransmission interval t during the retransmission phase cp2, of an estimation of the radio propagation channel seen by the receiving device DEST, in order to allow the relay nodes of each set B optto determine the precoding matrix to be used during the retransmission interval for which they were selected, or equivalently, information defining the precoding matrix to be used by the relay nodes of set B opt during this retransmission interval.

[0094] To overcome this deficiency, the invention relies on the shared knowledge by each device of the communication system 1, and more specifically by the receiving device DEST and the relay nodes involved during the retransmission phase cp2, of a dictionary (or "codebook") CODEB comprising a plurality of predefined and ordered precoders, each precoder being defined by a precoding matrix. This CODEB codebook is, for example, stored in the non-volatile memory of each device of the communication system 1. This makes it possible to limit the amount of information transmitted to the relay nodes of the set B opt in the DEST recipient device retransmission instructions.

[0095] In the embodiment described here, the CODEB codebook is composed more specifically of a plurality of CODEB(pl,p2) codebooks, where pl denotes an integer greater than or equal to 2 and p2 denotes an integer greater than or equal to 1, with p2 ≤ min(p1, NR). In this case, the devices of communication system 1 (the receiving device DEST and the relay nodes selected for the retransmission phase cp2) are configured to refer to the appropriate CODEB(pl,p2) codebook during a retransmission interval t of the retransmission phase cp2: more precisely, for such a retransmission interval t, the codebook considered by the devices of communication system 1 corresponds to the pair (pl,p2) = (N,v) where v denotes, as introduced previously, the number of undecoded source devices in the set S optselected by the receiving device DEST for the considered retransmission interval t and N denotes the number of relay nodes in set B opt selected for this set S opt , with v < min(N, NR).

[0096] Furthermore, to enable open-loop operation of the invention, the way in which the precoding matrices applied by the relay nodes of assembly B opt During a given retransmission interval t, values ​​are obtained from the CODEB codebook ( / V,v), and therefore evolve according to the N2 resource elements used by the relay nodes during the retransmission interval (in other words, the variation pattern of the precoding matrix on the N2 resource elements) is predefined and also known to the relay nodes. According to the invention, this variation pattern is known not only to the relay nodes but also to the receiving device DEST.

[0097] For example, such a variation pattern involves using a precoding matrix distinct from the CODEB codebook (N, v) every x resource elements (or equivalently, every x channel uses), where x is an integer such that l <x< N2, la matrice de précodage appliquée tous les x éléments de ressource étant obtenue en parcourant de façon séquentielle (i.e. dans l'ordre des index) le codebook CODEB( / V,v). A titre illustratif, on considère par exemple ici que x = 1, autrement dit la matrice de précodage appliquée par les nœuds de relayage de l'ensemble B opt selected for a retransmission interval r changes at each time / frequency resource element i (or RE for "Resource Element" in English) or each use of channel i, with i=0 N2-1.

[0098] The way in which CODEB( / V,v) codebooks are constructed for different pairs of values ​​(N,v) is now described in detail. In the embodiment described here, different possible constructions are considered depending on the values ​​of v under consideration.

[0099] Each CODEB( / V,v) codebook is by construction of dimension Kv, that is, it contains K v precoding matrices, where K denotes an integer such that: | / V| K = I— I + N mod v Therefore, if N2 > Kvx (in other words, Kv in the illustrative example considered here where x = 1), the precoding matrix to be applied to all x resource elements (that is, all uses of the channel in the illustrative example considered here where x = 1) can be obtained by sequentially and cyclically traversing the CODEB(N, v) codebook. This cyclic traversal of the CODEB(V, v) codebook is part of the known variation pattern of the relay nodes and the DEST receiving device of system 1.

[0100] For a number N of relay nodes in set B opt and a number v of undecoded devices in the set S opt , with N > 2 and v > 1, each precoding matrix applied collectively by the relay nodes of set B optEach resource element, or each use of channel i (for x = 1), is denoted P(i) for i = 0, ..., N2-1, and is extracted from the codebook CODEB( / V,v), which contains K.v precoding matrices. During a retransmission interval t, each relay node of set B opt selected for this retransmission interval t applies a separate row of the precoding matrix to the redundancy versions associated with the information messages of the v undecoded source devices in set S opt sent jointly by the relay nodes of set B opt using the same N2 resource elements. This retransmission is thus similar to spatial multiplexing on v spatial layers by the relay nodes for v > 1.

[0101] For v > 1: each precoding matrix P(i) is, in the embodiment described here, constructed from a combination, and more particularly from the product, of a complex matrix I47(j) of dimensions N xv, a complex diagonal matrix £>(i) of dimensions vxv introducing a diversity of cyclic delays (called "large delay CDD"), and a complex rotation matrix U of dimensions vxv. The "global" complex precoding matrix V(i) of dimensions N xv is thus defined, for the i-th use of the channel, by: V̄(i) = W(i)D(i)U for i=0,..., N2-1 (Eq. 1)

[0102] The diagonal matrix D(i), which is the origin of the diversity of cyclic delays, introduces a phase shift between the undecoded source devices retransmitted from S opt for v>l. It is defined by: ( ~j2m -j2Tr(y-l')i \ 1, evvj with j 2 =-1. In the illustrative example considered here where x = 1, this is equivalent to: £)(i) = D k , k = i mod v with: ( -j2nk -J2nk(yi) \ l,evvj, k E {0,..., v — 1}

[0103] The diagonal matrix D(i) therefore varies periodically with a period v.

[0104] The rotation matrix U is a fixed DFT (Discrete Fourier Transform) matrix (i.e., one that does not depend on the use of channel i), which is written as follows: U = [matrix with DFT entries as shown in image] L1 e v... Q vev J

[0105] The diagonal matrix of "large delay CDD" D(i) and the rotation matrix U are thus, in the embodiment described here, predefined as a function of the number v of source devices not decoded by the destination device DEST retransmitted during a retransmission interval t.

[0106] For N > min(1,v) and v > 1, the matrix W(i) is obtained from K matrices C₀, C₁,..., Cₖ₋₁ by applying the following relation: W(i) = C k , k = ⌊i / v⌋ mod K where Co, Ci,..., CK-I denote matrices of dimensions N x v. The matrix W(i) is therefore constant over v resource elements or, equivalently, v uses of the channel (during which the matrix D(i) varies). Thus, for example, for v = 3, N2=16 and K = 4, this amounts to selecting the following matrices W(i) for i=0,..., N2-1: Co, Co, Co, Ci, Ci, Ci, C2, C2, C2, C3, C3, C3, Co, Co, Co, Ci, for the resource / channel usage elements indexed by i=0, 1, 2, 3,..., 15 respectively.

