OPTIMIZED METHOD FOR DATA COLLECTION FROM COUNTERS COMMUNICATE VIA A CELLULAR NETWORK, AND SYSTEM FOR IMPLEMENTING THE METHOD

Description

TECHNICAL FIELD

[0001] The present invention relates to a method for collecting data available in smart meters, performed by a data collection system and operated via a cellular network. At least one embodiment of the invention relates to a method for collecting data from electricity, water, or gas meters configured to communicate with a data concentrator via a mobile phone operator's IP network. STATE OF PRIOR ART

[0002] Modern electricity meters are electronic devices, often referred to as "smart" and / or "communicating," capable of generating and transmitting data to a remote server or system that performs data collection functions, such as a consumption management server. This data is transmitted via various communication networks, particularly for pricing and data collection services related to electricity, water, or gas consumption. Some of these meters are configured to transmit the data they generate via a cellular communication network, such as a mobile phone operator's network. Such a network consists of a set of communication cells, each covered by a transmitting and receiving station known as a base station.In other words, the area covered by the cellular communication network is divided into cells, each containing a base station. Each cell contains one or more communicating meters that generate information useful for managing one or more service or product deliveries. It is therefore necessary to collect data from these meters, taking into account the communication capacity of each cell in the communication network. Current techniques for collecting information from the meters are likely to cause congestion in the cellular communication network, especially since the communicating meters are sometimes in standby mode between two data processing or collection operations. This means that it is first necessary to control the meter to wake it up, and then to collect the necessary information.At other times, meters are temporarily out of network range due to disruptions. When data collection targeting a meter fails, it must be repeated until the data is collected, resulting in redundant use of network resources for the same information. Therefore, there is a need to optimize data collection methods to obtain large amounts of data from a set of meters, according to imposed scheduling constraints, while limiting the use of network resources that are already being used for other purposes. This situation can be improved.

[0003] US document 8,767,744 B2 already discloses a method for prioritizing the transmission of counting data. DESCRIPTION OF THE INVENTION

[0004] The invention aims to provide a method for collecting data from communicating meters, according to calendar constraints, while limiting the use of network resources. The invention is defined in the attached set of claims.

[0005] To this end, a method is proposed for collecting, via a cellular communication network, data available in a set of communicating meters, the method being executed in a data collection system connected to said network and the method comprising: i) obtain initial information representative of the behavior of said meters, ii) establish parameters representative of the behavior of each of the communicating meters from said initial information, iii) establish a data collection schedule for all or part of said meters from said parameters representative of the behavior of each of the meters, and, iv) transmit data collection messages to said meters according to said established schedule.

[0006] Advantageously, by learning the typical behavior of each communicating meter, it is possible to schedule data collection from a meter only when it has the highest probability of responding (and therefore responding on the first attempt). The cellular communication network is then shared efficiently between the data collection system and other applications using the network. Furthermore, this method allows for the detection of seasonal patterns in the behavior of communicating meters connected to the cellular communication network, enabling these patterns to be taken into account when scheduling data collection.

[0007] The method according to the invention may further include optional features considered individually or in combination: The said first information includes at least (according to the invention) ∘ a cell identifier of said network with reference to a communicating meter identifier, ∘ a state of obtaining a response from a meter to a message addressed to it with reference to a timestamp, ∘ a response time of a meter to a message addressed to it, where appropriate, with reference to a timestamp.

[0008] Thus, it is possible to determine behavioral parameters for each of the counters, such as an average response time and an average response rate for each counter, depending on the day and time, and in reference to a cell through which it is accessible. Establishing a data collection schedule based on the defined parameters is performed with reference, for each communicating counter, to a cell identifier from which a communicating counter is accessible. Establishing the parameters representative of the behavior of each counter includes a statistical analysis (according to an alternative of the invention). The statistical analysis is performed with reference to a maximum collection time. The statistical analysis is performed with reference to minimal use of the communication network. Establishing the parameters representative of the behavior of each counter includes learning using a classifier-type module. Learning using a classifier-type module is performed with reference to a maximum collection time. Learning using a classifier-type module is performed with reference to minimal use of the communication network.

