FR3038172A1 - METHOD FOR DETERMINING THE TIME DIFFERENCE EXISTING BETWEEN THE CLOCKS OF THE BASE STATIONS OF A COMMUNICATION NETWORK - Google Patents

Abstract

The present invention relates to a method for determining the time difference existing between the clocks of the base stations (31) of a communication network. The method (50) comprises steps in which: - 51 / a rising message is sent by a remote terminal (20) to a central server (32), - 52 / the amount message is received by at least two stations base (31) of the communication network, - 53 / each of the base stations (31) having received the message amount stamps the latter according to its own clock, then transmits the message thus timestamped to the central server (32), - 54 / the central server (32) collects all the messages stamped by the base stations (31), extracts the time stamped messages comprising the same amount message from the first terminal (20A), and determines the time offsets between these time stamped messages.

Description

TECHNICAL FIELD The present invention belongs to the field of digital telecommunications, and more particularly relates to a method for determining the time difference existing between the clocks of the base stations of a communication network, said communications being carried out between at least one central server and remote terminals via at least one base station. STATE OF THE ART The present invention finds a particularly advantageous, though in no way limiting, application in ultra-narrowband wireless communication systems. By "ultra narrow band" ("Ultra Narrow Band" or UNB in the Anglo-Saxon literature), it is meant that the instantaneous frequency spectrum of the radio signals emitted by the terminals is of frequency width less than one kilohertz. Such UNB wireless communication systems are particularly suitable for applications of the type M2M (acronym for machine-to-machine) or the Internet of Things ("Internet of Things" or loT in the literature Anglo-Saxon).

In such a UNB wireless communication system, the data exchanges are essentially monodirectional, in this case on a rising link between terminals and an access network of said system. The terminals transmit uplink messages that are collected by base stations of the access network, without having to first associate with one or more base stations of the access network. In other words, the upstream messages sent by a terminal are not intended for a specific base station of the access network, and the terminal transmits its upstream messages assuming that they can be received by at least one forwarding station. based. Such provisions are advantageous in that the terminal does not need to make regular measurements, particularly greedy in terms of power consumption, to determine the most appropriate base station to receive its upstream messages. The complexity lies in the access network, which must be able to receive uplink messages that can be transmitted at arbitrary times and on arbitrary central frequencies. Each base station of the access network receives messages from the various terminals that are within its reach. Such a mode of operation, in which the data exchanges 5 are essentially monodirectional, is quite satisfactory for many applications, such as, for example, remote reading of gas meters, water meters, electricity meters, telemonitoring. buildings or houses, etc. Such a network is described in particular in the patent application EP 2,580,882 A1 ("METHOD OF USING A SHARED FREQUENCY RESOURCE 10, METHOD FOR MANUFACTURING TERMINALS, TERMINALS AND TELECOMMUNICATIONS SYSTEM") of the same applicant. It is a low bandwidth cellular network that aims to have a very wide geographical coverage, including in areas not covered by conventional communication networks whether they are wired or wireless (GSM, 3G, LTE , etc.).

This wireless network is characterized by an unorganized communication mode in which the remote terminals transmit messages on their initiative and where these messages are picked up by all the base stations which are within the reach of the terminal and where the base stations are not intended to filter messages received by multiple base stations. On the contrary, they in turn transmit any message picked up to a central server. The back-haul (base network formed by the base stations) of the messages from the base stations to the servers, uses a variety of communication channels according to their availability: wired connection to the Internet, radio link, cellular wireless connection (GSM, 3G, LTE, ...), satellite link 25 (low orbit or geostationary orbit) ... In certain back-haul or server congestion situations, the base stations may have to buffer the received messages to differ during transmission to the server. During the implementation of such a network, there are problems related to the desynchronization of the basic clocks of the various base stations of the network. These desynchronizations have their origin in many factors (delay, jitter ...). The order of magnitude of the time differences observed between various clocks can reach several seconds.

