WO2019185552A1

WO2019185552A1 - Communication method

Info

Publication number: WO2019185552A1
Application number: PCT/EP2019/057456
Authority: WO
Inventors: Henri Teboulle; Marc Le Gourrierec
Original assignee: Sagemcom Energy & Telecom Sas
Priority date: 2018-03-30
Filing date: 2019-03-25
Publication date: 2019-10-03

Abstract

Communication method executed in a LoRa network in which terminals communicate with a server by way of gateways from a set of gateways, each gateway being associated with a set of terminals comprising at least one terminal using said gateway to communicate with the server. The method allows a second gateway, in a downlink frame, to relay a response to a message received in an uplink frame by a first gateway, different from the second gateway, even if the second gateway has not received the uplink frame.

Description

COMMUNICATION METHOD

The invention relates to a communication method, executed in a long-range wireless network and allowing low power consumption, a device and a system implementing the method.

The Internet is gradually being transformed into an extensive network, called the "Internet of Things", connecting all kinds of objects that have become connectable. New network requirements then emerged, including wireless network requirements with greater coverage than conventional cellular networks and limiting power consumption of connected equipment. Among these wide range and low power Wide Area Network (LPWAN) wireless networks in English terminology are networks based on the LoRa (registered trademark) technology. Long Range "in Anglo-Saxon terminology). LoRa technology operates in frequency bands known as the ISM Band (Industry, Science and Medical) with frequency bands that can be used freely for industrial, scientific and medical applications. LoRa technology is based on a spread spectrum technology to obtain low bit rate communications with good robustness in a particularly noisy ISM band.

A network based on LoRa technology (called "LoRa network" later) uses a protocol called LoRaWAN. A LoRa network is composed of base stations or gateways ("gateways" in English terminology) generally placed on high points to cover a large geographical area. The gateways are able to detect messages transmitted in their area by devices or terminals ("endpoints" in English terminology) and to trace them to at least one server ("LoRa Network Server (LNS)" in English terminology ) who will treat them.

In a conventional operation of a LoRa network, a terminal wishing to transmit a message (ie data) to the server, transmits this message in a so-called rising frame, conforming to the LoRaWAN protocol. The upstream frame is transmitted in multicast mode ("broadcast" in English terminology). This rising frame is received by at least one gateway. Each gateway having received the frame decodes it and retransmits the message to the server in an HTTP request (hypertext transfer protocol, "HyperText Transfer Protocol" in English terminology). If multiple gateways received the frame, the server receives several HTTP requests containing the message. The server must then designate among the gateways having received the uplink, the gateway to be used to relay a response to the message contained in the uplink. The response is transmitted from the server to the designated gateway in an HTTP request and then point-to-point from the designated gateway to the terminal in a LoRaWAN compliant downlink. In a majority of implementations of LoRa networks, the designated gateway is the one offering the best transmission quality with the terminal having sent the uplink. A disadvantage of these implementations is that the relay of the responses to rising frames between the server and the terminals is often provided by the same gateways which can cause congestion of the downstream path of these gateways.

Another particularity of the LoRa networks is that the gateways are equipments operating in an alternating or half-duplex mode ("half duplex" in English terminology). Therefore, a gateway can not send and receive data from terminals simultaneously. Indeed, the LoRa gateways have very high emission levels, of the order of +27 dBm, compared to their reception level which can be of the order of -140 dBm. Such a difference between transmission and reception levels implies that a transmission of a message would greatly impede simultaneous reception of a message, although transmissions and receptions use two different frequency bands. This constraint of using bridges operating in half-duplex mode greatly hampers the performance of LoRa networks. One possible solution would be to divide the gateways into a transmitting and a receiving part and to geographically separate these two parts. This solution which would allow the gateways thus arranged to transmit and receive simultaneously (the gateways would thus become full-duplex equipment operating in "full duplex" in English terminology) poses a problem of synchronization between the two parties. Indeed, a gateway designated by a server to relay a response to a message it has received must transmit the response at a time set by the server. This time is calculated by the server by adding a predefined duration to a time of reception of the uplink containing the message by said gateway. But the moment of emission is calculated in a base time of the reception part of the gateway which, without synchronization between the transmission part and the reception part of the gateway, is not known to the transmission part. It is common to solve synchronization problems between remote equipment by inserting in each equipment a synchronization module on a common clock such as a GPS module (global positioning system, "Global Positioning System" in English terminology). However, inserting a GPS module into each gateway would increase both the cost and the complexity of the networks.

It is desirable to overcome these disadvantages of the state of the art. In particular, it is desirable to propose a method that allows a LoRa network comprising only half-duplex gateways to offer full duplex functionality, the full duplex functionality being achieved without inserting a synchronization module on a clock. common in each gateway. In addition, it would be desirable for this method to prevent the relay of responses to the messages of the terminals always seeking the same gateways.

It is also desirable to provide a method that is simple to implement and low cost.

According to a first aspect of the present invention, the present invention relates to a communication method, executed in a long-range wireless network and allowing a low power consumption in which terminals communicate with a server via gateways of the present invention. a set of gateways according to a communication technology implying that when the server receives a message from a terminal via a first gateway, it fixes a time at which a gateway must relay a response to said message according to a moment of reception of said message by said first gateway, each gateway being associated with a set of terminals comprising at least one terminal using said gateway to communicate with the server. The method implemented by the server comprises: receiving a request from a first gateway containing a message sent in a frame, said rising frame, by a terminal, said sending terminal; determining a second gateway different from the first gateway to be used to relay a response to said message intended for the sender terminal, the second gateway being determined in a subset of gateways of the set of gateways comprising the first gateway and at least one second gateway, each gateway of the subset being capable of relaying to a terminal a response to a message transmitted in an uplink by said terminal received by any other gateway of said subset, the gateways of the subset having at least one associated terminal in common and being associated with the transmitting terminal; ; transmitting a request containing said response to the second gateway, said response being transmitted by the second gateway to said transmitting terminal in a frame, said downstream frame, at a time indicated in the request, said instant being indicated in a time reference of the first gateway the second gateway taking into account a clock deviation between the first and second gateways to transmit the downlink to said instant.

According to one embodiment, the communication technology is the LoRa technology and each frame exchanged between a gateway and a terminal complies with the LoRaWAN protocol.

According to one embodiment, the set of gateways comprises a plurality of subsets of gateways, each subset being formed by an operator from bridging bridges to relay a response in a downlink to a message. received in an uplink by one of them or by an adaptive subset formation procedure periodically executed under the control of the server with a predefined period.

According to one embodiment, the server transmits to each gateway included in a subset of the characteristics of said subset comprising an identifier for each gateway included in said subset and an identifier for each terminal associated in common between the gateways of the sub-set. -together.