[0009] The invention also relates to a system (or device) for collecting, via a cellular communication network, data available in a set of communicating meters, the collection system comprising electronic circuitry configured to: i) obtain initial information representative of the behavior of said meters, ii) establish parameters representative of the behavior of each of the communicating meters from said initial information, iii) establish a data collection schedule for all or part of said meters from said parameters representative of the behavior of each of the meters, and, iv) transmit data collection messages to said meters according to said established schedule.

[0010] The collection system according to the invention may further include the optional features considered individually or in combination: The data collection system further comprises circuitry configured to process said initial information, including (according to the invention): a cell identifier of said network with reference to a communicating meter identifier; a status indicating whether a meter has received a response to a message addressed to it with reference to a timestamp; and a response time of a meter to a message addressed to it, where applicable, with reference to a timestamp. The data collection system includes electronic circuitry configured to establish a data collection schedule based on parameters established with reference, for each communicating meter, to a cell identifier from which a communicating meter is accessible.The data collection system further includes electronic circuitry configured to establish parameters representative of the behavior of each counter by performing a statistical analysis (according to an alternative of the invention). The data collection system further includes electronic circuitry configured to perform the statistical analysis with reference to a maximum collection time. The data collection system further includes electronic circuitry configured to perform the statistical analysis with reference to minimal use of the communication network. The data collection system further includes electronic circuitry configured to establish parameters representative of the behavior of each counter by means of a classifier-type module.The data collection system further includes electronic circuitry configured to perform learning by means of a classifier-type module, with reference to a maximum collection time.

[0011] Another object of the invention is a computer program product comprising program code instructions to execute the steps of a process as previously described, when this program is executed by a processor of a data collection system.

[0012] Finally, the invention also relates to an information storage medium comprising a computer program product as mentioned above. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The features of the invention mentioned above, as well as others, will become clearer upon reading the following description of at least one exemplary embodiment, said description being made in relation to the accompanying drawings, among which: [ Fig. 1 ] schematically illustrates a communicating counter generating data to be collected, as already known; [ Fig. 2 ] is a schematic representation of a cellular-type communication network to which communicating meters are connected and which includes an improved system for collecting data generated by the meters, according to one embodiment; [ Fig. 3 ] is a flowchart illustrating an improved data collection process executed in the communication network already represented on the Fig. 2 , according to one embodiment; and, [ Fig. 4] is a diagram illustrating an internal architecture of a system or device for collecting data generated by communicating meters, according to one embodiment. DETAILED DESCRIPTION OF IMPLEMENTATION METHODS

[0014] There Fig. 1 This schematically illustrates a communicating meter 10, already known in the prior art and configured to collect information on the supply of a physical quantity such as water, electricity, or gas, for example. Such a communicating meter is sometimes still called a smart meter or " smartmeter(From English). The communicating meter 10 is configured to count a distributed physical quantity transmitted between an input 12 of the communicating meter 10, connected to a network supplying the physical quantity, and an output 13 of the communicating meter 10, connected to a domestic, industrial, or commercial installation, for example, that consumes the distributed physical quantity and whose consumption must be measured and / or controlled remotely. The communicating meter 10 includes a human-machine interface 14, also called a user interface, capable of inputting and displaying information related to the use of the meter. The communicating meter 10 also includes internal electronic circuitry comprising, in particular, one or more microcontrollers and a radio communication interface (not shown in the figure) connected to an antenna system 11.Thus, the communicating meter 10 is configured to communicate in a cellular type communication network.