Such an offset is detrimental to the proper functioning of the communication network. Indeed, it may be necessary for some users of such a network to have accurate time information on the date of sending messages by the terminals, for example when it comes to messages of events such as than device failures to which they are connected. Synchronization of the base station clocks, allowing dating accuracy better than the second, is therefore generally desirable. Network Time Protocol (NTP) is already known in this field, which is the protocol widely used to synchronize the clocks of network equipment. Furthermore, to solve the synchronization problem of network equipment clocks, synchronization of the clocks by a GPS signal (Global Positioning System).

15 Due to the nature of the backhaul connectivity that may differ from one station to another and may change from one moment to another (eg loss of cellular network, disconnection of a link Internet, significant fluctuation of transmission delays ...) the use of the NTP protocol does not give reliable results.

Indeed, NTP is very efficient for Local Area Network (LAN) networks and makes it possible to reach clock accuracies of the order of 100 microseconds, however for high-bandwidth Wide Area Network (NTP) networks, NTP. begins to show limits (the accuracy can then fall to 100 ms). In the case of WAN networks with low flow rate, very irregular flow rate, very variable latency and especially non-symmetrical latency, NTP does not allow to obtain reliable clock synchronizations. In some cases, desynchronizations have been observed in the order of 10 seconds. NTP relies on a transmission latency between the nodes of the network which is essentially symmetrical, which is difficult to achieve when this latency is not constant and variable in time. GPS synchronization would require the equipping of all base stations with a GPS receiver, which implies a material overhead but also a complexity of deployment on the sites of the base stations (antenna 3038172 4 GPS on pylons, wiring and amplifier between l GPS antenna and equipment). The terminals on such a network must remain simple and therefore have no clock allowing them to time messages. In addition, the low rates limit the information transmitted and it is therefore unreasonable to transmit in each message a time stamp. However, this information can be important for the client, as it may allow him to date an event (for example). However, the base stations of the network are chosen voluntarily as inexpensive systems. It is therefore not possible to equip them with GPS receivers, or other sophisticated synchronization devices. PRESENTATION OF THE INVENTION The object of the present invention is to remedy all or some of the limitations of the solutions of the prior art, in particular those set out above, by proposing a solution that makes it possible to correct the time differences between the messages received. by the server, and / or synchronize the clocks of base stations at low cost.

For this purpose, and according to a first aspect, the invention relates to a method for determining the time difference existing between the clocks of the base stations of a communication network, said communications being made between at least one terminal and at least one terminal. central server via at least one base station, each base station being adapted to receive the upstream messages of the terminals within range, each base station being equipped with a clock of its own, each message amount thus received being transmitted by the base station to the central server. The method of determining the time offset includes steps in which: a rising message is transmitted by a first remote terminal to the central server, the upstream message is received by at least two base stations of the communication network, each base stations having received the message amount of the first terminal timestamp it according to its own clock, then transmits the message and stamped to the central server, 5 the central server collects the messages stamped by the base stations, extracts the messages stamped having the same amount message from the first terminal (20A), and determines the time offsets between these time stamped messages. By "base station" is meant here communication nodes of the network adapted to retransmit received received messages to a server or other communication node of the network. It will be understood that the invention is a method for determining the time shift existing between the clocks of the base stations, the method not being sensitive to the message transmission time between the base stations and the central server. This characteristic makes it possible to overcome the variability of back-haul means between stations and over time. It also makes it possible to overcome the variable latencies introduced in this backhaul by a temporary loss of connectivity or by a congestion situation in the network or in the central server.