[0107] Each matrix Ck, k=0,..., Kl has the following structure: Cₖ = [v₀⁽ᵏ⁾ v₁⁽ᵏ⁾ ... vᵥ₋₁⁽ᵏ⁾] HEIGHT="27" WIDTH="135" SRC="imgf000018_0002.tif" / > [ v 0 V 1 ■■■ v v-lJ where column v l (k) is the precoding vector to be applied for the undecoded source device l = 0, of the set S opt Each matrix Co, Ci,..., CK-I (and therefore, incidentally, each matrix W(i) obtained from these matrices) is remarkable in that its v columns are linear combinations of distinct vectors from a basis of N orthogonal vectors w n , n=0,.., N -1. By "linear combination", we mean here any weighted sum of basis vectors as long as at least one weighting coefficient considered in the sum and applied to a basis vector is non-zero.

[0108] In the embodiment described here, we consider more specifically the basis formed by the N orthogonal vectors w n , n=0,.., Nl, defined by: r w n = 1 / √N [1 e^(j2πn / N) e^(j2·2πn / N) ... e^(j2π(N-1)n / N)]ᵀ, n = 0,..., N - 1 (Eq. 2)

[0109] These vectors w n , n=0,.., / V - 1 each represent a so-called "IDFT" beam. They are also referred to hereafter as "IDFT vectors".

[0110] More specifically, in the embodiment described here, v - 1 columns (typically the first v - 1) of each matrix Ck, k=0,..., K - 1 result from a linear combination of distinct adjacent (consecutive) vectors from the orthogonal basis of the IDFT vectors, and one column of the matrix Ck, k=0,..., K - 1 (typically the last) results from a linear combination of + N mod v distinct adjacent (consecutive) basis vectors. Thus, the v vectors corresponding respectively to the v columns of each matrix Ck are orthogonal. This last column allows us to handle the case where N is not a multiple of v, if necessary. Of course, another position can be assigned to this column in the matrix C kAlternatively, other basis vectors besides adjacent vectors can be combined together. It should be noted that if N is a multiple of v, the v columns of the matrix C k are then linear combinations of distinct adjacent vectors of the IDFT vector basis).

[0111] Furthermore, in the embodiment described here, the K matrices C k distinct values ​​are obtained relative to each other by shifting the 1 er IDFT vector considered for the first column of each matrix C k , k=0,..., K - 1.

[0112] Thus, the matrix C0 has the following structure: v₀⁽⁰⁾ = (1 / β₀⁽⁰⁾) Σ wₙ [n=0 to ⌊N / v⌋-1]; v₁⁽⁰⁾ = (1 / β₁⁽⁰⁾) Σ wₙ [n=⌊N / v⌋ to 2⌊N / v⌋-1]; ...; vₗ⁽⁰⁾ = (1 / βₗ⁽⁰⁾) Σ wₙ [n=l⌊N / v⌋ to (l+1)⌊N / v⌋-1]; ...; vᵥ₋₁⁽⁰⁾ = (1 / βᵥ₋₁⁽⁰⁾) Σ wₙ [n=(v-1)⌊N / v⌋ to N-1] n where β l (0), l ∈ {0,...,v - 1} represent real weighting coefficients, at least one of these coefficients being non-zero. These weighting coefficients β l (0) , l ∈ {0,...,v - 1} are chosen for example in such a way as to normalize the precoding vectors v₀⁽⁰⁾, v₁⁽⁰⁾,...,vᵥ₋₁⁽⁰⁾ In the embodiment described here, all the weighting coefficients p ; (0) , the {0,...,v - 1} are non-zero.

[0113] To construct the C1 matrix = [v^ 1} ... v^, as mentioned previously, the IDFT vectors w n considered for each precoding vector Vj 7 are shifted cyclically. Thus, w0 is no longer considered for constructing the first column but to construct the last column v^. That is: l = 0,...,v - 2 (1) V, N 1 mod N ' ZV 1 l+i where {0,...,v - 1} represent real weighting coefficients. These weighting coefficients The {0,...,v - 1} are chosen, for example, in such a way as to normalize the precoding vectors In the embodiment described here, all the weighting coefficients the {0,.... v - 1} are non-zero.

[0114] The precoding vectors v l (k) of matrix C k are defined similarly, as follows: (i+i)[J]+ki v — 2 I = v — 1 where β l (k) , l ∈ {0,...,v - 1} represent real weighting coefficients. These weighting coefficients β l (k) , l ∈ {0,...,v - 1} are chosen, for example, in such a way as to normalize the precoding vectors v®,v®, In the embodiment described here, all the weighting coefficients p ; (k) , the {0,...,v - 1} are non-zero.

[0115] The construction just described leads to obtaining the K matrices C k , k=0,..., K -

[0116] In view of the above and the variation patterns of the matrices W(i) (which can take K distinct values ​​given by the K matrices C k , k=0,..., K - 1) and D(i) (which can take v distinct values ​​given by the v matrices D m (m=0,..., v-1), with the rotation matrix U being fixed, it is necessary to store in the codebook CODEB( / V, v ), in an ordered manner, the Kv possible combinations of the matrices C k , k=0,..., K - 1, D m , m=0,..., v-1 and U (fixed for Kv combinations). In the embodiment described here, the order is as follows: C₀D₀U, C₀D₁U,...,C₀Dᵥ₋₁U, C₁D₀U,C₁D₁U,...,C₁Dᵥ₋₁U,..., Cₖ₋₁D₀U,Cₖ₋₁D₁U,...,Cₖ₋₁Dᵥ₋₁U

[0117] Figure 5 illustrates the construction of the codebook CODEB( / V,v) for v = 3, N = 10 and K = 4.

[0118] By definition, the N vectors w n , n=0,.., Nl, and incidentally by construction, the vectors v l (k) , l ∈ {0,...,v - 1} of C matrix precoding k k=0,..., K - 1, allow for uniform scanning of the entire space covered by the antenna network formed by the N relay nodes of the set B opt selected.

[0119] Note that other constructions of precoding vectors v l (k)l ∈ {0,..., v - 1} can be considered as an alternative, provided that the precoding vectors are obtained from linear combinations of distinct IDFT vectors to ensure the orthogonality of the precoding vectors with each other. For example, one can combine vectors that are not adjacent (which amounts to applying a permutation to the previous formulas), or combine a different number of vectors for the different precoding vectors, or even consider only a reduced number of non-zero weighting coefficients per group of IDFT vectors (for example, a single non-zero weighting coefficient, applied to the IDFT vector in the middle of the group considered, or two non-zero coefficients applied to vectors duly distributed within the group).

