Characterization of at least a part of an internet network

Abstract

The present invention relates to a method for characterizing at least a portion of a network (20, 21) comprising at least one internet access device (2), characterized in that the internet access device (2) generates and transmits at regular intervals (E3), to at least one server (3), messages encapsulating (E2) characterization data stored in said device (2), said data comprising at least one performance parameter calculated (E1) by said device (2). Figure for the abstract: Fig. 1

Description

Title of the invention: Characterization of at least part of an internet network FIELD OF THE INVENTION

[0001] The invention relates generally to the provision of internet access and more specifically to the characterization of at least one part of an internet network comprising at least one internet access box (or IAD according to the generally used English terminology). STATE OF THE ART

[0002] To check the performance of a network connection, it is desirable to be able to monitor the connectivity and performance between a client and a server. For example, it is common for a user to complain that they no longer have adequate Internet access and call the Internet service provider's customer service. The latter must therefore be able to detect and locate the origin of the problem before the customer feels it and complains about it.

[0003] A first way of monitoring connectivity and performance between a client and a server consists of periodically sending a “ping” type message from a central server to the internet access boxes, in order to detect a potential sudden loss of connection called “clear loss”. Indeed, the server only needs to observe that the response to the ping is no longer received over several periods to note this “clear loss”.

[0004] However, this method only allows testing the reachability of the internet access boxes from the test server; it does not allow knowing whether the box has access to the different services to which the customer has subscribed and does not allow knowing the end-to-end performance for each of these services.

[0005] We also know, for example from document FR3020537, a method for evaluating the performance of a connection between two terminals, with at least one terminal comprising measuring equipment, comprising at least one step of acquiring data contained in the header of packets transmitted under a TCP protocol in the context of exchange sessions. This data may for example be the value of the SYN and SYN / ACK flags, the value of the time stamp field, or the packet number of the session.

[0006] Thanks to this data and the measuring equipment, it is possible, after several steps explained in document FR3020537, to determine the loop time, the latency value, the quantity of data in transit or even the congestion rate.

[0007] However, such a method does not allow for the detection of clear losses. In addition, the performance evaluation is limited and is based only on protocol data contained in the headers of the exchanged TCP packets.

[0008] Furthermore, this type of method only allows centralized information to be retrieved from an entire network during exchange sessions between terminals of the network. No evaluation takes place in the absence of these exchanges. A summary of the quality of the different connections of a given network requires exhaustive prior knowledge of the elements that make up the network. It also involves testing all possible connections from each of the different access boxes of the network, which is particularly complex to implement on a large scale. Statement of the invention

[0009] One aim of the invention is to remedy the aforementioned drawbacks.

[0010] In particular, it offers a solution that allows in a simple and deployable manner to large scale, to evaluate, practically in real time, connection performance on a given network and to detect failures or cuts on it, without prior knowledge of the network's network.

[0011] To this end, according to a first aspect of the invention, a method is proposed for characterizing at least one part of a network comprising at least one internet access box, characterized in that the internet access box generates and transmits at regular intervals, to at least one server, messages encapsulating characterization data stored in said box, said data comprising at least one performance parameter calculated by said box.

[0012] Thus, it is possible to characterize a part of the network without prior knowledge of its topology. In addition, it makes it possible to ensure that the user of the box has access to the services to which he has subscribed.

[0013] The characterization method according to the invention may also have at least one of the following characteristics:

[0014] - the characterization data are extracted from the received messages, said data in originating from different internet access boxes being centralized and processed to deduce at least one piece of information on the operation of all or part of the network,

[0015] - the server extracts the characterization data from the messages received and retransmits to a centralization server the characterization data that it receives from different boxes,

[0016] - the server adds to the characterization data, characterization data complementary, said complementary characterization data being stored by the server and including information on the connection environment of the internet access boxes,

[0017] - the server(s) transmit(s) the characterization data to a server remote centralization, said centralization server processing the data it receives from different servers and generating an alert when one or more messages expected by a server from which it receives data have not been received from an internet access box or when a performance parameter contained in the grouped data differs from a connection condition stored by the centralization server,

[0018] - the centralization server groups data from different internet access boxes and various servers and implements at least one statistical processing on these to deduce at least one performance indicator for at least one given zone on the network and, where appropriate, generate alerts for at least one indicator and at least one zone, and

[0019] - the characterization data are transmitted to the centralization server in flow stretched, the internet access boxes transmitting the messages encapsulating the characterization data at a frequency of less than one day, advantageously less than 6 hours, advantageously less than 1 hour, advantageously greater than 1 minute, preferably less than 10 minutes.