[0015] There Fig. 2This schematically illustrates a cellular communication network 1000, comprising a communication subnetwork 1001 to which three base stations, STA-A, STA-B, and STA-C, are connected. These base stations are configured to transmit and receive signals to and from third-party devices compatible with and connected to the communication network. In this example, base station STA-A is connected to communication subnetwork 1001 via a wired communication link 1000a, base station STA-B via a wired communication link 1000b, and base station STA-C via a wired communication link 1000c. Of course, each of the base stations STA-A, STA-B, or STA-C could also be connected to communication subnetwork 1001 via a wireless link. Each of the STA-A, STA-B and STA-C base stations covers a geographical area called a cell.According to the example described, base station STA-A covers cell A, base station STA-B covers cell B and base station STA-C covers cell C. Communicating meters 10a, 10b, 10c, 10d, 10e, 10f, 10g, 10h and 10i, similar to the communicating meter 10 previously described in relation to the . Fig. 1These devices are installed within the coverage area of ​​the 1000 communication network, established by combining the STA-A, STA-B, and STA-C base stations, which operate radio transmissions in cells A, B, and C, respectively. In the example described, the number of cells in the 1000 communication network, as well as the number of communicating meters, is intentionally reduced to facilitate reading and understanding. In reality, the number of cells for a data collection system can be in the tens, hundreds, or thousands, and the same applies to the number of meters involved in data collection.Although a smart meter can be geographically located in several cells of the 1000 communication network, it is considered here that such a smart meter, connected to a fixed installation, is fixed and is therefore connected to only one cell of the 1000 communication network at any given time. According to one embodiment, a smart meter located in several cells is connected to the 1000 communication network via the cell whose base station offers it the best communication performance. Thus, according to the embodiment described in Figure 1... Fig. 2The communicating meters 10a, 10b and 10c are connected to the communication network 1000 via the base station STA-A operating in cell A, the communicating meters 10d, 10e and 10f are connected to the communication network 1000 via the base station STA-B operating in cell B and the communicating meters 10g, 10h and 10i are connected to the communication network 1000 via the base station STA-C operating in cell C. The communication network 1000 further includes a data collection system 100 configured to collect data generated by the communicating meters 10a, 10b, 10c, 10d, 10e, 10f, 10g, 10h and 10i. According to the example described, the data collection system 100 is connected to the communication subnetwork 1001 via a wired communication link 101. According to a variant, the data collection system 100 is connected to the communication subnetwork 1001 via a wireless link.The data collection system 100 is also referred to here as a data collection device or a service provider's management server. Advantageously, the data collection system 100 is configured to implement a data collection process aimed at optimizing data collection by reducing the use of resources on the communication network 1000, which, being cellular, is intended to be shared with other uses or applications, including common mobile telephony or telecommunications applications over a mobile network (audio and / or video transmission, for example). Thus, the data collection system 100 is configured to execute an improved collection process described in relation to the [reference to relevant section]. Fig. 3 .

[0016] There Fig. 3 illustrates steps in the improved data collection process performed in the 1000 communication network or a similar communication network.

[0017] One step S0This corresponds to an initialization step at the end of which the data collection system 100, as well as all systems of the communication network 1000, are normally configured to perform data transmissions across the communication network 1000. Thus, at the end of step S0, the data collection system 100 is able to send messages to the various communicating meters 10a, 10b, 10c, 10d, 10e, 10f, 10g, 10h, and 10i via, depending on the communicating meter concerned by a message transmission, the base station to which it is connected. The data collection system 100 is also configured to receive messages sent by the communicating meters in response to messages addressed to them, provided that the latter have the capacity to respond.Indeed, some smart meters may be temporarily disrupted by obstacles to electromagnetic waves or by poor operating conditions of the communication network, or even depending on potential standby cycles aimed at saving energy.

[0018] Cleverly, the improved data collection process executed by the 100 data collection system generated by the various communicating meters 10a, 10b, 10c, 10d, 10e, 10f, 10g, 10h and 10i operates four main stages S1, S2, S3 and S4, also referred to here respectively as "labeling phase", "learning and planning phase", "scheduling phase" and "data collection phase".