[0120] The construction just described applies to a number N of relay nodes strictly greater than the number v of undecoded source devices retransmitted by these relay nodes. When N = v, in the embodiment described here, a fixed matrix W(i) is used for the N2 resource elements / for the N2 channel uses, and is taken to be equal to the normalized identity matrix of dimensions vxv, that is: 1 W(i) = — I v i.e. {0,..., N2 - 1} Vv

[0121] The overall precoding matrix V(i) is then given by: V(i) = 1 / √v D(i)U ie {0,..., N2 - 1} Vv

[0122] The v possible values ​​of V(i) (resulting from the v possible values ​​of £>(i)) are stored in codebooks CODEB(v,v) for the different values ​​of v that can be considered by the communication system 1.

[0123] For v = 1: In the case where only one undecoded source device is retransmitted during a retransmission interval t, the relay nodes of set B opt collectively applies to the redundancy version associated with the information message of this undecoded source device a matrix W(i) which reduces to a single precoding vector of dimension N. This precoding vector is selected from a CODEB( / V, 1) codebook of dimension N.

[0124] In the embodiment described here, the codebook CODEB( / V, 1) consists of the N IDFT vectors w n, n=0,..., N - 1 introduced previously and given by relation (Eq. 2) (for v = 1, the matrices £>(i) and U of equation (Eq. 1) reduce in fact to £)(i) = 1 and U = 1). When N2> N, the precoding vector applied by the N relay nodes is cyclically selected from the CODEE dictionary / V, 1), which amounts to scanning the different IDFT directions, as illustrated in figure 5.

[0125] In another embodiment, a different diversity technique called "small delay CDD" is considered, according to which the precoding vector applied by the N relay nodes is defined by: V(i) = D'(i)U' with: 1 U' = [1, 1, ..., 1]^T .1. of dimension N and: diag(1, e^(-j2πδ_cyc,2i / N2), ..., e^(-j2πδ_cyc,ni / N2), ..., e^(-j2πδ_cyc,Ni / N2)) where δ cyc,n is a positive real cyclic delay. For example, we can take δ cyc,n= 4n, leading to a transmission through the corresponding relay node delayed by 4nT e where T e denotes the sampling time. Of course, this is only an illustrative example and is not exhaustive; other values ​​can be considered as alternatives.

[0126] The plurality of CODEB( / V,u) codebooks thus constructed for different values ​​of N and v, make up the CODEB codebook which is, as mentioned previously, stored by each relay node and by the receiving device DEST of the communication system 1 in their respective non-volatile NVM memories, with the variation pattern of the precoding matrices P(i) as a function of the resource elements used. In the example considered here, according to this variation pattern, whatever the values ​​of N and v used for a retransmission between the relay nodes and the receiving device DEST, the "global" precoding matrix V(i) for the resource element / channel use i, i=0,..., N2-1, is obtained by sequentially traversing the CODEB( / V,u) codebook, and cyclically when N2> K. vx (in other words Kv in the example considered here where x = 1).

[0127] The precoding matrix applied is also assumed here to be normalized and such that: - n u- o H no =- 1i v where P(i) H is the transposed conjugate matrix of P(i).

[0128] Figure 6 represents the main steps of the activation and retransmission processes as implemented, in a particular embodiment, by the devices of the communication system 1, and more particularly by the receiving device DEST and by the relaying nodes selected to participate in the retransmission phase cp2 during at least one retransmission interval t.

[0129] The different devices (source devices, receiving device and, where applicable, relay) of communication system 1 are assumed to be synchronized, and the source devices si, S2,..., SM are assumed to be statistically independent (no correlation between them).

[0130] As mentioned previously, during the cpO initialization phase that precedes the transmission of data frames by the source devices si, S2,..., SM, the receiving device DEST determines, for each source device Sm, m=l,...M, the modulation and coding scheme (MCS) that it must use to transmit data (step E10). The way in which the receiving device DEST does this involves link adaptation techniques, known to those skilled in the art and not described here. The receiving device DEST sends to each source device Sm, m=l,..., M, via a limited-rate control channel, information that unambiguously identifies the MCSm scheme assigned to it (step E20). This information is, for example, in the context of a 5G network, an MCS scheme index pointing in a specific table to a MOD modulation, a coding efficiency R mto obtain at the output of a given COD coding scheme (e.g., an LDPC code), and a spectral efficiency value to use for transmitting its useful data.

[0131] No limitations are attached to the modulations or coding types that can be envisaged in the context of the invention. For example, one can consider using phase shift keying (PSK) or quadrature amplitude modulation (QAM) of various orders, LDPC (Low Parity Check Code) coding schemes, turbo codes or convolutional codes, systematic or non-systematic, or any other modulation and / or coding scheme.

[0132] Following this cpO initialization phase, the cpl transmission phase begins.

[0133] Each source device Sm, m=l,..., M, attaches to the useful data (information message in the sense of the invention) that it wishes to transmit to the receiving device DEST a CRC (for "Cyclic Redundancy Check" in English) for error detection, then codes the useful data completed by the CRC using the COD coding scheme.

[0134] When using an IR-HARQ type protocol based on incremental coding, the information message including the CRC is encoded by the source device. m with a low-efficiency master code (e.g., 1 / 3 or 1 / 5), typically lower than the efficiency of the MCSm scheme indicated during the initialization phase. The number of bits K m of the information message from the source device s m depends on the coding efficiency and MOD modulation of the MCSm scheme indicated during the cpO initialization phase. More specifically, K m = R m . NI. q m where qm denotes the number of bits per symbol carried by the modulation. The coded information message resulting from this encoding (or codeword) is then stored by the source device. m For example, in its non-volatile NVM memory, in a circular buffer as illustrated in Figure 7. In the example shown in Figure 7, a systematic master code with a 1 / 3 efficiency is considered. As is well known, the coded bits obtained as output from a systematic code consist of the information bits (called "systematic" bits) supplied as input to the code (possibly punched) and redundancy bits.

[0135] The coded bits stored in the circular buffer are organized into a number of so-called redundancy versions, denoted RVOm, RVlm, etc., starting at specific positions POSOm, POSlm,..., on the buffer, and whose dimensions are fixed according to the number of resource elements available for each (re)transmission. Thus, the first redundancy version, RVOm, has dimensions Nl.q m , then the subsequent redundancy versions are of dimension N2.q m The first version of the RVOm redundancy is also chosen to be self-decodable. For example, it includes systematic bits as illustrated in Figure 7. Note the configuration of the POSOm, POSlm, and P0S2 positions. m The POS3m configuration illustrated in Figure 7 is that proposed by the 3GPP standard and is only an illustrative example, not a limiting one. Other configurations can be considered.