[0020] Thus, it is possible to dissociate a unitary connection problem, i.e. on an isolated box, and a regional problem, i.e. on a plurality of boxes. In addition, such centralization of data makes it possible to process the data more quickly and to promote the customer experience as well as the efficiency of the operator network maintenance services. The method makes it possible to characterize a part of the network according to different geographical scales. It also promotes scalability. Finally, it is possible to adjust the nominal performance of the connections dynamically.

[0021] According to the invention, an internet access box for part of a network is provided, the internet access box being configured to:

[0022] - calculate a performance parameter,

[0023] - storing characterization data including the performance parameter,

[0024] - encapsulate the stored characterization data in a message, and

[0025] - transmitting a message thus prepared to a server at regular intervals.

[0026] The invention also provides a centralization server for a network being configured for said centralization server to process the data that it receives from different servers and to generate an alert when one or more messages expected by a server from which it receives data have not been received from an internet access box or when a performance parameter contained in the grouped data differs from a connection condition stored by the centralization server.

[0027] A computer program product comprising code instructions may be provided for executing the method according to the invention when said program is executed by a computer.

[0028] Storage means readable by computer equipment may be provided on which a computer program product comprises code instructions for executing the method according to the invention. DESCRIPTION OF FIGURES

[0029] Other characteristics, aims and advantages of the invention will emerge from the following description, which is purely illustrative and non-limiting, and which must be read in conjunction with the appended drawings in which:

[0030] [Fig.l] schematically illustrates an example of a connection between an Internet access box and remote servers and service platforms;

[0031] [Fig.2] details the connection between the internet access box and the connected remote server of [Fig.l];

[0032] [Fig. 3] is a flowchart illustrating a mode of implementation of the method of transmitting data from at least one internet access box to a server;

[0033] [Fig.4] is a flowchart illustrating a mode of implementation of the method for characterizing a connection by the server connected to at least one Internet access box;

[0034] [Fig.5] shows diagrammatically several types of connections between Internet access boxes and a multiplexer;

[0035] [Fig.6] shows a diagram of a type of connection between several multiplexers and an operator network;

[0036] [Fig.7] is a graph representing several messages sent by an internet access box and the evolution of the latency of the connection with the server as a function of time;

[0037] [Fig.8] is a graph representing the evolution of latency as a function of time at the level of several multiplexers;

[0038] [Fig.9] is a map obtained according to the method of implementing the process illustrated in [Fig.4] in which the areas having suffered a clear loss are represented;

[0039] [Fig. 10] is another map obtained according to the method of implementing the method illustrated in [Fig.4] in which the connection zones are distinguished according to the number of defects counted;

[0040] [Fig. 11] is a different mapping from [Fig.9] and [Fig. 10] according to the mode of implementation of the method illustrated in [Fig.4] in which the connection zones are distinguished according to the level of latency;

[0041] [Fig. 12] is a bar chart according to the mode of implementation of the method illustrated in [Fig.4] in which the number of anomalies of each connection is classified in descending order; and

[0042] [Fig. 13] is another bar chart according to the mode of implementation of the method illustrated in [Fig.4] in which the latency of each connection is ranked in descending order.

[0043] Throughout the figures, similar elements bear identical references. DETAILED DESCRIPTION OF THE INVENTION

[0044] With reference to [Fig. 1], a method is proposed for characterizing all or part of an internet network comprising at least one internet access box 2 and one or more servers 3.

[0045] By “characterization” we mean at least the capacity to detect and report a problem in the connection between a box and a remote service.

[0046] Typically, an alert on failures concerning a geographical area must be able to be sent in practically real time (for example in less than 10 minutes) to a customer service likely to be questioned by customers on the failures that they themselves notice (for example, if the user complains about their Wi-Fi speed). Architecture - General

[0047] In [Fig. 1], the box 2 is connected to the internet network 20 via an operator network 21. Remote servers 3, typically servers of the operator of the internet access box 2, are arranged in the operator network 21. The various network characterization data are also uploaded to a central server 300 of the operator where they are processed. The box 2 can also connect to several platforms 30 providing the remote services. These platforms 30 can be located in the internet network 20 or the operator network 21.