[0019] During step S1, and throughout a predefined period T (for example, a full week), the data collection system 100 sends protocol messages to the various communicating meters 10a, 10b, 10c, 10d, 10e, 10f, 10g, 10h, and 10i, and records a wealth of information about the behavior of the different meters. This involves determining whether a meter responds to a protocol message addressed to it and, if so, how long it takes. This information is recorded in correlation with an identifier for the communicating meter in question. During this step S1, information relating to the cell of the communication network 1000 through which a communicating meter responded or did not respond is obtained and stored, so that all the information can also be stored with reference to a cell identifier of the communication network 1000.The term "protocol message" here refers to any message in a predefined format that a meter is expected to be able to receive and interpret, so as to respond to it, and that the data collection system 100 also understands, or at the very least is capable of generating, and for which the data collection system 100 can identify a response or lack thereof. The format of the exchanged protocol messages is not described here, as it is not essential to understanding the invention. The messages exchanged between the data collection system 100 and the communicating meters may be exchanged as part of transmissions related to the provision of a service, or simply as part of the search for information representative of the operation of the communicating meters, or through a combination of these two methods.Thus, at the end of period T of census of representative information of the operation of the meters, the data collection system 100 has information relating to each of the communicating meters, due to the numerous attempts at communication made via the exchanges of messages described.

[0020] According to the invention, the data collection system 100 identifies and records at least some correlated information in the form of: a communication network cell identifier 1000 with reference to a communicating meter identifier, a status of obtaining a response from a meter to a message addressed to it with reference to a timestamp, for example "response" or "no response" at the current time, a response time of a meter to a message addressed to it, if applicable, and with reference to a timestamp, for example a given meter responded at such time, after a delay of 10 seconds.

[0021] In one embodiment, the data collection system 100 has information about the meters for which data collection is to be carried out, and therefore the meters to which the improved data collection process should be applied. Thus, the data collection system 100 has, for example, a list of identifiers of the relevant communicating meters and does not need to carry out any phase of discovering communicating meters present or potentially present in a geographical area covered by the communication network 1000.For example, the 100 data collection system can send one or more messages to each of the meters in a list and then use information about the communication protocol of the cellular-type 1000 communication network to identify through which base station a meter could have been reached and, in fact, in which cell of the communication network it is located.

[0022] According to an example implementation, the information collected and stored during step S1 is grouped by cells of the communication network 1000, that is to say with reference to a cell identifier of the communication network 1000.

[0023] For example, a set of information obtained during the so-called labeling phase, during step S1, can be organized as in the table below: [Table 1] Cell identifier (Ceid) Counter ID (Coid) Date (d) Time (h) Status of obtaining a response (r) Response time (s) (tr) A 10a 29 / 09 / 2023 00 :00 answer 0,1 A 10b 29 / 09 / 2023 00 :00 answer 0,2 A 10c 30 / 09 / 2023 00 :00 no response - B 10d 30 / 09 / 2023 00 :00 answer 0,1 B 10e 01 / 10 / 2023 00 :15 answer 0,2 B 10f 01 / 10 / 2023 00 :30 no response - C 10g 01 / 10 / 2023 00 :15 answer 0,3 C 10h 01 / 10 / 2023 00 :45 answer 0,2 C 10i 02 / 10 / 2023 00 :45 no response -

[0024] Thus, cell ID, counter ID, timestamp, response status and response time information are correlated with each other in a memory of the data collection system 100.

[0025] During the stage S2,The data collection system 100 establishes operating parameters specific to each of the communicating meters involved in the data collection. In other words, the data collection system 100 establishes a behavioral profile for each communicating meter, based on the information recorded during step S1 over a period of time T and stored. This learning phase is carried out by meter and by time slots, for example, by meter and by hour, by meter and by quarter-hour, or by meter and by five-minute intervals; these examples are not exhaustive. According to a first embodiment, the data collection system 100 performs this learning phase by conducting a statistical analysis to determine when each communicating meter has the highest probability of responding to a message addressed to it, in particular a data collection message.According to a second embodiment, the data collection system 100 operates this learning phase by operating a type of learning. "machine learning" capable of providing a probability of successful communication for each of the meters involved in an upcoming collection.

[0026] Each of these embodiments can be implemented according to variations whose respective objectives are: to respect a time constraint for the end of collection, such as for example aiming to have collected the data from all the meters before the next day at eight o'clock in the morning, or to significantly reduce the use of the resources of the communication network used (here the communication network 1000) in order to avoid any redundancy of sending messages to the meter until the data to be collected is obtained.