[0136] During the cpl transmission phase, the first redundancy version RVOm is transmitted by the source device Sm, after modulating the coded bits contained in this redundancy version RVOm with the MOD modulation corresponding to the MCSm scheme determined by the receiving device DEST (step E30). The other redundancy versions are transmitted subsequently if necessary during the cp2 retransmission phase, as described later.

[0137] The source devices S1, S2, ..., SM transmit their respective first redundancy versions RVO1, RVO2, ..., RVOM to the destination device DEST, in turn during M time intervals of the cpl transmission phase. Note that when a source device Sm transmits its first redundancy version RVOm during a time interval, the other devices of communication system 1 (i.e. the destination device DEST, the other source devices S1, j=l, ..., M and j*m, and where applicable the relays RM+1, RM+2, ..., RM+L) listen to the transmission channel (without transmitting).

[0138] At the end of the PLC transmission phase (or each time interval of the PLC transmission phase), the receiving device DEST attempts to decode the messages received from the source devices si, S2,..., SM (step E40). For simplicity, "decoding (correctly) a message received from a device" is sometimes referred to as "decoding the device" in question.

[0139] The receiving device DEST identifies the source devices whose messages it has correctly decoded (referred to here as "decoded devices") using the CRCs attached to the information messages of the source devices, in a way that is known per se. We denote S DEST0 the so-called decoding set of the receiving device DEST comprising the source devices correctly decoded by the receiving device DEST at the end of the powerline transmission phase, and by S DEST 0 the complementary set of the S decoding set DESTfiwhich includes so-called "undecoded" source devices whose destination device DEST was unable to correctly decode the messages.

[0140] The other devices j, je X of communication system 1 proceed identically from the messages they received during the cpl transmission phase (step E50). We denote by 5)0, je X the decoding set of a device ] of communication system 1, and by S jfi , i X, the set of source devices that were not correctly decoded by device ]. Note that if device j is a source device of communication system 1, the decoding set S j,0 includes at least device j, since it knows its own information message.

[0141] If S̄ DEST,0If the cpl transmission phase is not empty (i.e., there is at least one source device not decoded by the receiving device DEST at the end of the cpl transmission phase), the receiving device DEST informs the other devices j, j ∈ Σ, of the communication system 1, for example by sending a non-acknowledgment message NACK in a control or feedback channel (step E60). Otherwise, the receiving device DEST sends an acknowledgment message ACK in the control or feedback channel, and a new cpl transmission phase (possibly preceded by a new cpO initialization phase if propagation conditions require it) can be implemented by the source devices if, S2,..., SM.

[0142] Note that the acknowledgment (ACK) and non-acknowledgment (NACK) messages are not necessarily explicit. For example, the receiving device DEST can send the set of source devices S̄ to the devices of communication system 1. DEST,0 that it was unable to decode during the transmission phase, and if this set is not empty, this is interpreted by the devices of communication system 1 as a non-acknowledgment message. Conversely, if the set S̄ DEST,0 The transmitted message is empty; the devices of communication system 1 interpret it as an acknowledgment message.

[0143] The sending of the non-acknowledgment message by the receiving device DEST triggers the cp2 retransmission phase. The cp2 retransmission phase is designed to allow the receiving device DEST to correctly decode the information messages from the source devices in its set S̄ DEST,0and lasts for a maximum number T max of time intervals (in other words, a maximum number Tmax of retransmissions is allowed).

[0144] The steps described below are repeated for at least T used ≥1 time interval(s) t (called retransmission interval(s) t) of the retransmission phase cp2, until a predetermined stopping criterion is met (test step E70). In the embodiment described here, this stopping criterion is reaching the maximum number Tmax of retransmissions (i.e., t = Tmax) or obtaining, at the end of the retransmission interval t of the retransmission phase cp2, a set S DEST t empty for the DEST recipient device. Of course, other stopping criteria can be considered.

[0145] In the following, we will denote by S DEST t and S j tthe respective decoding sets of the receiving device DEST and each device j of the communication system 1, with je X, at the end of the retransmission interval t of the retransmission phase cp2, and by S̄ DEST,t and S̄ j,t their complementary sets (i.e. including the source devices not correctly decoded respectively by the destination device DEST and by the other devices indexed by j of the communication system 1).

[0146] During the retransmission interval t of the retransmission phase cp2, 1≤t≤ T used A device j of the communication system 1 can help the receiving device DEST to decode a source device s m of set S DEST and therefore serve as a relay node for this source device if it knows the information message associated with this source device Sm, either because this device is itself the source device s meither because it correctly decoded its information message during a previous retransmission interval of the cp2 retransmission phase or at the end of the cpl transmission phase, or because it is aware of it by some other means. As mentioned previously, if device j is a source device of communication system 1, the decoding set S j t- includes at a minimum device j.

[0147] Following the sending of the non-acknowledgment message from the receiving device DEST, the receiving device DEST is informed of the devices in communication system 1 that are aware of the information messages associated with the source devices of set S̄ DEST,t-1 (step E80).

[0148] To this end, different approaches can be considered.

[0149] For example, upon receiving the non-acknowledgment message NACK from the receiving device DEST, the devices j, iX, of the communication system 1 inform the receiving device DEST, via their respective information modules 6 (for example, by means of a message sent to the receiving device DEST in the control or feedback channel), of their respective decoding sets S j,t-1 , j ∈ Σ.

[0150] In another example, the receiving device DEST informs the devices j, je X of the communication system 1 of its decoding set S DEST,t-1 or its set of undecoded source devices S̄ DEST,t-1and the devices j, i X, of the communication system 1 inform the receiving device DEST, via a message sent back by their respective information modules 6 (for example in the control or feedback channel), when they know one or more information messages not decoded by the receiving device DEST by identifying the known message(s) (for example by sending the intersection of their set of decoded source devices with the set S̄ DEST,t-1 of undecoded devices of the recipient device DEST).

[0151] The receiving device DEST then selects, for the considered retransmission interval t, a set B opt comprising several relay nodes among the devices j, i X, of the communication system 1 to retransmit, using the same N2 resource elements, the redundancy versions of the information messages associated with v source devices of the same set Sopt including at least one source device of the set (i.e., v > l). It should be noted that set B opt groups N devices of communication system 1 that know all the information messages associated with each of the v source devices of set S opt , with N > min(2,v).