According to one embodiment, the adaptive subset formation procedure includes: classifying the gateway set gateways in descending order of a number of falling frames that each gateway has transmitted during the last predefined period or in descending order a number of upstream frames that each gateway has received during the last predefined period; browse the gateways thus classified in their order of classification and for each gateway, classifying each terminal associated with said gateway in descending order of a number of upstream frames received from said terminal by said gateway during the last predefined period; browse the gateways in their order of classification and for each gateway, define a subset of gateways from said gateway, said gateway, called base gateway, belonging to said subset, the subset definition comprising: as long as a number of terminals in a first set of terminals associated with the subset of gateways being defined is greater than a predefined percentage of a number of terminals of a second set of terminals; terminals associated with the basic gateway: identifying among the gateways thus classified that have not yet been inserted in the subset, a gateway having the most terminals in common with the base gateway; insert the gateway thus identified into the subset; and, forming the first set of terminals as an intersection between the second set of terminals and each set of terminals associated with a gateway inserted in the subset; and, deleting each terminal included in the first set of terminals from the terminal sets associated with the gateways thus classified.

According to one embodiment, when the gateways are ranked in descending order of the number of falling frames that each gateway has transmitted during the last predefined period, only those gateways that have issued at least a predetermined number of falling frames during the last predefined period are classified, and when the gateways are ranked in descending order of the number of upstream frames that each gateway has received during the last predefined period, only those gateways that have received at least a predetermined number of upstream frames during the last predefined period are classified.

According to one embodiment, when a gateway has been inserted into a subset formed from a base gateway, this gateway can no longer be inserted into a subset formed from any other base gateways.

According to one embodiment, the adaptive subset formation procedure stops when all the classified gateways have been inserted into a subset or when all the terminals of the network are included in a set of terminals associated with a sub-set. together.

According to one embodiment, the clock difference existing between two gateways is determined by a synchronization procedure, the synchronization procedure being executed by one of the two gateways, called the initiating gateway, in collaboration with the other gateway, said collaborating gateway and comprising: transmitting an uplink to the collaborating gateway, the initiating gateway putting itself on hold for a predetermined period before listening to a response to said uplink frame, said response being transmitted in a frame down by the collaborating gangway after a period equal to the predetermined period following reception of said frame by the collaborating gateway, a first information representative of a moment of reception of said request by the collaborating gateway to which the duration of the predetermined period has been added being inserted in said response by the collaborating gateway; receiving said response and obtaining second information representative of a moment of reception of said response by the second gateway; and, calculating the clock difference between the initiator gateway and the collaborating gateway as a difference between the second and the first information.

According to one embodiment, the synchronization procedure is executed in each subset for each pair of possible gateways of said subset, each gateway of a pair of gateways being alternatively initiator or collaborator, at predefined regular intervals or adaptive intervals of durations dependent on an evolution of the clock deviations over time.

According to one embodiment, the second gateway is randomly determined in the subset of gateways comprising the first gateway or, randomly, among a predefined number of gateways of said subset offering a communication quality with the highest sending terminal among the gateways of said subset.

According to one embodiment, the second gateway is determined by taking into account a counter value representative of a number of falling frames transmitted by the gateways of said subset in response to messages transmitted in upstream frames by terminals.

According to one embodiment, the determined second gateway is associated with the lowest counter value among the counter values associated with the gateways of said subset.

According to one embodiment, the second gateway is determined by performing a selection procedure comprising: preselecting a predefined number of gateways associated with the lowest counter values among the counter values associated with the gateways of said subset; selecting a gateway associated with a value representative of a quality of communication with the sending terminal greater than or equal to a predefined quality threshold among the gateways of the predefined number of preselected gateways, the chosen gateway becoming the second gateway. According to one embodiment, when no gateway of the predefined number of gateways is associated with a value representative of a quality of communication with the sending terminal greater than or equal to the predefined quality threshold, choose a gateway of the associated subset at the lowest counter value among the counter values associated with the gateways of said subset and associated with a value representative of a quality of communication with the transmitting terminal greater than or equal to the predefined quality threshold; and, if no gateway of said subset is associated with a value representative of a quality of communication with the sending terminal greater than or equal to the predefined quality threshold, choosing a gateway associated with the value representative of a quality of communication with the maximum transmitter terminal among the values representative of a quality of communication with the transmitter terminal associated with the gateways of said subset.

According to one embodiment, when the server receives a plurality of requests each containing the same message sent in an uplink by a transmitting terminal associated with no subset, the method comprises: choosing from the first gateways having reassembled said message to the subscriber; server, a gateway whose reception quality of said frame is greater than or equal to a predefined quality threshold; and, transmitting a request containing said response to the selected gateway, said response being retransmitted by the selected gateway to said transmitting terminal in a downlink.

According to a second aspect of the invention, the invention relates to a server-type device included in a long-range wireless network and allowing a low power consumption in which terminals communicate with the server via gateways. a set of gateways according to a communication technology implying that when the server receives a message from a terminal via a first gateway, it fixes a time at which a gateway must relay a response to said message according to a moment of reception of said message by said first gateway, each gateway being associated with a set of terminals comprising at least one terminal using said gateway to communicate with the server. The device comprises: receiving means for receiving a request from a first gateway containing a message sent in a frame, said rising frame, by a terminal, said transmitting terminal; determining means for determining a second gateway different from the first gateway to be used to relay a response to said message intended for the sender terminal, the second gateway being determined in a subset of gateways of the set of gateways comprising the first gateway and at least one second gateway, each gateway of the subset being capable of relaying to a terminal a response to a message transmitted in an upstream frame by said terminal received by any other gateway of said subset, the gateways of the subset having at least one associated terminal in common and being associated with the transmitting terminal; transmission means for transmitting a request containing said response to the second gateway, said response being retransmitted by the second gateway to the transmitting terminal in a so-called downgoing frame at a time indicated in the request, said instant being indicated in a time reference of the first gateway, the second gateway taking into account a clock deviation between the first and second gateways to transmit the downlink to said instant.

According to a third aspect of the invention, the invention relates to a system included in a long-range wireless network and allowing a low power consumption comprising a server-type device according to the second aspect, a plurality of terminals and a plurality gateways, each gateway of the subset being capable of relaying to a terminal a response to a message transmitted in a rising frame by said terminal received by any other gateway of said subset, the gateways of the subset having at least one associated terminal in common.

According to a fourth aspect of the invention, the invention relates to a computer program, comprising instructions for implementing, by a device, the method according to the first aspect, when said program is executed by a processor of said device.

According to a fifth aspect of the invention, the invention relates to storage means storing a computer program comprising instructions for implementing, by a device, the method according to the first aspect, when said program is executed by a processor. said device.