The method is based on the fact that a message sent by a remote terminal is generally received by several base stations; the base stations time stamp the received messages and transmit the time stamped messages to a central server configured to identify the identical messages received by different base stations and to deduce an offset of the timestamp clocks of the base stations having received the same message . In particular modes of implementation, the method for determining the time difference existing between the clocks of the base stations may further comprise one or more of the following characteristics, taken individually or in any technically possible combination. The central server determines one of the timestamps as a reference value, and transmits a synchronization request to the base stations on the basis of the reference value. the central server maintains a database of time offsets between the clocks of the base stations, this database being used to correct the "timestamp" of each of the upstream messages transmitted by each of the base stations as and when as they are received. The database of time offsets is completed for the base stations of a second set of base stations within reception range of a second terminal, said set comprising at least one common base station which is also part of a first set of base stations within reception range of the first terminal, by adding the time offset value found for the clock of a base station common to the two sets, to the offsets of the clocks of the stations of the set by report to this station taken as a reference. each base station that has received a resynchronization request corrects its clock by a value equal to the time difference noted between the time of its time stamp and the time of the time stamp of the clock that served as a reference value. in step 53, all the received messages received from all the terminals are time stamped by the base stations receiving them, before retransmission to the server. In step 53, the timestamp of the received uplink messages is only made at recurring intervals, for a predetermined period of time - step 54 of calculating the time offset of the base station clocks is by the server only from time to time, at a predetermined interval calculated according to an estimate of the maximum drift of the least reliable clocks. in step 55, the central server chooses the clock of one of the base stations as a common reference from its NTP synchronization status, in step 56, each base station having received this resynchronization request corrects its clock by a value equal to the time difference found between the time of its time stamp and the time of the time stamp of the clock that served as a reference value.

In particular modes of implementation, the method for determining the time difference existing between the clocks of the base stations is such that, in a seventh step 57, the base stations within reception range of a second terminal are synchronized with respect to a base station and an element common to both sets of base stations, within reception range of the first terminal and the second terminal, respectively, the base station being selected as a reference clock for the set of stations base station within range of the second terminal, this step 57 being repeated as long as there is at least one set of base stations receiving range of the same remote terminal, not yet resynchronized, the set comprising at least one base station already resynchronized in the same resynchronization operation, the identifiers of the base stations having been resynchronized being stored in memory by the server.

According to a second aspect, the present invention relates to a base station comprising means configured to implement a method according to any one of the embodiments of the invention. According to a third aspect, the present invention relates to an access network comprising means configured to implement a method 20 according to any of the embodiments of the invention. According to yet another aspect, the present invention relates to a central server comprising means configured to implement a method for determining the time difference existing between the clocks of the base stations according to any one of the implementation modes of the invention. 'invention. According to another aspect, the invention relates to a wireless communication system, of the UNB type, comprising a plurality of terminals, an access network comprising a plurality of base stations, a server, comprising means configured to implement a method such as forth.

PRESENTATION OF THE FIGURES The invention will be better understood on reading the following description, given by way of non-limiting example, and with reference to FIGS. 3038172 which represent: FIG. 1: a schematic representation of a system FIG. 2 is a diagram illustrating the main steps of a method for determining the time difference between base station clocks; FIG. 3: a schematic representation of the case where two sets of base stations receiving messages from two remote terminals have a non-empty intersection.

In these figures, identical references from one figure to another designate identical or similar elements. For the sake of clarity, the elements shown are not to scale unless otherwise stated. DETAILED DESCRIPTION OF EMBODIMENTS The invention finds its place in the context of a wireless communication system 10, for example of the UNB ("Ultra Narrow Band") type, comprising several terminals 20 and an access network 30. comprising a plurality of base stations 31, as shown diagrammatically in FIG. 1. In a normal operating mode of the system, the latter comprises a large number of terminals 20 and base stations 31. For the implementation of the method described here only one terminal 20 and two base stations 31 are sufficient. The terminals 20 and the base stations 31 of the access network 30 exchange data in the form of radio signals. By "radio signal" is meant an electromagnetic wave propagating via non-wired means, the frequencies of which are included in the traditional spectrum of radio waves (a few hertz to several hundred gigahertz). The terminals 20 are adapted to transmit uplink messages on a uplink link to the access network 30. The uplink messages are for example transmitted asynchronously. By "asynchronously transmitting" is meant that the terminals 20 autonomously determine when they transmit, without coordination of said terminals 20 between them and with the base stations 31 of the access network 30. Each base station 31 is adapted to receive the upstream messages of the terminals 20 which are within range. Each uplink message thus received is for example transmitted by the base station 31 to at least one server 32 of the access network 30, possibly accompanied by other information such as an identifier of the base station 31 which received it, the measured power of said received amount message, the date of receipt of said amount message, etc. Each base station 31 is here assumed to have means for dating a received message, with respect to a local clock integrated in the base station. The server 32 processes, for example, all the amounts received from the various base stations 31. The following description is considered the case of a single central server 32. The invention is naturally adaptable to the case of more central servers 32.