[0152] The S sets opt and B opt are selected for the retransmission interval t by the receiving device DEST via its selection module 3, for example at the end of the retransmission interval t-1 (retransmission interval 0 corresponding to the end of the power line communication (PLC) transmission phase), in order to optimize a given performance criterion (step E90). For example, the receiving device DEST selects sets S opt and B optoptimizing (i.e., maximizing) the mutual information of the equivalent propagation channel between the relay nodes of set B opt and the NR receiving antennas of the DEST recipient device.

[0153] More specifically, during step E90, knowing the decoding sets (or their complement) of all devices in communication system 1 and having the link quality between each of these devices and itself known in itself, the receiving device DEST calculates the mutual information of the equivalent channels for all eligible undecoded source device subsets S and for all possible relay node subsets B in communication system 1 for these subsets S, with: S ⊆ Part NR (S̄ DEST,t-1 ) and |S| < |B| where |B| and |S| denote respectively the cardinalities of sets B and S, and: Part NR(S̄ DEST,t-1 ) = ∪ NR Q=1 Part Q (S̄ DEST,t-1 ) the part y (A) denoting all possible subsets of cardinality y of elements of set A.

[0154] Alternatively, in order to limit the complexity of the evaluation of the performance criterion by the receiving device DEST, it can be envisaged that the receiving device DEST is limited to certain configurations of ( / V,v) when optimizing the performance criterion.

[0155] The Annex presents an example of pseudocode that can be used during step E90 by the selection module 3 of the DEST recipient device. According to this example, to select sets B opt and S opt used for the retransmission interval t, the receiving device DEST calculates the set of parts. Q (S̄ DEST,t-1 ) of the set S̄ DEST,t-1undecoded source devices of the destination device DEST at interval t-1, limiting ourselves to subsets with cardinality less than or equal to a given value Qmax (for example, Qmax is taken to be less than or equal to the number of NR receiving antennas of the destination device DEST) and which are different from the empty subset. By part of a set A, denoted Part(A), we mean the set of all possible non-empty subsets consisting of elements of A. The resulting set is denoted Part(S̄ DEST,t-1 ). The subsets of P Part Q (S̄ DEST,t-1 ) of cardinality Q ∈ {1,..., Qmax} are denoted Part Q (S̄ DEST,t-1 ) and are such that Part(S̄ DEST,t-1 ) = ∪min{NR,Qmax} Q=1 Part Q (S̄ DEST,t-1 ) -

[0156] Then, for each subset of source devices S e PartQ^S^sp ^), Q e {1,..., Qmax), the receiving device DEST determines the set B of devices i X, of the communication system 1 that have decoded at least the source devices of S with the constraint |S| < |B| for S to be eligible. If no subset is eligible for a given value Q o Therefore, the receiving device DEST does not test for values ​​of Q > Q o .

[0157] According to the invention, the receiving device DEST determines the mutual information of an equivalent channel between a possible set B for a set S of eligible undecoded source devices by taking into account the precoding matrix that would be collectively applied by the relay nodes of set B, and its variations on the N2 resource elements used by the relay nodes in question to retransmit the associated redundancy versions. As previously stated, this precoding matrix is ​​extracted from the CODEB dictionary (and varies within this dictionary), and more specifically from the CODEB(|B|,|S|) dictionary. The method by which the receiving device DEST calculates this mutual information for each pair of sets B and S is described in more detail later.

[0158] Following the selection made during step E90, the receiving device DEST activates, via its activation module 4, the relay nodes of assembly B opt selected so that they transmit, during the retransmission interval t, using the same N2 resource elements, the same redundancy version of each information message from each source device in set S opt selected (step E100).

[0159] In the embodiment described here, this activation includes the sending by the receiving device DEST, in the return channel, of retransmission instructions to each of the relay nodes of set B opt selected. The retransmission instructions can, for example, be broadcast to all devices in communication system 1. Here, they include the S sets opt and B opt selected, and designate for each relay node of set Bopt , the row of the precoding matrix P(i) (i.e., which denotes the precoding vector) that it must apply to the redundancy versions associated with the information messages of the source devices of set S opt .

[0160] To limit the amount of information sent in the return channel, the S set opt is sent in the form of a sequence of indices m0,... m v-1 , with m L e {1,... M},1 = 0,...,v - 1 ordered in ascending order and designating the undecoded source devices selected for the retransmission interval t by the receiving device DEST during step E90. Set B opt is provided in the form of a vector b = [b^b^...,b N-1 ] T ( T designating the transposition operator), or equivalently, an ordered sequence of indices b n ,n = 0,..., N - 1 with b n∈Σ, identifying the relay nodes of communication system 1 selected from set B opt The index of a relay node in the sequence designates the row of the precoding matrix P(i) (in other words, the precoding vector) to be applied by the relay node identified by that index. For example, the relay node with index b n must apply the n ème row of the precoding matrix P(i), and more specifically the coefficient v̄ n,l (i) to the message associated with the l-th source device indexed by m l designated in the set S opt It should be noted that similarly, the order of an index in the sequence designating the undecoded source devices of set S opt indicates the column of the precoding matrix (precoding vector) to be applied by the relay nodes of set B optto the redundancy version associated with the information message of the source device identified by this index.

[0161] Alternatively, the receiving device DEST can be considered to send individually to each selected relay node of set B opt the retransmission instruction concerning it. In this variant, the retransmission instruction includes the set S opt , a designation of the row of the precoding matrix to be applied by the relay node, as well as an indication of the redundancy versions to be transmitted for the source devices of the S set opt .

[0162] As previously mentioned, the invention operates in an open loop: therefore, it is not necessary for the receiving device DEST to transmit in the retransmission instructions the precoding matrix V(i), i=0,..., N2-1 to be applied by the nodes of set B optfor each resource element i=0,..., N2-1 used, during the retransmission interval t. This can be extracted, without additional information provided by the receiving device DEST, by each relay node from the CODEB codebook it has in memory for each resource element i=0,..., N2-1, knowing the variation pattern of the precoding matrix as a function of the resource elements used (i.e., the channel usages) and the values ​​of ( / V,v). Typically, in the example considered here of a precoding matrix varying with each resource element, a relay node of set B opt obtains each precoding matrix P(i), i=0,..., N2-1 to be applied to the redundancy versions of the source devices of the set S optby cyclically traversing the CODEB( / V,v) dictionary at each resource element and applying the resulting row of the precoding matrix V(i) for a resource element i designated for it in the retransmission instructions. Note that the values ​​of N and v can be obtained by each relay node from the sets B opt and S opt received in the retransmission instructions, N and v corresponding to the respective cardinals of these two sets.