[0027] According to one embodiment, when the analysis is of a statistical type and when the objective is to respect a time constraint, the data collection system 100 determines, for each of the counters, a time slot for which this counter has the fastest collection time while having a minimum threshold of 50% in response rate to messages addressed to it.

[0028] According to one embodiment, when the analysis is of a statistical type and when the objective is to reduce the use of the resources of the communication network used, the data collection system 100 determines a time slot for which a counter has a maximum rate of response to messages addressed to it.

[0029] According to one embodiment, when the learning phase performs a learning of the type " machine learning »The objective of this algorithm is to determine the probability of successful communication based on the circumstances of sending a message to a communicating counter (for example, the day of the week, date, time, whether a day is a holiday, etc.). A decision tree-type algorithm is used. In one embodiment, the algorithm used is a classifier for each time slot, classifying it into two categories, one associated with a communicating counter and the other with a non-communicating counter. Thus, each time slot has a probability of belonging to a category for a given counter. In one embodiment, if a time slot has a probability of belonging to a category greater than 0.6, then that time slot belongs to that category.The aim of such an algorithm is to maximize the success rate of a data collection by respecting a collection deadline or to minimize the number of communications required to collect data available in a given communicating meter.

[0030] According to one embodiment, the training of the decision tree-type classifier is based on two sets of information, the first of which includes all the first information previously collected in step S1, time-stamped (recorded in association with the current date and time), enriched with second pieces of information such as the day of the week, the month of the year, whether the day concerned is a holiday or not, the urban or rural nature of the meter's position identified by the network cell through which it communicates, and the second of which includes the class to which each piece of information belongs (for example, a class 0 according to which the meter communicates and a class 1 according to which the meter does not communicate).

[0031] Cleverly, the algorithm thus trained is used during step S2, if necessary (a variant of a "machine learning" type learning mode) to determine the best time slot for each of the communicating meters concerned by a planned and future collection.

[0032] At the end of the learning process carried out in step S2, the data collection system 100 has the ability to determine, in step S3, a collection schedule based on predefined constraints, for example based on calendar constraints of a data collection (which day, which duration, etc.).

[0033] According to one embodiment, the collection scheduling is carried out by proceeding cell by cell and with the aim of subsequently operating a collection in parallel (simultaneously) in each of the relevant cells of the communication network.

[0034] According to one embodiment, and depending on collection organization constraints, such as scheduling constraints (for example, during the night from Tuesday to Wednesday, between 9 p.m. and 9 a.m., in urban areas), the scheduling is established so as to collect data from all meters identified as being able to respond at the scheduled collection time and, moreover, as being unable to respond in the periods following collection. The collection then covers the maximum number of meters in each cell meeting this criterion, using the maximum amount of communication network resources allocated (or dedicated) to the collection for the duration of the collection.

[0035] According to one embodiment, data collection tables are generated so as to establish a collection order (a sequence), with reference to each of the cells A, B and C, of ​​the communication network 1000.

[0036] When a collection schedule has been established by the data collection system 100 by surveying information on the observed behavior of meters concerned by the collection subject to the scheduling, then by determining operating parameters (profile) for each of the meters, and taking into account a defined objective in terms of calendar constraints and / or level of network resources used, the data collection system 100 can execute the data collection during the S4 step, proceeding according to the established sequence and operating in parallel (simultaneously) in the different cells A, B and C of the communication network.

[0037] According to one embodiment, and when data collection is completed at the end of step S4, i.e., when the allotted time for data collection has expired, an optional step S5is executed by the data collection system 100, aimed at determining the rate of communicating meters that could be successfully read, that is to say the rate of communicating meters that were able to send the data to be collected in response to a collection message from the data collection system 100.