[0048] An example of a connection between a server 3 and the device 2, illustrated in [Fig.2], presents from upstream to downstream,

[0049] - box 2 (at the subscriber's premises),

[0050] - a multiplexer 4 adapted to DSL, FFTH or FFTO technology, etc., located on the local operator network enabling data to be transmitted from or to subscribers by grouping them onto a single medium,

[0051] - a BRAS type aggregation and encapsulation means 5 (“BNG” in English) in order to enter the operator network 21, and

[0052] - an internet router 6 allowing entry into the internet network 20.

[0053] Conventionally, the internet access box 2 and the server 3 comprise respective data processing means of the processor type, and where appropriate data storage means (a memory), for example a hard disk.

[0054] The box 2 also generally generates a local network (called LAN for Local Access Network) for example via a wired connection, typically Ethernet. It can also be connected to one or more other devices 1a, 1b wirelessly. (Wifi, Bluetooth or any other protocol). Internet access box

[0055] Conventionally, an internet access box 2 has a certain number of characterization data concerning it. This characterization data may be data relating to the operation of said box and in particular performance parameters calculated by it, for example the average WIFI flow rate, the latency and more generally any performance measurement.

[0056] The characterization data may also be data stored in the internet access box 2 and relating to said box: type of box, type of WIFI, and more generally any information on the box and its environment.

[0057] The box 2 has the particularity of being adapted to generate and transmit at regular intervals to one or more proximity servers 3 messages encapsulating characterization data stored by said box, and in particular one or more performance parameter values.

[0058] The sending of messages to the servers 3 by the internet access box(es) 2 is done in a unidirectional manner (Unicast message).

[0059] A server 3 to which different boxes 2 send their characterization messages is itself configured to receive and process the messages that it receives at regular intervals, extract the data from these messages and return in a just-in-time flow the characterization data received from several boxes to a central server 300.

[0060] This characterization data is, where appropriate, enriched by the server 3.

[0061] Server 3 is also configured to detect possible absences of reception of expected messages from an internet access box 2. Information regarding a failure to receive a message by a box 2 is part of the information transmitted in the data flow by the server 3 to the server 300.

[0062] The central server 300 to which the servers 3 are connected is configured to store the characterization data that it receives from different proximity servers 3 and to process this data. In particular, it groups them by category and / or by network zone and determines values ​​for performance indicators from these. In particular, it implements statistical processing on this data. to determine values ​​of local or large-scale performance indicators on the network.

[0063] The central server 300 also processes the performance indicator values ​​to generate alert messages.

[0064] For example, an alert may be generated when the value of a performance indicator diverges from a previously stored value for a zone of the network. Alternatively, an alert processing operation may take into account the evolution of a performance indicator and generate an alert based on this evolution over time.

[0065] Different alert levels can be provided depending on the observed development. Example of characterization

[0066] As indicated previously, an internet access box 2 generates and transmits at regular intervals to the server 3 messages encapsulating data and is configured for this purpose. The boxes 2 can for this purpose integrate this functionality as soon as they are deployed. This functionality can also be programmed during an upgrade by download.

[0067] The main operating steps are, for example, the following.

[0068] During a step E1, the internet access box 2 accesses various characterization data to be transmitted, and in particular performance parameters determined during the operation of the box 2 or even data specific to the box and its environment (type of box, type of WIFI).

[0069] A performance parameter may for example be the throughput of a connection to a remote service of a platform 30, the response time of a web server to the box 2 or even the average throughput of the WiFi of the box 2.

[0070] During a step E2, the device 2 encapsulates the data in a communication protocol using a connectionless transmission mode to form a message.

[0071] Finally, during a step E3, the message is sent by the internet access box 2 to one or more @anycast IP addresses. The case where several addresses are used in fact makes it possible to ensure redundancy in the transmission of information and scalability.

[0072] Scalability means the ability to distribute the load of the processed data, for example if the quantity of data to be processed increases on one or more servers 3.

[0073] According to one embodiment of the method, the periodicity of the transmission may be less than one day, advantageously less than 6 hours, advantageously less than 1 hour, advantageously greater than 1 minute, preferably less than 10 minutes.

[0074] A server 3 intended to receive the characterization messages from an internet access box 2 is configured to receive these messages at regular intervals and to detect any absence of reception of the expected messages. Such a server 3 further extracts the characterization data contained in these messages (step E11). If necessary, it enriches this data with data relating to the environment of the box 2 (ADSL, FTTH, line operator, etc.) (step E12).