[0163] Furthermore, in the embodiment described here, no information regarding the power to be applied needs to be transmitted by the receiving device DEST to each relay node: each relay node applies it to each redundancy version associated with a source device in the set S optthe total transmission power P available at its level, the precoding matrix P(i) being effectively here without amplitude modulation.

[0164] Upon receiving the retransmission instructions from the DEST receiving device, each relay node of set B opt retransmits to the destination device DEST, using the same N2 resource elements, during the retransmission interval t, redundant versions of the information messages associated with the source devices of set S opt to which are applied the coefficients of the row of the precoding matrix P(i) associated with it (step El 10). Thus, during step El 10, the relay nodes of set B opt execute, via their respective retransmission modules 8, the retransmission instructions that have been transmitted to them.

[0165] It should be noted that when retransmission instructions are broadcast by the receiving device DEST to all devices in communication system 1, each device is able to keep track of the redundancy version to be retransmitted for a source device at each retransmission interval of the retransmission phase cp2, and thus identify the redundancy version to be retransmitted for the retransmission interval t for which it has been selected, if applicable. The redundancy version transmitted during the retransmission interval t for the source device is denoted by s mi with, in the particular embodiment described here based on Figure 7, and 4 distinct redundancy versions, t mi = t m ' l mod 4 where t m ' l denotes the number of retransmissions of the information message associated with the source device s m;(including that of the retransmission interval t) in the form of redundancy versions.

[0166] Each relay node indexed by bn, n=0,..., Nl, thus transmits simultaneously, during the retransmission interval t, using the same resource elements i=0,..., N2-1, the linear combination of the v redundancy versions RVt mi RVt m2 ,..., RVt mv associated respectively with the v source devices s mi ,...,s mv of set S opt weighted by the coefficients of n ième row of the precoding matrix P(i), i.e.: x̃ n (i) = Σ v-1 l=0 v̄ n,l (i). RVt ml (i) where v̄ n,l (i), l = 0,..., v - 1, denote coefficients of the n ième line of the precoding matrix P(i), and designates the i ième (i.e., {0,..., N2 - 1}) coded modulated symbol of the t mi-th version of redundancy resulting from the encoding of the information message associated with the source device s mi The RVt redundancy version mi transmitted at time interval t is obtained by the relay node indexed by bn from a circular buffer such as the one shown in Figure 7, generated from its knowledge of the information message associated with the source device s mi The RVt redundancy version ml The transmitted data is obtained by reading the encoded bits from the POSt position into the buffer. ml Thus, for the retransmission interval indexed by t=l, the redundancy version RV1 ml is read from position POS1 ml in the circular buffers corresponding to each information message associated with the source device s ml , for l = 0,...,v - 1. Subsequently, the redundancy versions are transmitted, in the embodiment described here, in the following order: RV2 mi RV3mi RV0 mi RVl mi RV2 mi RV3 mi , etc.

[0167] During the retransmission interval t and step El 10, all relay nodes of set B opt therefore simultaneously transmit on the same N2 resource elements a linear combination of the same redundancy versions associated with the source devices s m of set S opt The linear combination differs from one relay node to another depending on the rows of the precoding matrix P(i) applied. During this retransmission interval t, all relay nodes therefore identically use the same number N2 of resource elements (i.e., the same number N2 of channel uses) and retransmit the same redundancy versions of the information messages associated with the undecoded source devices of set S opt .

[0168] Upon receipt by the receiving device DEST (via its decoding module 5) of the redundancy versions transmitted during the retransmission interval t by the relay nodes of set B opt The receiving device DEST attempts to decode the messages received during the cpl transmission phase from the source devices of the set S opt by using the new redundancy versions transmitted by the relay nodes (step E120). The method by which the receiving device DEST accomplishes this is known and relies on techniques commonly used within an IR-HARQ protocol, not described here. All other devices in communication system 1 can proceed similarly.

[0169] The receiving device DEST identifies new source devices whose messages it has successfully decoded using the CRCs attached to the information messages of these source devices, as described previously, and updates its decoding set S D t and the complementary set S DEST t at the end of the retransmission interval t.

[0170] If the decoding set S̄ DEST,t is empty (all source devices SI,..., SM are correctly decoded), the receiving device DEST sends an acknowledgment message ACK in the control or feedback channel, and the cp2 retransmission phase is completed (step E130). A new cpl transmission phase (possibly preceded by a new cpO initialization phase if propagation conditions require it) can be implemented for new information messages from the source devices SI,..., SM.

[0171] If S̄ DEST,tIf the channel is not empty (i.e., if at least one source device remains undecoded by the receiving device DEST after the retransmission interval t), the receiving device DEST informs the other devices j, iX, of communication system 1 by sending a non-acknowledgment (NACK) message in the feedback control channel, identical to what was done previously in step E60 (step E130). If the maximum number of retransmissions is not reached (test step E70), steps E80 to E130 are repeated for a new retransmission interval t+1.

[0172] For illustrative purposes, we consider for example an OMAMRC system comprising M=2 source devices si and S2 and a destination device DEST.

[0173] We assume that at the retransmission interval t-1: S̄ DEST,t-1 = ∅ S̄ 2,t-1 = {s1, s2}

[0174] During the E90 selection step, three possibilities exist for the DEST recipient device for sets S and B: S={s1} and B={s1, s2} S={s2} and B={s1, s2} S ={s1, s2} and B={s1, s2}

[0175] We assume that S opt ={s1, s2} and B opt ={s1, s2} maximize the equivalent mutual information, i.e., N = v = 2. It follows that during the retransmission interval t, the two source devices si and S2 selected as relay nodes retransmit the same redundancy versions associated with their respective information messages.

[0176] The pre-coded vector transmitted by the selected N=2 relay nodes s1 and s2 is given by: x̃(i) = W(i)D(i)Ux(i), i ∈ {0,..., N2 - 1} x(i) denotes a complex column vector of dimension v comprising the redundancy versions associated with the information messages of the source devices si and S2 and: D(i) = diag[1, e^(-jπi)], i ∈ {0,..., N2-1} that is, for resource elements with even indices i: D(i) = [1 0; 0 1] and for resource elements with odd indices i: D(i) = [1 0; 0 -1]

[0177] The rotation matrix U is fixed and is written as follows: U = [1 1; 1 e^(-jπ)] = [1 1; 1 -1]

[0178] Therefore, we obtain: D(i)U = {[1 1; 1 -1], i even [_\ J]- * im P a i r

[0179] Since N = v = 2: W(i) = I2, i ∈ {0,..., N2 — 1}

[0180] The overall precoding matrix V(i) can therefore be written as follows: V̄(i) = W(i)D(i)U = {[1 0; 0 1][1 1; 1 -1] = [1 1; 1 -1], i even V̄(i) = W(i)D(i)U = 1] j, i odd [1 0; 0 1][-1 1; 1 1] = either after normalization: i pair V̄(i) = W(i)D(i)U = i odd

[0181] The precoding vector v0(i) applied to the redundancy version associated with the source device if corresponds to the l ère column of the precoding matrix V(i) i is: i even Vo (!) = i odd

[0182] The precoding vector (i) applied to the redundancy version associated with the source device S2 corresponds to the 2 ème column of the precoding matrix V(i), i.e.: i pair Vi (i) i odd

[0183] The effective channel applied to the information messages x(i) is given by: H(i)W7(i)D(i)t / = [h® h^]D(i)U f -= [h^ + - h^], i is even i A / 2 L JJ J — [h^ - + h% )], i is odd where and ft® represent the channels between the source device Si and the destination device DEST, and between the source device s2 and the destination device DEST respectively.