[0038] If the collection rate determined in step S5 is higher than a predefined collection target (step S5, output "yes"), then the process is completed and no further collection operations will be carried out until a new collection phase is scheduled according to the defined collection policy. Conversely, if the collection rate determined in step S5 is insufficient (step S5, output "no"), then the process loops back to step S1, modifying, for example, the parameter T representing the duration of the labeling phase, which aims to obtain new information representative of the behavior of the communicating meters present in the 1000 communication network.Each new iteration of step S1, performed when the data collection completeness rate is deemed insufficient, is considered a learning reinforcement phase aimed at identifying the communicating meters that contributed to the failure to achieve the target data collection rate. In one embodiment, a new iteration of step S1 is performed with a data collection period for information representative of the meters' behavior that is longer than initially defined (for example, longer than seven days).

[0039] According to one embodiment, and considering that some meters may be identified as unresponsive, for example due to interference in electromagnetic communications, a communication strategy with these meters can be advantageously established. This strategy might involve not systematically repeating a communication attempt for every elementary time period considered by the data collection system 100, but rather attempting to collect data only from a predefined number of communicating meters to be read during an elementary time period. In other words, if the system repeats communication attempts with "difficult-to-reach" meters every five minutes, it will only attempt to reach 70% or 60% of the remaining meters, for example, per five-minute period, to avoid overloading the communication network 1000.Of course, the percentage of counters that are the subject of a new attempt can be any number of times, between 0 and 100%, depending on the results previously obtained.

[0040] According to one embodiment, when a new smart meter is inserted into the communication network 1000, the data collection system 100 plans to implement a data collection schedule identical in behavior to that planned for a smart meter geographically close to a smart meter already known to the collection system 100. If such a strategy does not allow data to be collected from this meter without difficulty, then a learning phase must take place according to the described procedure.

[0041] It should be noted that on the Fig. 3 The optional nature of step S5 is illustrated by dotted lines.

[0042] There Fig. 4schematically illustrates an example of the internal architecture of the data collection system 100, also referred to here as the data collection device or management server of a service provider(s). According to the hardware architecture example shown in the Fig. 4, the collection system 100 then comprises, connected by a communication bus 120: a processor or CPU (“Central Processing Unit”) 121; a RAM (“Random Access Memory”) 122; a ROM (“Read Only Memory”) 123; a storage unit such as a hard disk drive (or a storage media reader, such as an SD card reader (“Secure Digital”) 124; at least one communication interface 125 enabling the data collection system 100 to communicate with devices present in the communication network comprising the data collection system 100 including in particular the communicating meters 10a, 10b, 10c, 10d, 10e, 10f and 10h, via the base stations STA-A, STA-B and STA-C.

[0043] The processor 121 is capable of executing instructions loaded into RAM 122 from ROM 123, external memory (not shown), storage media (such as an SD card), or a communication network. When the data collection system 100 is powered on, the processor 121 is able to read instructions from RAM 122 and execute them. These instructions form a computer program causing the processor 121 to implement the process described in relation to the Fig. 3 or one of its variants.

[0044] All or part of the process implemented by the data collection system 100, or its described variants, can be implemented in software form by executing a set of instructions by a programmable machine, for example a DSP (Digital Signal Processor) or a microcontroller, or implemented in hardware form by a dedicated machine or component, for example an FPGA (Field-Programmable Gate Array) or an ASIC (Application-Specific Integrated Circuit). In general, the data collection system 100 comprises electronic circuitry configured to implement the described process in relation to itself and to devices connected to the communication network 1000.Of course, the 100 data collection system also includes all the elements usually present in a system comprising a control unit and its peripherals, such as a power supply circuit, a power supply monitoring circuit, one or more clock circuits, a reset circuit, input / output ports, interrupt inputs, bus drivers, this list being non-exhaustive.

[0045] The invention is not limited to the embodiments and examples described but relates more broadly to any collection method, operating via a cellular communication network, adapted to the collection of data available in a set of communicating meters and executed in a data collection device connected to this network, and comprising: obtaining initial information representative of the behavior of the meters, then establishing representative parameters (or profiles) of the behavior of each of the communicating meters from this information, in order to establish a data collection schedule of all or part of the meters from the parameters (or profiles) representative of the behavior of each of the meters, and then transmitting data collection messages to the communicating meters according to the schedule which has been established.In particular, a classifier of a different type than a decision tree can be used for the learning phase carried out during step S2.