The characteristics of the invention mentioned above, as well as others, will appear more clearly on reading the following description of an exemplary embodiment, said description being given in relation to the attached drawings, among which: FIG. 1 schematically illustrates a LoRa network in which the invention is implemented;

FIG. 2A schematically illustrates a processing module included in a server;

FIG. 2B schematically illustrates a processing module included in a gateway;

FIG. 3 schematically illustrates a communication method according to the invention;

FIG. 4 schematically illustrates a synchronization procedure according to the invention;

FIG. 5A schematically illustrates a first example of an adaptive procedure for forming subsets of gateways according to the invention;

FIG. 5B schematically illustrates a second example of an adaptive procedure for forming subsets of gateways according to the invention;

FIG. 6 schematically illustrates in detail the formation of subsets of gateways; and,

FIG. 7 schematically illustrates a procedure for choosing a gateway to relay a response to a message from a terminal.

The invention is described later in a LoRa network context. However, the invention applies in other contexts for all types of long-range and low-power wireless networks in which terminals communicate with a central server via half-duplex gateways.

Fig. 1 schematically illustrates an exemplary LoRa 1 network in which the invention is implemented.

In the example of FIG. 1, the LoRa 1 network comprises a server 10, a set of gateways comprising at least two gateways (in Fig. 1 only two gateways 11A and 11B are shown) and at least one terminal (in Fig. 1 only a terminal). 12 is shown). Each gateway communicates with the server 10 via a wired or wireless communication link. For example the gateway 11 A (respectively the gateway 11B) communicates with the server 10 via a wired or wireless communication link 14A (respectively 14B). In addition, each communication between a gateway and a terminal uses a wireless communication link. For example, the gateway 11A (respectively the gateway 11B) communicates with the terminal 12 via a wireless communication link 13A (respectively 13B).

The server 10 comprises a processing module 100. Each gateway comprises a processing module. For example, the gateway 11A (respectively 11B) comprises a processing module 110A (respectively 110B).

In one embodiment, certain gateways may communicate with each other. To do this, each gateway wishing to communicate with another gateway uses a feature that is to allow a gateway to pass temporarily for a terminal. For example, the gateways 11A and 11B can communicate with each other via a wireless communication link 15.

It should be noted that in the LoRa 1 network, the communications between the terminals and the gateways and the communications between the gateways use frames compatible with the LoRaWAN protocol. The document LoRaWAN 1.1 ("draft LoRaWAN 1.1" in English terminology) of August 2015 defines the communications between the terminals and the gateways of a LoRa network.

Fig. 2A schematically illustrates an example of hardware architecture of the processing module 100 included in the centralized server 10.

According to the example of hardware architecture shown in FIG. 2A, the processing module 100 then comprises, connected by a communication bus 1000: a CPU or CPU ("Central Processing Unit" in English terminology) 1001; random access memory RAM ("Random Access Memory" in English terminology) 1002; a ROM memory ("Read Only Memory" in English terminology) 1003; a storage unit such as a hard disk or a storage medium reader, such as an SD card reader ("Secure Digital" in English terminology) 1004; at least one communication interface 1005 enabling the processing module 100 to communicate with other modules or devices. For example, the communication interface 1005 enables the processing module 100 to communicate with the gateways of the LoRa network 1 such as the gateways 11A and 11B or with remote servers belonging to the cloud ("cloud" in English terminology). .

The processor 1001 is capable of executing instructions loaded into the RAM 1002 from the ROM 1003, an external memory (not shown), a storage medium (such as an SD card), or a communication network. When the server 10 is powered up, the processor 1001 is able to read RAM 1002 instructions and execute them. In one embodiment, these instructions form a computer program causing the processor 1001 to completely or partially implement the methods described below in relation with FIGS. 3, 4, 5A, 5B and 6.

Fig. 2B schematically illustrates an example of hardware architecture of the processing module 110A included in the LoRa gateway 11 A. The processing module 110B is identical.

According to the example of hardware architecture shown in FIG. 2B, the processing module 110A then comprises, connected by a communication bus 1100: a processor or CPU 1101; a random access memory RAM 1102; ROM ROM 1103; a storage unit such as a hard disk or a storage medium reader, such as an SD card reader 1104; at least one communication interface 1105 enabling the processing module 110A to communicate with other modules or devices. For example, the communication interface 1105 enables the processing module 110A to communicate with the server 10, with other gateways of the LoRa network 1 such as the gateway 11B or with terminals of the LoRa network 1 such as the terminal 12. one embodiment, the processing module 110A further comprises a counting module capable of supplying a count value CPRTi encoded on "32" bits and incremented by one unit at each microsecond, where i is an identifier of the processing module ( ie i = 110A for the count module of the processing module 110A).

The processor 1101 is capable of executing instructions loaded into the RAM 1002 from the ROM 1003, an external memory (not shown), a storage medium (such as an SD card), or a communication network. When the gateway 11A is turned on, the processor 1101 is able to read instructions from RAM 1102 and execute them. In one embodiment, these instructions form a computer program causing the processor 1101 to completely or partially implement the methods described hereinafter with reference to FIGS. 3 and 4.

The methods described in connection with FIGS. 3, 4, 5A, 5B and 6 can be implemented in software form by executing a set of instructions by a programmable machine, for example a DSP ("Digital Signal Processor" in English terminology) or a microcontroller, or be implemented in form hardware by a machine or a dedicated component, for example an FPGA ("Field Programmable Gate Array" in English terminology) or an ASIC ("Application-Specific Integrated Circuit" in English terminology).

Fig. 3 schematically illustrates a communication method according to the invention.

In a step 300, the terminal 12 transmits a LoRaWAN compliant frame to the server 10. This frame, called the uplink frame, is transmitted in multicast mode so that each gateway in range of the terminal 12 receives this upstream frame. In the example of the LoRa 1 network, it is assumed that the gateway 11 A, called the LM gateway (Uplink, in English terminology), receives each uplink transmitted by the terminal 12.

In a step 301, the processing module 110A of the gateway 11A therefore receives the rising frame.

When a gateway of a LoRa network receives an uplink containing a message destined for a server, it decodes it and inserts the message contained in this uplink into an HTTP request, called a rising HTTP request, and then transmits this rising HTTP request to LoRa network server in point-to-point. In a step 302, the processing module 110A decodes and inserts the message contained in the upstream frame into a rising HTTP request, and then transmits this upstream HTTP request to the server 10 in point-to-point. In addition, the request includes information representing a time of reception of the uplink by the gateway 11 A. This information representative of the reception time of the uplink frame is supplied to the processing module 110A by its counting module. . This information representative of the moment of reception of the rising frame has therefore been measured in a time frame of the gateway 11 A.

In a step 303, the processing module 100 of the server 10 receives the uplink HTTP request containing the message sent in the uplink by the terminal 12.