[0075] For example, the data relating to the environment of the box 2 may be commercial information concerning a termination of a contract by a customer or their move, but also technical information such as the way in which the box is connected to the network. It is also possible that this information comes from another peripheral, for example a multiplexer 4 communicating information on the loss of a box 2 on the network 21.

[0076] The server 3 then implements calculations and the enriched data from several boxes 2 are aggregated (step El3) before transmitting the aggregated data and the results of the calculations in a just-in-time flow to a centralization server 300 (step El4).

[0077] For example, a calculation carried out by the server 3 may consist of incrementing a counter each time there is a loss of synchronization with the boxes 2 in order to note the loss of messages sent by the boxes 2.

[0078] Aggregation can consist of grouping the enriched data from several boxes 2 according to several scales:

[0079] - at the level of the multiplexer 4, where a set 200 of devices are connected,

[0080] - at the level of cards 7a, 7b, 7c of the multiplexer 4. Where several sub-assemblies 201 contain a part of all devices 2 connected to multiplexer 4, and

[0081] - at the level of loops 22, comprising all of the multiplexers 4 connected to the 6 internet access router 21.

[0082] Several examples of groupings are illustrated in [Fig.5] and [Fig.6].

[0083] Different levels of characterization can thus be envisaged in order to dissociate a unitary connection problem, i.e. on an isolated box 2, and a regional problem, i.e. on a plurality of boxes 2 connected to a multiplexer 4.

[0084] During a step E15, the data received by the central server 300 from different servers are aggregated.

[0085] An additional level of characterization can thus be envisaged.

[0086] Typically, for example, the server 300 implements statistical analyses by performance indicator: determination of average and percentiles by server or by multiplexer.

[0087] In one possible implementation mode, the server 300 detects abnormal developments, for example by means of Machine Learning-based algorithms to which we submit as input the evolutions of the performance indicators, by grouping the messages from several boxes according to the possible characterization levels.

[0088] The processing of the data thus grouped may for example consist of taking an average on each of the groups, or more generally, statistical operations, then comparing the processed data with the connection conditions stored by the server 300. The comparison may be done for example by means of artificial intelligence algorithms. This comparison makes it possible to obtain a diagnosis of each connection over a large territory and to adjust the nominal performances of these connections.

[0089] During a step El6, the server 300 generates an alert message when at least one connection problem is detected: clear loss on a box; clear losses on a set of boxes in the same part of a network; or even loss of performance on a part of the network.

[0090] Alert messages can be sent to customer service. They can also be sent automatically to the customers themselves, if necessary using communication channels other than the tested network (sending emails and SMS messages in parallel).

[0091] During a step El7, the central server 300 displays the contents of the database to the operator.

[0092] Displays of maps at different scales of the network are also possible, particularly in the context of quasi-real-time monitoring of the state of a network.

[0093] For example, the content can be displayed on a graph 50 showing the evolution of a quantity corresponding to a criterion of the connection condition.

[0094] [Fig.7] illustrates an example of curves that can be displayed by the central server 300. On the abscissa is the time in minutes, and on the ordinate is the value of the “DNS” latency, i.e. the response time between an internet access box 2 and a web server, in ms. In this example, the criterion chosen in the connection condition is a nominal value of the latency represented by a horizontal line 10 on the graph. For each 10-minute interval, a peak 8 corresponds to the data received by the server 3 when the data is generated at regular intervals. Curve 9 corresponds to the value of the latency over time, the latency being able to be one of the pieces of information contained in the message data. In zone 11, no message arrives at the server, corresponding to a clear cutoff.In areas 12, 13 and 14, messages are received correctly and the information regarding latency indicates an increase in its value because curve 9 is above the nominal value of latency 10, this nominal value could for example correspond to a performance parameter.

[0095] [Fig.8] illustrates another example in which several curves 51, 52, 53 correspond to the value of the loop latency 22 over time. In this example, the criterion chosen in the connection condition is a nominal value of the latency. Each curve 54, 55, 56 increases sharply, which corresponds to a significant increase in the latency 57. The abnormal increase in the latency moves the curve away from the nominal value of the latency, indicating a loss of connection.

[0096] The contents of the database may be displayed on a map 40 in which different levels of criticality related to a fault on the network may be highlighted. Three examples of maps are given below.

[0097] [Fig.9] illustrates the mapping of connectivity losses detected at an instant in an area on the scale of a loop 22. Points 41 correspond to all the loops where a clear loss was diagnosed in the area.