Claims

1. Method for collecting, via a communication network (1000) of the cellular type, data available in a set of smart meters (10a, 10b, 10c, 10d, 10e, 10f, 10g, 10h, 10i), the method being implemented in a data collection system (100) connected to said network (1000), and the method comprising: - i) obtaining first information (S1) representing a communication-performance behaviour of said meters (10a, 10b, 10c, 10d, 10e, 10f, 10g, 10h, 10i), said first information comprising at least one response time (tr) and a state of obtaining a response for each meter with reference to a timestamp (d, h) and a cell identifier (Ceid), - ii) establishing parameters (S2) representing a communication-performance behaviour of each of the smart meters (10a, 10b, 10c, 10d, 10e, 10f, 10g, 10h, 10i) by statistical analysis or machine learning based on communication data observed, - iii) establishing a scheduling of collection (S3) of data from all or some of said meters (10a, 10b, 10c, 10d, 10e, 10f, 10g, 10h, 10i) from said parameters representing the behaviour of each of the meters, wherein said scheduling is established to optimise the data collection on the basis of the communication-performance behaviour learned from the meters, and - iv) transmitting data collection messages (S4) to said meters (10a, 10b, 10c, 10d, 10e, 10f, 10g, 10h, 10i) in accordance with said scheduling established.

2. Data collection method according to claim 1, wherein establishing said data-collection scheduling (S3) from said parameters is implemented with reference, for each of the smart meters, to an identifier (Ceid) of a cell from which a smart meter is accessible.

3. Data collection method according to one of claims 1 to 2, wherein the statistical analysis is implemented with reference to a maximum collection time.

4. Data collection method according to one of claims 1 to 2, wherein the statistical analysis is implemented with reference to a minimum use of said communication network.

5. Data collection method according to one of claims 1 to 2, wherein establishing said parameters representing the behaviour of each of the meters comprises a learning by means of a module of the classifier type.

6. Data collection method according to claim 5, wherein the learning by means of a module of the classifier type is implemented with reference to a maximum collection time.

7. Data collection method according to claim 5, wherein the learning by means of a module of the classifier type is implemented with reference to a minimum use of said communication network.

8. System (100) for collecting, via a communication network (1000) of the cellular type, data available in a set of smart meters, the collection system comprising electronic circuitry configured to: i) obtain first information (S1) representing a communication-performance behaviour of said meters, said first information comprising at least one response time and a state of obtaining a response for each meter with reference to a timestamp and a cell identifier, ii) establish parameters representing a communication-performance behaviour (S2) of each of the smart meters by statistical analysis or machine learning based on communication data observed, iii) establish a scheduling of collection (S3) of data from all or some of said meters from said parameters representing the behaviour of each of the meters, wherein said scheduling is established to optimise the data collection on the basis of the communication-performance behaviour learned from the meters, and iv) transmit data collection messages (S4) to said meters in accordance with said scheduling established.

9. Data collection system according to claim 8, comprising electronic circuitry configured to establish said data-collection scheduling from said parameters with reference, for each of the smart meters, to an identifier of a cell from which a smart meter is accessible.

10. Data collection system according to one of claims 8 to 9, furthermore comprising electronic circuitry configured to make the statistical analysis with reference to a maximum collection time.

11. Data collection system according to one of claims 8 to 9, furthermore comprising electronic circuitry configured to make the statistical analysis with reference to a minimum use of said communication network.

12. Data collection system according to one of claims 8 to 9, furthermore comprising electronic circuitry configured to establish said parameters representing the behaviour of each of the meters by learning by means of a module of the classifier type.

13. Data collection system according to claim 12, furthermore comprising electronic circuitry configured to implement said learning by means of a module of the classifier type, with reference to a maximum collection time.

14. Data collection system according to claim 12, furthermore comprising electronic circuitry configured to implement said learning by means of a module of the classifier type, with reference to a minimum use of said communication network.

15. Computer program product characterised in that it comprises program code instructions for performing the steps of the method according to any one of claims 1 to 7, when said program is executed by a processor of said data collection system.

16. Information storage medium comprising a computer program product according to claim 15.