In a step 304, the processing module 100 determines a second gateway, called LD gateway ("downstream" in English terminology) different from the gateway 11A (ie LM gateway) to be used to relay a response to the content message. in the uplink for the terminal 12. The set of gateways LoRa network 1 is composed of a plurality of subsets of gateways. Each subset of gateways includes gateways capable of relaying to a terminal a response to a contained message in a rising frame, transmitted by said terminal, received by any other gateway of said subset. Each gateway of a subset may alternatively be an LM gateway or an LD gateway. Each gateway of a subset that has received a rising frame is considered to be an LM gateway. Each gateway of a subset relaying a response to a message contained in an uplink received by another gateway of the subset, that is, relaying a response to a message contained in an upstream frame for the account another gateway of the subset, is considered an LD gateway. Each subset therefore comprises at least a first and a second gateway. The first gateway can alternatively be an LM gateway or an LD gateway. The second gateway may be an LD gateway when the first gateway is an LM gateway, or an LM gateway when the first gateway is an LD gateway. Gateways of a subset have at least one associated terminal in common with each other. A gateway is considered to be associated with a terminal when it has received upstream frames from said terminal for a recent period. Furthermore, each gateway of a subset is associated at least with the terminal having sent the uplink received by a gateway LM of the subset. In step 304, the gateway LD is therefore determined in a subset comprising the gateway 11A and in which each gateway is at least associated with the terminal 12. In the example described with reference to FIG. 3, the determined LD gateway is the gateway 11B. Note that, although the upstream frame was transmitted in multicast mode, the gateway 11B did not necessarily receive this frame. We describe later in connection with Figs. 5A and 5B two examples of procedures for creating subsets of gateways from the set of gateways LoRa network 1. It is noted that in the invention is the server 10 or a remote server cloud to the request of the server 10 which holds characteristics of each subset. The server 10 (or remote server of the cloud under the control of the server 10) has either defined these subsets, or obtained these subsets during a LoRa 1 network configuration by an operator. These characteristics comprise an identifier for each gateway belonging to said subset and an identifier for each terminal associated in common between all the gateways of said subset. The server 10 transmits the characteristics of each subset to each gateway belonging to said subset each time said subset is defined or updated. In a step 305, the processing module 100 transmits an HTTP request containing said response to the gateway 11B. This request further comprises information representative of a desired transmission time of the downlink corresponding to said request by the gateway 11B. This transmission time is calculated by the processing module 100 with respect to the information representative of the reception time of the uplink by the gateway 11 A. This calculation consists in adding a predefined duration D to said reception time. For example, D = ls. The instant of transmission is thus calculated with respect to a time reference of the gateway 11A (ie of the gateway LM).

In a step 306, the processing module 110B of the gateway 11B receives the HTTP request containing said response.

In a step 307, the processing module 110B generates a downlink containing the response and transmits it point-to-point to the terminal 12 at a time corresponding to the information representative of the desired transmission time. As described above, the desired transmission time has been calculated with respect to the time reference of the gateway 11 A. Or the gateway 11B does not operate with the same time reference. Indeed, the time reference of the gateway 11B is given by the counting module of the processing module 110B, which, although identical to the counting module of the processing module 110A, is generally offset with respect to the module counting module 110A treatment. Moreover, even if the two counting modules had been synchronized at a time, they can both undergo different drifts over time. To compensate for this offset, the processing module 110B takes into account a clock deviation D between the gateway 11A and the gateway 11B. This clock deviation D is obtained by the processing module 110B by regularly implementing a synchronization procedure described hereinafter with reference to FIG. 4.

In a step 308, the terminal 12 receives the downlink containing the response to the message contained in the upstream frame.

In one embodiment, during step 304, if the subset of gateways in which the gateway 11B (ie the LD gateway) has been determined comprises a plurality of gateways capable of becoming an LD gateway on behalf of the gateway 11A (ie the LM gateway), the processing module 100 determines the LD gateway among the gateways of the plurality in a random manner. In one embodiment, during step 304, if the subset of gateways in which the gateway 11B has been determined comprises a plurality of gateways capable of becoming an LD gateway on behalf of the gateway 11A, the module processing 100 determines the gateway LD among n number of gateways of the plurality offering the best quality of communication with the terminal 12 in a random manner. For example, if the subset includes a number m of gateways, m being equal to 100, n = 10.

Fig. 4 schematically illustrates a synchronization procedure according to the invention.

The synchronization procedure is executed, in each subset, for each possible pair of gateways included in said subset. The synchronization procedure between two gateways of a pair is unidirectional. Indeed, assuming that one of the two gateways of the pair, called initiating gateway, is the one that initiated the synchronization procedure, the synchronization procedure allows the initiating gateway to know the clock difference D between the two bridges of the pair. The other gateway of the pair, said collaborating gateway, not having initiated the synchronization procedure, does not know the clock deviation D at the end of the execution of the synchronization procedure. It is therefore necessary to execute the synchronization procedure twice so that the two gateways of a pair know the clock deviation D, each gateway of a pair becoming alternatively initiator and collaborator. This procedure is executed for example at predefined regular intervals of duration T. In an embodiment 70s.

In FIG. 4, the synchronization procedure is executed between the gateways 11A and 11B, the gateway 11B being the one that initiates said procedure. In this example therefore, only the gateway 11B knows the clock deviation D at the end of the execution of the synchronization procedure.

In a step 400, the processing module 110B transmits an upstream frame in multicast mode. To do this, the gateway 11B behaves like a terminal called TERM l 1B. The upstream frame is considered as an empty frame since it does not contain any useful data but only LoRaWAN compliant header data including an TERM addr address l 1B of the TERM terminal 11B. In a step 401, following this transmission, the processing module 110B goes on standby for a duration D. In one embodiment, D = 1 s.

At the end of the duration D, the processing module 110B listens for a response to its empty uplink in a step 402.

In a step 403, the processing module 110A of the gateway 11 A receives the empty rising frame of the gateway 11B. The processing module 110A recognizes the terminal TERM 11B (and therefore the gateway 11 B) with the addr address TERM l 1B contained in the empty uplink. Indeed, the address addr TERM l 1B is an identifier of the gateway 11B contained in the characteristics of the subset including the gateway 11A transmitted by the server 10 to all the gateways included in this subset.

In a step 404, the processing module 110A saves a current value CPRT _110A of its counting module. The current value CPRT _110A is information representative of a time of reception by the gateway 11A of the empty uplink from the TERM terminal 1B (ie of the gateway 11B).

In a step 405, following reception of the empty uplink, the processing module 110B goes on standby for the duration D.

At the end of the duration D, the processing module 110A transmits a response to the empty uplink in a step 406. This response is transmitted in the form of a point-to-point downlink to the gateway 11 B. This response comprises the information representative of the instant of reception by the gateway 11 A of the empty uplink to which the duration D has been added, that is to say CPRT _110A + D.

In a step 407, the processing module 110B receives the downlink containing the response.