Each base station 31 further comprises wireless communication means, known to those skilled in the art, enabling said base station to receive upstream messages and to transmit downstream messages. The base stations 31 and the server 32 also comprise respective network communication means, considered as known to those skilled in the art, enabling the server 32 to exchange data with each base station 31. base 31 and the server 32 comprise respective processing modules (not shown in the figures), each processing module comprising for example one or more processors and storage means 25 (magnetic hard disk, electronic memory, optical disk, etc. .) in which is stored a computer program product, in the form of a set of program code instructions to be executed to implement the different steps of a method of determining time shift between clocks of stations of base, and synchronization 30 of these clocks. In a variant, each processing module comprises one or more programmable logic circuits of the FPGA, PLD, etc. type, and / or specialized integrated circuits (ASIC) adapted to implement all or part of said process steps.

In other words, the access network 30 comprises a set of means configured in software (specific computer program product) and / or hardware (FPGA, PLD, ASIC, etc.) to implement the different stages of the process.

Mode of operation of the method As seen in FIG. 2, the method for determining the time offset existing between the clocks of the base stations 31 of the network comprises several steps.

Consider a remote terminal 20 (an object connected in the case of the network here considered by way of non-limiting example) within range of several base stations 31. A message sent by this terminal 20 will be received by all these base stations. 31 essentially at the same time. There may be a slight reception time offset related to a difference in flight time (distance between the terminal 20 and different base stations), but it can reasonably be ignored because of its extremely low value with respect to the time of flight. desired timing accuracy (about 0.1 seconds). In a first step 51, a rising message is sent by this remote terminal 20 to a central server 32. This message can be any, it does not have any particular format required. Any message sent by one of the terminals 20, in accordance with the format of the considered network, and received by a plurality of base stations 31 can be used for implementing the method for determining the time difference existing between the clocks of the base stations.

In a second step 52, the uplink message is received by at least two base stations 31 of the communication network 30. Each of these base stations 31 carries out, to receive the uplink message, tasks whose details are supposed to be known in and as such come out of the scope of the invention.

In a third step 53, each of the base stations 31 having received the message amounts it timestamps it according to its own clock, then transmits the message thus timed to the central server 32. Here again, the steps and details of the method of issuing a rising message between a base station 3038172 11 31 and the server 32 are assumed to be known per se, and are therefore not detailed further here. In the embodiment of the method described here, all the received messages received from the terminals 20 are time stamped before retransmission to the server 32.

In another mode of implementation, this time stamp of the received amounts messages is only made at recurring intervals, for example every few tens of minutes and for a predetermined period of time. This second option makes it possible to reduce the length of the transmitted messages, which is favorable in the case of a low-speed network.