[0098] [Fig. 10] illustrates the mapping of the number of anomalies detected following saturation of the data exchange between several access providers at the scale of a loop 22. Each pattern 42, 43, 44, 45 corresponds to a different number of anomalies.

[0099] [Fig. 11] illustrates the mapping of latency at the national level on the scale of a multiplexer 4. Each pattern 46, 47, 48, corresponds to a different latency level.

[0100] Finally, it is possible to visualize the results on a bar graph 60, the results being able to be classified in an ascending or descending order. Two examples of bar graphs 60 are illustrated below.

[0101] [Fig. 12] illustrates a bar graph with several entries and classifying at the scale of a city the number of anomalies detected following a saturation of the exchange of data between several access providers. A first column 61 lists the selected multiplexers, a second column 62 lists the cities associated with the selected multiplexers and a last column displays a number of anomalies 63.

[0102] [Fig. 13] illustrates a bar graph with several entries and classifying the value of the latency at the scale of a region. A first column 64 lists the selected multiplexers 4, a second column 65 corresponds to the operator of the multiplexer 4, a third column 66 lists the regions associated with the selected multiplexers, and a last column associates a number of anomalies 67.

[0103] The method may be executed by a computer program comprising code instructions, when the program is executed by a computer.

[0104] The computer program is recorded in storage means readable by computer equipment.

Claims

Claims

1. Method for characterizing at least one part of a network (20, 21) comprising at least one internet access box (2) and several servers (3) having at least one same anycast IP address, the method being characterized in that: - the internet access box (2) generates and transmits at regular intervals (E3), to the same anycast IP address, messages encapsulating (E2) characterization data stored in said box (2), said data comprising at least one performance parameter calculated (El) by said box (2), at least one performance parameter thus encapsulated being a connection rate of the box to remote services to which the box has access on the internet;and - the servers corresponding to the anycast IP address (3) transmit (El3) the characterization data received from different boxes to a remote centralization server (300), said centralization server (300) processing the data that it receives from these different servers and generating an alert (E15) when one or more messages expected by a server (3) from which it receives the data, has not (have not) been received from an internet access box (2) or when a performance parameter contained in the grouped data differs from a connection condition stored by the centralization server (300).;

2. Method according to claim 1, in which the characterization data are extracted from the messages received, said data coming from different internet access boxes (2) being centralized and processed to deduce at least one piece of information on the operation of all or part of the network.

3. Method according to the preceding claim, in which the server (3) extracts the characterization data from the received messages (Eli) and retransmits to a centralization server (300) the characterization data that it receives from different boxes.

4. Method according to the preceding claim, in which the server (3) adds to the characterization data, complementary characterization data (El2), said complementary characterization data being stored by the server (3) and comprising information on the connection environment of internet access boxes (2).

5. Method according to one of claims 1 to 4, in which the centralization server (300) groups together the data (E14) coming from different internet access boxes (2) and different servers (3) and implements at least one statistical processing on these to deduce therefrom at least one performance indicator of at least one given zone on the network and, where appropriate, generate alerts for at least one indicator and at least one zone.

6. Method according to one of claims 1 to 5, in which the characterization data are transmitted to the centralization server (300) in a just-in-time flow, the internet access boxes (2) transmitting the messages encapsulating the characterization data according to a periodicity of less than one day, advantageously less than 6 hours, advantageously less than 1 hour, advantageously greater than 1 minute, preferably less than 10 minutes.

7. Internet access box (2) of a part of a network (20, 21), the internet access box (2) being configured to: - calculate at least one performance parameter being a connection rate of the internet access box (2) to remote services to which the box has access on the internet, - store characterization data comprising the performance parameter, - encapsulate the stored characterization data in a message, and - transmit at regular intervals a message thus prepared to at least one anycast IP address of several servers (3).

8. Centralization server (300) of a network (20, 21), the server (300) being configured to process data that it receives from different servers and generate an alert (E15) when one or more messages expected by a server from which it receives data, have not been received from an internet access box (2) or when a performance parameter contained in the grouped data differs from a connection condition stored by the centralization server (300).

9. Computer program product comprising code instructions, for the execution of a method according to one of claims 1 to 6 when said program is executed by a computer

10. Storage means readable by computer equipment on which a computer program product comprises code instructions for executing a method according to one of claims 1 to 6.

Images