In a step 408, the processing module 110B saves a current value CPRT _110B of its counting module. The current value CPRT _110B is information representative of a time of reception by the gateway 11B of the downlink coming from the gateway 11 A.

In a step 409, the processing module 110B calculates a clock deviation D between the clocks of the gateways 11A and 11B:

D = CPRT _11QB - (CPRT _11QA + D)

In a step 410, the processing module 110B stores the clock deviation D. The clock deviation D is used in the step 307. In a step 411, the processing module 110B waits for the duration of the regular preset interval T and then returns to step 400 to update the clock deviation D.

In one embodiment, the duration of the interval between two launches of the synchronization procedure is not predefined but adaptive. For example, this interval duration, denoted T _adap depends on a change in deviations D over time. If the processing module of the initiating gateway (here gateway 110B) finds that an absolute value of the clock deviation | D | then increases the duration of the interval T is _adapted diminished. If the absolute value of the clock deviation | D | decreases, then the duration of the _adaptive interval T is increased. For example, if _curr is the last calculated difference and A _pre is the previous difference:

In the example of FIG. 4, we show how the gateway 11B obtains the clock deviation D between the gateway 11A and the gateway 11B. In order for the gateway 11A to obtain the clock deviation D between the gateway 11 A and the gateway 11B, it must initiate the synchronization procedure and therefore execute the steps 400, 401, 402, 407, 408, 409, 410 and 411, the gateway 11B executing steps 403, 404, 405 and 406.

In one embodiment, each gateway of a subset implements, via its processing module, a software module (i.e. an application) specific for each other gateway of said subset with which it forms a pair. For each other gateway, this software module manages the reception of the empty uplink (step 403), each processing applied following this reception (step 404 and 405) and the transmission (step 406) of the downlink containing the representative information the time of reception of the empty uplink to which the duration D has been added.

As we have seen above, the invention is based on the use of subsets of gateways. Each subset of gateways comprises gateways capable of relaying to a terminal a response to a message contained in an upstream frame, transmitted by said terminal, received by any other gateway of said subset.

In one embodiment, the subsets are obtained during a configuration of the LoRa 1 network by an operator. In this configuration, the operator defines the subsets. For example, the operator forms subsets from bridges which he knows will be able to supplement each other. to relay in a downlink a response to a message received in an upstream frame by one of them. For this, the operator can rely on a geographical position of each gateway. Thus, a subset associated with a first gateway may be formed from gateways within a radius of R of the first gateway (eg, R = 10 kilometers). In this embodiment, it is assumed that a terminal that is capable of transmitting an upstream frame to the first gateway is capable of receiving a downlink from a second gateway located within a radius of R kilometers around the first gateway.

This first embodiment, besides the fact that it is tedious to implement if the LoRa network comprises a large number of gateways and terminals, is not adaptive. It does not take into account a possible malfunction of a gateway or a change of environment of a gateway (for example a construction of a building in the vicinity of a second gateway that would prevent communications between the first and the next. second gateways or between the second gateway and the terminal having transmitted the upstream frame). In addition, each insertion of a new gateway into the LoRa 1 network requires operator intervention to modify the characteristics of at least some subsets. This embodiment can, however, be used when the LoRa 1 network is started when the server 10 has no information on the gateways actually associated with each terminal.

Figs. 5A and 5B schematically illustrate a first example and a second example of an adaptive procedure for forming subsets of gateways according to the invention.

Each adaptive gateway subsystem training procedure is periodically executed with a period P by the server 10 or by a remote server of the cloud under the control of the server 10. P is for example a period of two hours.

The first and second examples of adaptive procedure for subset subsystem formation are adapted to different contexts.

The context of the example of FIG. 5A is that of a LoRa network in which, the LoRa network gateways can transmit on the same channels, said gateways can transmit downlink frames to and receive upstream frames of each LoRa network terminal that is within their reach. The context of the example of FIG. 5B is that of a LoRa network in which the LoRa network gateways can not receive upstream frames from each LoRa network terminal that is within their reach. In this context, it is considered that the gateways use a dynamic reception channel which traverses the entire frequency band allocated to LoRa so as to be able to receive terminals that they could not otherwise receive.

For the execution of these training procedures of the subsets of gateways, the server 10 (or a remote server of the cloud under the control of the server 10) uses information on the requests that it receives from the gateways of the LoRa 1 network or it sends to the LoRa 1 network gateways. This information includes for each request:

• time of transmission or receipt of the request;

• type of request: rising or falling;

• identifier of the terminal concerned by the request;

· Identifier of the gateway concerned by the request.

Subsequently, we describe the adaptive procedures for forming subsets of gateways when they are executed by the processing module 100 of the server 10. These procedures could just as easily be executed by an equivalent processing module of a server remote from the cloud which would then provide a result of executing the procedures to the server 10.

Subsequently, it is assumed that the LoRa 1 network includes N _p gateways.

With reference to FIG. 5A, in a step 501, the processing module 100 classifies the LoRa network gateways 1 in decreasing order of a number of falling frames that each gateway has sent to the terminals of the LoRa 1 network during the last period P.

In one embodiment, the processing module 100 classifies only those gateways that have transmitted at least a predetermined number of descending frames during the last period P. In one embodiment, l = 100. The other gateways are not considered in the rest of the procedure. There remains only N _p £ N _p gateways to consider, each noted GW _L thereafter where i is an index varying from '1' to N _p .

In a step 502, the processing module 100 traverses the GWi gateways in ascending order of the index i. For each gateway GW the processing module 100 classifies each terminal associated with the gateway GWi in descending order of a number of upstream frames received from said terminal by said gateway GW _L during the last period P. This gives the gateway GW _t a set of terminals classified GT _L comprising M _t terminals associated with the gateway GW _t .

In a step 503, the processing module 100 traverses the GWi gateways in increasing order of the index i, but this time to form the subsets.

Fig. 6 schematically illustrates in detail the formation of the subsets corresponding to step 503.

In a step 600, the processing module 100 initializes a variable i to "1". The variable i makes it possible to traverse the gateways GW _L. A gateway GW _L is a gateway from which the processing module 100 forms a subset, the gateway GWi belonging to this subset. In step 600, the processing module also initializes a variable n _p 'to the value of the number of gateways to consider N _p .

In a step 601, the processing module 100 initialises a set of terminals ^{GT1 NTER} to the set of terminals GT _L associated with the gateway GW _t .

In a step 602, the processing module 100 initializes a variable K to the number of gateways that can be considered to form the subset from the gateway GW _t . The variable K is therefore initialized to n _p '- 1 because the gateway GW _t is removed from the set of gateways that can be considered because by definition it is part of the subset being formed.

In a step 603, the processing module 100 identifies, among the remaining K gateways, a gateway GW _j having the most terminals in common with the gateway G Wi.