[0184] The precoded signal transmitted during the retransmission interval t, using resource elements indexed by i = 0, ..., N2-1, is therefore: x(i) = H(i)W7(i)D(i)t / x(i) | — {(h® + h^RVt1Çi) + (ft® - ft®) / ?7t2(i)}, i is even il ( — ((ft® — ^2 ' i est odd

[0185] We will now describe how mutual information is evaluated by the receiving device DEST in the embodiment described here, for a pair of sets S and B during the selection step E90.

[0186] If x(i) denotes the transmitted vector containing the redundancy versions of the messages from the source devices of the subset S, the signal received at the destination device DEST for the i-th resource element or channel usage ie {0,..., N2 - 1}, is as follows: y(i) = H(i)P(i)x(i) + n = H(i)x(i) + n where W(i) denotes the propagation channel between the relay nodes of set B and the receiving device DEST and n the AWGN noise vector (“Additive White Gaussian Noise” in English).

[0187] For example, we assume that an interference rejection filter of the LMMSE-IRC type (for "Linear Minimum Mean Squared Error - Interference Rejection Combiner") is applied at the receiver by the receiving device D. The output of this filter is: x(i) = F(i) H y(i)with: F(Ï) H = A i H i u H(i A i H i) u+ « ' 1 where A(i) = diag(Pg\ denotes the transmission powers of the versions of redundancy of the source devices of the set S, and R n is the covariance of the noise n, i.e., R n = E{nn H}.

[0188] The signal-to-noise-plus-interference ratio (SINR) for the spatial layer {0,...,v - l} (i.e., the l-th source device of S) and the resource element or channel usage {0,..., N2 - 1} is denoted y (l) with: Or: e L designating the Z-th column of the identity matrix l v .

[0189] Mutual information I sThe redundancy for retransmitting the versions of the source devices in set S via the relay nodes in set B is calculated by the receiving device DEST (via its selection module 3) based on the SINR outputs of the LMMSE-IRC filter for each spatial layer (i.e., for each source device in set S). It is summed over the v spatial layers and averaged over the N2 channel usages. V-1 N2-1 v-lKv-1 Is = ^Ê È = 7 / 2 Ë S 1=0 i=0 1=0 j=0 Yes denotes the mutual information of a Gaussian additive white noise channel for a signal-to-noise ratio equal to y^ given by (in the case of Gaussian inputs where the mutual information is equal to the capacity): / = log2(1 + y, 0) )

[0190] In the preceding expression, Kv is the periodicity of the precoding coefficients in the codebook CODEB(V, v) and crQ) is the occurrence of the precoding matrix VQ) on the N2 channel uses for i {0,..., Kv - 1}, i.e.: N2 mod Kv fâ + l ' ) |fV7| j > N2 mod Kv

[0191] This coefficient can be approximated to a(J) ≈ 1 when N2 is large. We then obtain: h

[0192] Note that to estimate the mutual information I s Knowledge of the propagation channel H(i) or the equivalent propagation channel H(i) (i.e., including precoding) is necessary for the receiving device DEST. Such an estimation can be performed in a way that is known per se.

[0193] For example, to estimate the propagation channel H(i), the receiving device DEST can use uncoded reference signals transmitted by devices in communication system 1 (source devices and, where applicable, relays; in other words, all devices in communication system 1 that could be selected as relay nodes during the cp2 retransmission phase). Such reference signals are typically Sounding Reference Signals (SRS) for a 4G LTE or 5G NR network. This transmission of reference signals can be performed by devices in communication system 1 prior to the power line communication (PLC) transmission phase, assuming that the propagation channels between devices that could be selected as relay nodes do not vary or vary very little. Appendix selection Inputs: set X comprising all potential relay nodes of communication system 1, sets S DEST t-1 and Sj t -i,j £ X Outputs: set of source devices S opt for the retransmission interval t of the retransmission phase and set of nodes B opt corresponding relaying Initialization: Max = 0 1 for all Q in {1,..., Q1111} do # loop over all possible sets of source devices not decoded by DEST 2 e <- False # boolean variable indicating if at least one subset is eligible for a given value of Q 3 for all c > in PctrtQ (5^ do # for each possible subset of source devices not decoded 4 B <- 0 # initialization of set B of relay nodes 5 for all j in Z do # for all potential relay nodes 6 if SJ = Sj then # if node j has decoded all sources in S 7 |B <- BU { / } # it is integrated into set B 8 end if 9 end for 10 if |S| < |B| then # the number of source devices with redundancy versions being retransmitted must not exceed |B| 11 e <- True # set S is eligible 12 Calculate Information equivalent mutual insurance 1^ 13 if\ §> Max then 14 Max <- I5# update of the maximum value of the equivalent mutual information 15 S°P'^ s # update of the sets selected for retransmission B opt <- B 16 end if 17 end if 18 end for 19 if e = False then 20 Break # no eligible subset S e Part Qo for a given QO value 2 1 end if 22 end for

Claims

Demands

1. A method for activating, by a multi-antenna receiving device (D) receiving a plurality of information messages from a plurality of source devices, a retransmission using a plurality of resource elements of at least one redundancy version resulting from an encoding of at least one information message from a source device not decoded by the receiving device, said method comprising, for at least one retransmission interval: a step (E90) of selecting a first set comprising at least one source device not decoded by the receiving device and a second set comprising a plurality of relay nodes knowing each information message of each undecoded source device in the first set, said first and second selected sets satisfying a performance criterion evaluated by taking into account variations on said resource elements of a precoding matrix applied collectively by said relay nodes on the same redundancy version of each information message of each source device in the first set, said precoding matrix being derived from a dictionary known to the receiving device and the relay nodes and varying within this dictionary according to a variation pattern known to the receiving device and the relay nodes; and a step (E100) of activating the relay nodes of the second selected set so that they retransmit, during said retransmission interval, said same redundancy version of each information message from each source device of said first selected set by collectively applying said precoding matrix.