In a fourth step 54, the central server 32 collects and stores in memory all the messages stamped by the base stations 31. The server 32 then identifies the received messages concerning the same initial upstream message, and determines the time offsets At between the Timestamp data attached to these timestamped messages. The difference of "timestamp 15" between two messages received from two stations gives the time difference between the clocks of the two stations. Here again, this operation of calculating the time shift of the clocks of the base stations 31 is carried out permanently in the embodiment described here by way of example, but could, alternatively, be carried out only from time to time. , at a predetermined interval calculated according to an estimate of the maximum drift of the least reliable clocks. In a particular embodiment, the server 20 stores in memory a history of time shifts At observed between the clocks 25 of the different base stations 31, with a view to a possible subsequent statistical analysis, or to a probable failure detection. 'a clock. In a fifth step 55, the central server 32 selects the clock of one of the base stations 31 as a common reference (from its NTP synchronization status, for example) for the calculation of this time offset At. embodiment described here by way of example, one chooses a clock among those which have a good accuracy of synchronization. Synchronization accuracy is given by NTP. One can also choose the base station 31 which has the best round-trip-time between the server 32 3038172 12 and the base station 31. In an exemplary implementation, the central server 32 asks each of the base stations 31 having issued the same message to shift its clock to compensate for its time offset from this reference clock. In a sixth step 56, each base station 31 having received this resynchronization request corrects its clock by a value equal to the time offset At found between the time of its timestamp and the time of the time stamp of the clock having served as a reference value. Alternatively, these fifth and sixth steps are avoided by maintaining, at the server 32, a database of time offsets At between the clocks of the base stations 31. This database is used to correct the "timestamp" of each of the amount messages transmitted by each of the base stations 31 as and when they are received.

In this way, it is possible to resynchronize all the clocks of the base stations 31 which are within the reach of the same remote terminal 20, or alternatively to have a database of time offsets between base stations 31 at reception range. of the same terminal 20, this database for correcting the timestamps attached to these messages.

More generally, this principle is applied to synchronize the entire network. Consider a first remote terminal 20A and a second remote terminal 20B in range, respectively of a first set 310A, and a second set 310B of base stations 31. The two sets 310A, 310B of base station are assumed to have a Non-empty intersection comprising at least one base station 31C (see Figure 3). It is possible to synchronize the base stations 31 within reach of the first terminal 20A (the first set 310A on the left in FIG. 3) by the method described above, or to create a database of time delays At observed between the clocks of the base stations of this first set 310A. In a seventh step 57, the base stations 31 with a reception range of the second terminal 20B (the one on the right in FIG. 3) can then be synchronized with respect to a base station 31C which is an element of the two sets 310A. , 310B of base stations 31, respectively receiving the terminal 20A and the terminal 20B, the base station 31C being chosen as the reference clock for the assembly 310B. By transitivity, all the base stations 31 of the two sets 310A, 310B are thus synchronized. This step 57 is repeated as long as there is at least one set 310N of base stations 31 within reception range of the same remote terminal 20N, not yet resynchronized, the set 310N comprising at least one base station 31 already resynchronized in the same resynchronization operation. Alternatively, if a database of time offsets between the clocks of the base stations 31 of the first set 310A is maintained by the central server 32, this base is completed for the base stations of the second set 310B of base stations 31 within range. receiving the second terminal 20B, adding the time offset value At (31C) found for the clock of a base station 31C common to the two sets 310A, 310B, to the clock offsets of the stations of the set 310B compared to this station 31C taken as a reference. It is therefore apparent that, step by step, one can synchronize all the base stations of the global network, or at least all the base stations within range of the same set of terminals. This is a viral propagation synchronization of a reference value based on time shift data between clocks of the base stations. In the present exemplary embodiment, the central server 32 keeps in memory the identity of the base stations 31 which have been resynchronized, the instant of their resynchronization, and the shift of clock time which they had to carry out. Of the various steps illustrated in FIG. 2, only the steps 53 for time stamping of the upstream messages and 56 for resynchronization must necessarily be executed at least partially by a base station 31. Here, in a nonlimiting manner, the case where the data collection step 54 and the reference clock determining step 55 are executed by the server 32, which then transmits the time offset data to be corrected to each base station 31, which then executes the step 56 of resynchronization.