When the gateway GW _j is found, the processing module 100 inserts, in a step 604, the gateway GW _j into the subset formed from the gateway GW _t . In other words, the processing module adds an identifier of the gateway GW _j to a set of identifiers representing the subset formed from the gateway GW _L , this subset comprising an identifier of the gateway G Wi .

In a step 605, the processing module 100 resets the set of terminals ^{GT1 NTER} as the intersection between the current set of terminals ^{GT1 NTER} and the set of terminals GT _j associated with the gateway GW _j:

GT j INTER _ = GT [ ^{NI ER} n GT _j In a step 606, the processing module 100 checks whether the number of terminals NB_GTl ^NTER in the set of terminals ^{GT1 NTER} is greater than or equal to a predefined percentage Q of the number M _t of terminals in the set of terminals GW _t . In one embodiment, Q = 25%.

If NB_GT- ^NTER ³ Q x M, the processing module 100 continues with a step 607. Otherwise, the processing module continues with a step 609.

In step 607, the processing module decrements the variable K by one unit to switch to another gateway GW _j to be inserted in the subset of gateways formed from the gateway GW _t .

If during a step 608 the variable K is equal to zero, the processing module executes step 609. Otherwise, the processing module returns to step 603.

In step 609, the processing module 100 deletes the terminals belonging to the set of terminals ^{GT1 NTER} of all sets of terminals GTi. In this way, the same terminal can not be associated with several subsets.

During a step 610, the processing module 100 verifies that following the deletion of the terminals belonging to all the terminals GT [ ^NTER , there remain terminals in at least two sets GTi. If there are no longer any GT _L sets including terminals, this means that all terminals have been associated with a subset. If there is only one set GTi including terminals, it means that we can never associate said terminals to a subset. In one embodiment, for those terminals that are not associated with a subset, the behavior of a conventional LoRa network applies. In these two cases, if there is no more GTi set including terminals or if only a set GT _t remains the processing module 100 terminates the algorithm of FIG. 6 in step 613 and thus to the adaptive procedure for forming gate subsets of FIG. 5A.

Otherwise, during a step 611, the processing module 100 updates the variables i and n _p '. In one embodiment, a gateway may appear in more than one subset. In this embodiment, in step 610, the processing module 100 increments the variable i by one unit to switch to another gateway GWi and the variable n _p 'remains unchanged.

Step 611 is followed by a step 612 in which the processing module 100 checks whether all the bridges from which a subset can be formed were considered. In the embodiment where a gateway can appear in several subsets, the processing module 100 checks whether the variable i is less than N _v '. If this is the case, the processing module 100 returns to step 601. Otherwise, the processing module 100 terminates the algorithm of FIG. 6 during a step 613 and therefore to the adaptive procedure for forming gateway subassemblies of FIG. 5A.

Note that at the end of the algorithm of FIG. 6, each of the N _v 'gateways is included in a subset, each gateway of a subset being associated with each terminal belonging to the intersection GT [ ^NTER of sets of terminals GTi associated with each gateway of said subset .

According to the algorithm of FIG. 6, a gateway may appear in more than one subset.

In one embodiment, when a gateway GW _j has been inserted into a subset formed from a gateway GWi during an execution of step 604, this gateway GW _j can no longer be inserted into a subscriber. subset formed from all other gateways. In this embodiment, a gateway can therefore appear only in a single subset. In this embodiment, in step 611, the variable i takes a value of an index of a gateway GW _L that does not belong to a subset already formed. Moreover, the variable n _p 'takes a value corresponding to a number of gateways included in the N _p gateways not appearing in any subset already formed. In step 612, the processing module 100 checks whether there are at least two gateways that do not appear in any subset already formed. If no gateway remains, it means that all gateways have been inserted into a subset. If there is only one gateway, it means that it is not possible to form a subset from the remaining gateway since at least two gateways are needed to form a subset. In both cases, the processing module 100 terminates the algorithm of FIG. 6 in step 613 and thus to the adaptive procedure for forming gate subsets of FIG. 5A.

Referring now to FIG. 5B, in a step 511, the processing module 100 classifies the LoRa network gateways 1 in descending order of a number of rising frames that each gateway has received from the terminals of the LoRa 1 network during the last period P. Similar to the procedure of FIG. 5A, in one embodiment, the processing module 100 only classifies the gateways that have received at least a predetermined number of upstream frames during the last period P. The other gateways are not considered in the following procedure. It remains then only N _p £ N _p gateways to consider noted GW _L thereafter where i is a variant index of

"1" to N _p .

Step 511 is followed by steps 512 and 513 which are strictly identical to steps 502 and 503. In one embodiment, each uplink HTTP request transmitted by a gateway GW _t containing a message sent in an uplink by a terminal TERM _k ( k ranging from "0" to the number of terminals included in the LoRa network 1) further comprises a value q _ik representative of a reception quality of said uplink by said gateway GW _t . For each pair of gateway GM / _j and terminal TERM _k LoRa network 1 possible, the processing module 100 stores a variable Q _ik . The variable Q _ik is initialized by the processing module 100 to a value of a predefined quality threshold SQ at the beginning of a predefined period D. For example, the period D is set to the period P. In another embodiment, the period D is independent of the period P and for example equal to "24" hours. At each reception of a rising frame from a terminal TERM _k by a gateway GW, the processing module 100 retrieves the value q _ik contained in the rising HTTP request and stores this value in the variable Q _ik . In the Q _ik variable, the processing module 100 therefore stores a value representative of the reception quality of the last uplink received by the gateway GW _{t of} said terminal TERM _k . The processing module 100 therefore stores reception quality values Q _ik representative of the last information of known quality for couples consisting of a gateway GW _t and a terminal TERM _k .

In this embodiment, when at least one gateway GW _t relays in a rising HTTP request a message transmitted in an upstream frame by a TERM terminal _k which is not associated with a subset, the processing module 100 chooses among gateways GW _t having reassembled said message, a gateway GW _L whose reception quality Q _ik of said frame is greater than or equal to the predefined quality threshold SQ. The processing module 100 can only choose among the gateways GW _t having received the rising frame to relay a response to said message since the gateways GW _t having sent a rising HTTP request containing a message from a TERM terminal _k that is not associated with any subset is not expected to be synchronized. The choice is made, for example, randomly among the gateways GW _t whose reception quality of said frame is greater than or equal to the predefined quality threshold SQ. If no gateway GW _L has a reception quality greater than or equal to the predefined quality threshold SQ, the processing module 100 selects the gateway having the highest reception quality.