2. Activation method according to claim 1 wherein the activation step (E100) comprises sending retransmission instructions comprising the first and second sets and designating, for each relay node of the second set, a line of the precoding matrix to be applied by said relay node.

3. An activation method according to claim 2 wherein the second set is provided in the form of an ordered sequence of indices identifying the relay nodes of the second set, the order of an index designating the line of the precoding matrix to be applied by the relay node identified by that index.

4. Activation method according to any one of claims 1 to 3 wherein the selection and activation steps are repeated for at least one other retransmission interval until a predefined stopping criterion is verified and considering new redundancy versions at each new retransmission interval.

5. A method for retransmitting, via a relay node selected by a multi-antenna receiving device (DEST), a plurality of information messages from a plurality of source devices, said method comprising: a receiving step (E100) from the receiving device (DEST) of a retransmission instruction comprising a first and a second set selected by the receiving device for a retransmission interval, the first set comprising at least one source device, said to be undecoded, associated with an information message not decoded by the receiving device, and said second set comprising a plurality of relay nodes, including said selected relay node, knowing each information message from each undecoded source device of the first set; and a retransmission step (El 10) during said retransmission interval, using the same resource elements as the other relay node(s) of the second set, for each information message associated with each undecoded source device of the first set, of the same redundancy version from the encoding of said information message, by applying a line designated by said retransmission instruction of a precoding matrix applied collectively by the relay nodes of the second set and from a dictionary known to the receiving device and the relay nodes of the second set, the precoding matrix varying within this dictionary according to said resource elements according to a variation pattern known to the receiving device and the relay nodes.

6. A method according to any one of claims 1 to 5 wherein the precoding matrix varies every x resource elements, x denoting an integer such that 1 < x < N2, where N2 denotes the number of resource elements used during the retransmission interval.

7. A method according to any one of claims 1 to 6 wherein the dictionary has dimension Kv, K denoting an integer such that: | / V| K = I— I + N mod v where N denotes the number of relay nodes in the second set and v the number of undecoded source devices in the first set.

8. Method according to claims 6 and 7 wherein when N2 > Kvx, the precoding matrix is ​​obtained by cyclically traversing the dictionary.

9. A method according to any one of claims 1 to 8 wherein each precoding matrix is ​​constructed from a matrix W of dimensions N xv, where N denotes the number of relay nodes of the second set and v the number of undecoded source devices of the first set, and wherein each column of the matrix W is a linear combination of distinct vectors from a basis of N orthogonal vectors.

10. A method according to claim 9 wherein said N orthogonal vectors, denoted w n , n = 0,, N - 1, are defined by: 1 r j2nn j2n2n j2n(_N w n = -j= |^1 e N e N... e N j

11. Method according to claim 9 or 10 wherein v - 1 columns of matrix W result from a linear combination of |^| distinct basis vectors and one column of the matrix results from a linear combination of Q^j + N mod v distinct basis vectors.

12. A method according to any one of claims 9 to 11 wherein each column of the matrix is ​​a linear combination of distinct adjacent basis vectors.

13. A method according to any one of claims 1 to 12 wherein each dictionary precoding matrix is ​​further constructed from a complex diagonal matrix D of dimensions vxv introducing a diversity of cyclic delays and a complex rotation matrix U of dimensions vxv.

14. A method according to any one of claims 1 to 8 wherein where said first set comprises a single undecoded source device, each precoding matrix is ​​composed of a single column vector chosen from N orthogonal basis vectors by cyclically traversing said basis, where N denotes the number of relay nodes of the second set.

15. A method according to any one of claims 1 to 8 wherein, where the first set comprises a single undecoded source device, each dictionary precoding matrix is ​​composed of a single vector obtained from a product of a complex diagonal matrix D' of dimensions N x N introducing a diversity of small cyclic delays and a column vector U' of dimension N all of whose components are equal to 1, N denoting the number of relay nodes of the second set.

16. A multi-antenna device (DEST) receiving a plurality of information messages from a plurality of source devices, said receiving device being configured to enable retransmission using a plurality of resource elements of at least one redundancy version resulting from an encoding of at least one information message from a source device not decoded by the receiving device, said receiving device comprising modules enabled for at least one retransmission interval and comprising: a selection module (3), configured to select a first set comprising at least one source device not decoded by the receiving device and a second set comprising a plurality of relay nodes knowing each information message from each undecoded source device in the first set,said first and second selected sets satisfying a performance criterion evaluated by taking into account variations on said resource elements of a precoding matrix applied collectively by said relay nodes on the same redundancy version of each information message from each source device of the first set, said precoding matrix being derived from a dictionary known to the receiving device and the relay nodes and varying within this dictionary according to a variation pattern known to the receiving device and the relay nodes; and an activation module (4), configured to activate the relay nodes of the second selected set to retransmit, during said retransmission interval, said same redundancy version of each information message from each source device of said first selected set by collectively applying said precoding matrix.

17. A relay node capable of being selected by a multi-antenna device (DEST) receiving a plurality of information messages from a plurality of source devices, said relay node comprising: a receiving module (7), configured to receive from the receiving device a retransmission instruction comprising a first and a second set selected by the receiving device for a retransmission interval, the first set comprising at least one source device, said to be undecoded, associated with an information message not decoded by the receiving device, and said second set comprising a plurality of relay nodes, including said selected relay node, knowing each information message from each undecoded source device of the first set; and a retransmission module (8), configured to retransmit during said retransmission interval, using the same resource elements as the other relay node(s) of the second set, for each information message associated with each undecoded source device of the first set, of the same redundancy version from the encoding of said information message, by applying a line designated by said retransmission instruction of a precoding matrix applied collectively by the relay nodes of the second set and from a dictionary known to the receiving device and the relay nodes of the second set, the precoding matrix varying within this dictionary according to said resource elements according to a variation pattern known to the receiving device and said relay nodes.

18. Communication system (1) comprising: a multi-antenna receiver device (DEST) according to claim 16, a plurality of source devices (SI,... SM) associated with a plurality of information messages intended for the receiver device, and a plurality of relay nodes according to claim 17, said relay node being able to be said source device or a third device distinct from said source devices.

19. Communication system (1) according to claim 18 using an orthogonal multiple access multiple-relay channel scheme of the OMAMRC type, between source devices, relay nodes and the destination device.

Images