Advantages of the method It is understood from reading the description above that the method for determining the time difference existing between the clocks of the base stations, as explained here, makes it possible, in particular, to resynchronize step by step, to low cost, all the base stations of a communication network whose connection to the heart of the network is unreliable and of poor quality. Moreover, this solution eliminates the need for terminals capable of time stamping all transmitted messages, which would be expensive in terms of equipment, and would increase the need for data rate, and therefore 15 bandwidth . This invention is also particularly useful in the case of a network requiring synchronized clocks, or at least whose relative time shifts are known, for operation in bidirectional mode. 20

Claims (11)

CLAIMS1 - Method (50) for determining the time shift existing between the clocks of the base stations (31) of a communication network, said communications being made between at least a first terminal (20A) and at least one central server (32). ) via at least one base station (31), each base station (31) being adapted to receive terminal messages (20, 20A) within reach, each base station being equipped with a clean clock, each received amount message being transmitted by the base station (31) to the central server (32), characterized in that said method (50) comprises steps in which: 51 / a rising message is transmitted by the first terminal (20A) to the central server (32), 52 / the upstream message is received by at least two base stations (31) of the communication network, 53 / each of the base stations (31) having received the message rising from the first er terminal (20A) timestamp it according to its own clock, then transmits the message and stamped to the central server (32), 54 / the central server (32) collects the messages stamped by the base stations (31), extracted the time stamped messages comprising the same amount message from the first terminal (20A), and determines the time offsets between these time stamped messages. 2 - Method (50) according to claim 1, further comprising a step 55 / wherein the central server (32) determines one of the timestamps as a reference value, and transmits to a base station (31) a synchronization request time of the base station clocks on the basis of the reference value. 3 - Method (50) according to claim 2, wherein, in step 55, the central server (32) chooses the clock of one of the base stations (31) as a common reference from its status of NTP synchronization. 4 - Method (50) according to claim 2 or 3, further comprising a step / 56 / in which each base station (31) having received this resynchronization request corrects its clock by a value equal to the time difference found between the time of its timestamp and the time of the timestamp of the clock that served as the reference value. 5 5 - Method (50) according to claim 4, wherein, in a seventh step 57, the base stations (31) receiving range of a second terminal (20B) are synchronized with respect to a base station (31C ) which is a common element to the two sets (310A, 310B) of base stations (31), receiving range respectively of the first terminal (20A) and the second terminal (20B), the base station (31C) being selected as a reference clock for the set (310B) of base stations within reach of the second terminal (20B), this step 57 being repeated as long as there is at least one set (310N) of base stations (31) within range of receiving the same remote terminal (20N), not yet resynchronized, the set (310N) comprising at least one base station (31) already resynchronized in the same resynchronization operation, the base station identifiers having been resynchronized being memorized by the serve ur (32). A method (50) according to any one of claims 1 to 5, wherein the central server (32) maintains a database of time offsets between the clocks of the base stations (31), this database being used to correct the "timestamp" of each of the upstream messages transmitted by each of the base stations (31) as and when they are received. 25 The method (50) of claim 6, wherein the time offset database is completed for the base stations of a second set (310B) of base stations (31) within reception range of a second terminal (20B), said set comprising at least one common base station (31C) also forming part of a first set (310A) of base stations (31) within reception range of the first terminal (20A), by adding the time offset value At (31C) found for the clock of a base station (31C) common to the two sets 3038172 17 (310A, 310B), to the offsets of the clocks of the stations of the set (310B) relative to at this station (31C) taken as a reference. 8 - base station (31) characterized in that it comprises means configured to implement a method (50) according to one of claims 1 to 7. 9 - access network (30) characterized in that it comprises means configured to implement a method (50) according to one of claims 1 to 7. 10 - Server (32) characterized in that it comprises means configured to implement a method (50) according to one of claims 1 to 7. 11 - UNB type wireless communication system (10) comprising a plurality of terminals (20), an access network (30) comprising a plurality of base stations (31), a server (32), characterized in that it comprises means configured to implement a method (50) according to one of claims 1 to 7.