In one embodiment, said embodiment with a counter, the processing module 100 associates a counter CPT Ri with each gateway GW _L of the network LoRa 1. The counter CPT Ri makes it possible to count for each gateway GW a number of transmitted downlink frames. by said gateway GW _L in response to messages transmitted in upstream frames by terminals TERM _k (ie the number of times that a gateway GW _L has been chosen to relay a response to a message transmitted in a rising frame by a terminal TERM _k ). In this embodiment, each counter CPT Ri is reset at each beginning of period D. In this embodiment, a gateway GWi has been chosen to relay a response to a message transmitted in a rising frame by a terminal. TERM _k which is not associated with a subset or which is associated with a subset, the counter CPTRi associated with said gateway GW _t is incremented by one unit. The counter embodiment makes it possible to better distribute the load of the transmission of the responses. Indeed, a gateway that has been little used to relay responses to messages transmitted in uplink frames TERM terminals by _k , is preferred over another gateway more often used to relay responses.

In a first variant of a procedure for selecting a gateway to relay a response to a message from a terminal adapted to the counter embodiment, in step 304, if the subset of gateways GWi in which has been determined the gateway 11B (ie the LD gateway) comprises a plurality of gateways GWi likely to become an LD gateway on behalf of the gateway 11A (ie the LM gateway), the processing module 100 determines the gateway LD among gateways GWi of the plurality using the counter values CPT Ri associated with each gateway GWi of the plurality. The processing module 100 thus chooses the gateway GW _t associated with the lowest counter value CPTR _t . When the lowest counter value is shared by several GW gateways, the processing module 100 randomly chooses a gateway GW _t among the gateways GW _t associated with said counter value. The counter CPT Ri of the gateway GW _t chosen is then incremented by one unit.

Fig. 7 schematically illustrates a second variant of a gateway selection procedure to relay a response to a message from a terminal adapted to the counter embodiment. The method of choice illustrated in FIG. 7 corresponds to a particular implementation of step 304.

Here again it is assumed that the subset of gateways in which the LD gateway is to be determined comprises a plurality of gateways capable of becoming an LD gateway on behalf of the LM gateway.

In a step 700, the processing module 100 preselects a number L of gateways GfV _j . The GW L _L L preselected gateways are gateways GW _L said subset associated with the lowest CPT Ri counter values. In one embodiment, L = 5.

In a step 701, the processing module 100 checks whether at least one of the L gateways GWi preselected is associated with a reception quality value Q _ik for an uplink transmitted by the terminal 12 (which then acts as terminal TERM _k having transmitted the upstream frame) greater than or equal to the predefined quality threshold SQ.

If only one preselected gateway GWi is associated with a reception quality value Q _ik greater than or equal to the predefined quality threshold SQ, during a step 702, the processing module 100 chooses this gateway to relay a response to the message contained in the rising frame sent by the terminal 12.

If several preselected GWi gateways are associated with a reception quality value Q _ik greater than or equal to the predefined quality threshold SQ, during step 702, the processing module 100 randomly selects one of these gateways GM / _j to relay a response to the message contained in the uplink sent by the terminal 12.

During step 702, the processing module 100 increments by one unit the value of the counter CPT Ri of the selected gateway GWi.

If each preselected GWi gateway is associated with a reception quality value Q _ik less than the predefined quality threshold SQ, the processing module 100 proceeds to a step 703.

In step 703, the processing module 100 determines whether at least one gateway of the subset is associated with a reception quality value Q _ik greater than or equal to the predefined quality threshold SQ. If this is the case, the processing module 100 proceeds to a step 705.

In step 705, the processing module 100 chooses, among the gateways G Wi of the subset, the gateway GW _L associated with a reception quality value Q _ik greater than or equal to the predefined quality threshold SQ having the value of counter CPT R _t the weakest. If the lowest CPTRi counter value is shared by several GW gateways, the processing module 100 randomly chooses a GWi gateway among these gateways. During step 705, the processing module 100 increments the value of the counter CPTRi of the selected gateway GW _t by one unit.

If each gateway of the subset is associated with a reception quality value Q _ik less than the predefined quality threshold SQ, the processing module 100 executes a step 704. In the step 704, the processing module 100 selects the GWi gateway associated with the highest Q _ik value among the gateways of the subset. If the highest Q _ik value is shared by several GW gateways, the processing module 100 randomly chooses a GW gateway _t among these gateways. In step 704, the processing module 100 increments by one the value of the counter CPT Ri of the gateway GWi chosen.

Claims

A communication method, executed in a long range wireless network and allowing low power consumption in which terminals communicate with a server (10) via gateways (11A, 11B) of a set of gateways according to a communication technology implying that when the server receives a message from a terminal via a first gateway, it fixes a time at which a gateway must relay a response to said message according to a time of reception of said message by said first gateway , each gateway being associated with a set of terminals comprising at least one terminal using said gateway for communicating with the server, characterized in that the method implemented by the server comprises:

receiving (303) a request from a first gateway (11A) containing a message sent in a frame, said rising frame, by a terminal (12), said transmitting terminal;

determining (304) a second gateway (11B) different from the first gateway to be used for relaying a response to said message intended for the sender terminal, the second gateway being determined in a subset of gateways of the set of gateways comprising the first gateway and at least one second gateway, each gateway of the subset being capable of relaying to a terminal a response to a message transmitted in a rising frame by said terminal received by any other gateway of said subset, the gateways of the sub-set. -etembly having at least one associated terminal in common and being associated with the transmitting terminal, the second gateway being determined by executing a selection procedure comprising:

preselecting (700) a predefined number of gateways associated with the lowest counter values among counter values associated with the gateways of said subset representative of a number of falling frames transmitted by the gateways of said subset in response to messages transmitted in upstream frames by terminals;

selecting (701, 702) a gateway associated with a value representative of a quality of communication with the sending terminal greater than or equal to a predefined quality threshold among the gateways of the predefined number of preselected gateways, the chosen gateway becoming the second gateway. transmitting (305) a request containing said response to the second gateway, said response being retransmitted (307) by the second gateway to said transmitting terminal in a frame, said downlink frame, at a time indicated in the request, said instant being indicated in a time reference of the first gateway, the second gateway taking into account a clock deviation between the first and second gateways to transmit the downlink to said instant.

2) Method according to claim 1, characterized in that the communication technology is LoRa technology and each frame exchanged between a gateway and a terminal is in accordance with the LoRaWAN protocol.

3) Method according to claim 1 or 2, characterized in that the set of gateways comprises a plurality of subsets of gateways, each subset being formed by an operator from bridges that can be substituted between them to relay a responding in a downlink to a received message in an upstream frame by one of them or by an adaptive subset formation procedure periodically executed under the control of the server with a predefined period. 4) Method according to claim 3, characterized in that the server transmits to each gateway included in a subset of the characteristics of said subset comprising an identifier for each gateway included in said subset and an identifier for each associated terminal in between the bridges of the subset.

5) Method according to claim 3 or 4, characterized in that the adaptive procedure for forming the subassemblies comprises:

classifying (501) gateways of the gateway set in descending order of a number of falling frames that each gateway issued during the last predefined period or in descending order of a number of upstream frames that each gateway received during the last predefined period;

browse (502) the gateways thus classified in their order of classification and for each gateway, classify each terminal associated with said gateway by descending order of a number of upstream frames received from said terminal by said gateway during the last predefined period;

browse (503) the gateways in their order of classification and for each gateway, define a subset of gateways from said gateway, said gateway, called base gateway, belonging to said subset, the subset definition comprising :

as long as a number of terminals in a first set of terminals associated with the subset of gateways being defined is greater than a predefined percentage of a number of terminals of a second set of terminals associated with the base gateway ( 606):

identifying (603) among the gateways thus classified that have not yet been inserted into the subset, a gateway having the most terminals in common with the base gateway;

inserting (604) the gateway so identified in the subset; and,

forming (605) the first set of terminals as an intersection between the second set of terminals and each set of terminals associated with a gateway inserted in the subset;

and, deleting each terminal included in the first set of terminals from the terminal sets associated with the gateways thus classified.

6) Method according to claim 5, characterized in that, when the gateways are classified in descending order of the number of falling frames that each gateway has transmitted during the last predefined period, only gateways that have issued at least a predetermined number of frames during the last predefined period are classified, and when the gateways are ranked in descending order of the number of upstream frames that each gateway received during the last predefined period, only those gateways that received at least a predetermined number of upstream frames during the last predefined period are ranked.

7) Method according to claim 5 or 6, characterized in that when a gateway has been inserted into a subset formed from a base gateway, this gateway can no longer be inserted into a subset formed by from all other basic gateways. 8) Method according to claim 5, 6 or 7, characterized in that the adaptive subset formation procedure stops when all the classified gateways have been inserted in a subset or when all the terminals of the network are included in a set of terminals associated with a subset.

9) Method according to any one of the preceding claims, characterized in that the existing clock gap between two gateways is determined by a synchronization procedure, the synchronization procedure being executed by one of the two gateways, said gateway initiator, in collaboration with the other gateway, said collaborating gateway and comprising:

transmitting (400) an upstream frame to the collaborating gateway, the initiating gateway queuing (401) for a predetermined period of time before listening for a response (402) to said upstream frame, said response being transmitted (406) in a downlink frame by the collaborating gateway after a period equal to the predetermined period following reception (403) of said frame by the collaborating gateway, a first piece of information representative of a moment of reception of said request by the collaborating gateway to which has been added the duration of the predetermined period being inserted in said response by the collaborating gateway;

receiving (407) said response and obtaining (408) second information representative of a time of receipt of said response by the second gateway; and calculating (409) the clock difference between the initiator gateway and the collaborating gateway as a difference between the second and the first information.

10) Method according to claim 9, characterized in that the synchronization procedure is executed in each subset for each pair of possible gateways of said subset, each gateway of a pair of gateways being alternatively initiator or collaborator, at intervals regular predefined or adaptive intervals of durations dependent on a change in clock deviations over time. 11) Method according to any one of the preceding claims, characterized in that the second gateway is determined randomly in the subset of gateways comprising the first gateway or,

randomly among a predefined number of gateways of said subset offering a quality of communication with the highest transmitting terminal among the gateways of said subset.

12) Method according to claim 1, characterized in that, when no gateway of the predefined number of gateways is associated with a value representative of a quality of communication with the sending terminal greater than or equal to the predefined quality threshold, choose a subset gateway associated with the lowest counter value among the counter values associated with the gateways of said subset and associated with a value representative of a quality of communication with the sending terminal greater than or equal to the quality threshold predefined; and,

if no gateway of said subset is associated with a value representative of a quality of communication with the sending terminal greater than or equal to the predefined quality threshold, choosing a gateway associated with the value representative of a quality of communication with the maximum transmitter terminal among the values representative of a quality of communication with the transmitter terminal associated with the gateways of said subset.

13) Method according to claim 1 characterized in that, when the server receives a plurality of requests each containing the same message sent in an uplink by a transmitter terminal associated with no subset, the method comprises:

choosing from the first gateways having sent back said message to the server, a gateway whose reception quality of said frame is greater than or equal to a predefined quality threshold; and,

transmitting a request containing said response to the selected gateway, said response being retransmitted by the selected gateway to said transmitting terminal in a downlink. A server-type device (10) included in a long-range wireless network and enabling low power consumption in which terminals communicate with the server via gateways (11A, 11B) of a set. gateways according to a communication technology implying that when the server receives a message from a terminal via a first gateway, it fixes a time at which a gateway must relay a response to said message according to a time of receipt of said message by said first gateway, each gateway being associated with a set of terminals comprising at least one terminal using said gateway to communicate with the server, characterized in that the device comprises:

receiving means for receiving (303) a request from a first gateway (11A) containing a message sent in a frame, said rising frame, by a terminal (12), said transmitting terminal;

determining means for determining (304) a second gateway (HB) different from the first gateway to be used for relaying a response to said message intended for the transmitting terminal, the second gateway being determined in a subset of gateways of the set of gateways comprising the first gateway and at least one second gateway, each gateway of the subset being capable of relaying to a terminal a response to a message transmitted in an upstream frame by said terminal received by any other gateway of said subset the subsystem gateways having at least one associated terminal in common and being associated with the transmitting terminal, the determining means executing a selection procedure for determining the second gateway comprising: preselecting (700) a predefined number of gateways associated with the lowest counter values among counter values a associated with the gateways of said subset representative of a number of falling frames transmitted by the gateways of said subset in response to messages transmitted in uplink frames by terminals;

selecting (701, 702) a gateway associated with a value representative of a quality of communication with the sending terminal greater than or equal to a predefined quality threshold among the gateways of the predefined number of preselected gateways, the chosen gateway becoming the second gateway. ;

transmission means for transmitting (305) a request containing said response to the second gateway, said response being retransmitted (307) by the second gateway to the transmitting terminal in a so-called downgoing frame at a time indicated in the request, said instant being indicated in a time reference of the first gateway, the second gateway taking into account a clock deviation between the first and the second gateways for transmit the downlink to the instant.

A system included in a long-range, low-power wireless network comprising a server-type device according to claim 14, a plurality of terminals and a plurality of gateways, the plurality of gateways comprising subsets. gateways, each gateway of the subset being capable of relaying to a terminal a response to a message transmitted in a rising frame by said terminal received by any other gateway of said subset, the gateways of the subset having at least one associated terminal in common.

16) Computer program, characterized in that it comprises instructions for implementing, by a device (100), the method according to any one of claims 1 to 13, when said program is executed by a processor of said device (100).

17) Storage means, characterized in that they store a computer program comprising instructions for implementing, by a device (100), the method according to any one of claims 1 to 13, when said program is executed by a processor of said device (100).