Method for identifying items of equipment present in a home network

EP4460776A1Pending Publication Date: 2024-11-13SOFTATHOME
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
EP2023700272
Authority / Receiving Office
EP · EP
Patent Type
Applications
Current Assignee / Owner
Priority Date
2022-01-04
Filing Date
2023-01-04
Publication Date
2024-11-13

AI Technical Summary

Technical Problem

Existing methods for identifying equipment in home networks, particularly those with thousands of devices and unbalanced distributions, are inefficient and unsuitable for recognizing less common devices due to their reliance on multi-class classification approaches.

Method used

A method utilizing a statistical model based on neural networks to generate a digital fingerprint of equipment, comparing it to pre-recorded fingerprints in a reference base, and employing a KNN algorithm to identify similar devices, allowing for recognition of new and less popular equipment through machine learning and a hybrid solution combining expert systems and automatic learning.

Benefits of technology

Enables efficient identification of a large number of equipment types, including less common devices, by using machine learning to generate digital fingerprints and compare them to a non-exhaustive reference base, reducing computational costs and improving recognition accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 1.1
    Figure 1.1
Patent Text Reader

Abstract

The invention relates to a method for identifying a first item of equipment present in a communication network, this method being implemented by a processing unit and comprising the following steps: - receiving identification data from said first item of equipment, - using a neural network-based statistical model to compute a digital fingerprint of the first item of equipment based on the identification data, - successively determining distances between the computed digital fingerprint and, respectively, digital fingerprints pre-recorded in a reference base; these pre-recorded digital fingerprints being digital fingerprints of known items of equipment, - identifying the first item of equipment as being a known item of equipment when the distance between the digital fingerprint of the first item of equipment and the pre-recorded digital fingerprint of said known item of equipment is less than a predetermined threshold.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] DESCRIPTION

[0002] TITLE: Method for identifying equipment present in a home network.

[0003] The present invention relates to a method for identifying a first device present in a communication network.

[0004] Generally speaking, device identification in a home network is based on multi-class classification type machine learning solutions. These solutions are all based on environments composed of twenty to thirty different devices.

[0005] The environment of a telecom operator is very different. Millions of users use several thousand device models. Moreover, among these devices, some are very popular and much more present than others (several hundred thousand times more). Having several thousand classes and a very unbalanced distribution of models makes the multi-class classification approach inappropriate for networks of several thousand devices.

[0006] Document US20200382376A1 describes a method for classifying equipment by classifying equipment into a first category according to a first classification technique for a predetermined duration. A second classification technique is applied if the equipment has not changed category during the predetermined duration.

[0007] Document US10652116B2 describes a method for classifying equipment. Various information is used to enable classification.

[0008] The present invention aims to identify a very large number of devices.

[0009] Another aim of the invention is to design a new identification method capable of easily identifying numerous new pieces of equipment.

[0010] The invention also aims to provide a new identification method capable of reliably recognizing unpopular equipment. At least one of the objectives is achieved with a method for identifying a first piece of equipment present in a communications network, in particular a home network, this method being implemented by a processing unit and comprising the following steps:

[0011] - reception of identification data from said first equipment,

[0012] - use of a statistical model based on a neural network to calculate a digital fingerprint of the first equipment from the identification data,

[0013] - successive determination of distances between the calculated digital fingerprint and respectively digital fingerprints pre-recorded in a reference base; these pre-recorded digital fingerprints being digital fingerprints of known equipment,

[0014] - identification of the first equipment as known equipment when the distance between the digital fingerprint of the first equipment and the pre-recorded digital fingerprint of said known equipment is less than a predetermined threshold.

[0015] With the method according to the invention, a new digital fingerprint approach is used rather than a classification as in the prior art. The invention makes it possible to recognize in particular equipment that was not known during the training of the model.

[0016] The distance of the equipment to be identified can advantageously be calculated in relation to all the fingerprints in the reference base, in particular with a KNN (K nearest neighbors) type algorithm. This makes it possible to effectively know which equipment in the reference base is most similar to the equipment to be identified.

[0017] The statistical model is derived from machine learning technology based on neural networks. The input data is used to predict a digital fingerprint. A search is then initiated in the reference database to find a pre-registered digital fingerprint that most closely matches or resembles the fingerprint just predicted or determined.

[0018] The invention uses machine learning instead of a simple signature creation algorithm. Using machine learning is less time-consuming. Identifying a device involves identifying the operating system and the type of device.

[0019] Preferably, the following information can be identified for each piece of equipment:

[0020] - the type of equipment, for example a computer, a camera, a television...

[0021] - the seller of the equipment, for example “apple”,...

[0022] - the product line of the equipment, for example “iphone”,...

[0023] - the model of the device, for example an “iPhone 13 Pro Max”,...

[0024] - the operating system used by the equipment, for example “iOS”,...

[0025] - the version of the operating system used by the equipment, for example “15.1.0”,...

[0026] The invention makes it possible to detect a very large number of types of equipment by using machine learning to generate a digital fingerprint and comparing it to digital fingerprints contained in a reference database.

[0027] Using machine learning and the reference database allows for a limited number of digital fingerprints in the reference database. Indeed, the comparison is not necessarily made identically, but we are looking for a digital fingerprint that is closest. Therefore, it is not necessary to accurately list all the fingerprints of all the equipment.

[0028] This allows us to have a non-exhaustive reference base.

[0029] In the method according to the invention, if the equipment is not recognized, a manual or automatic process can be implemented to retrieve a set of data from the equipment being identified. This can be done in particular by means of an application communicating with the equipment. The digital fingerprint of this equipment can then be recorded in the reference database.

[0030] According to an advantageous characteristic of the invention, the pre-recorded digital fingerprints can be vectors obtained from identification data of known equipment. In addition to the above, each vector can be determined from:

[0031] - on the one hand, identification data containing textual information on which processing is applied by means of a subset of the neural network based on recurrent neurons to determine a first sub-vector,

[0032] - on the other hand, identification data containing categorical information to which an ordinal encoding is applied to determine a unique code per category, then an embedding to associate each category with a second sub-vector,

[0033] - the first and second sub-vectors are then combined using at least one fully connected layer (“dense layers” in English) to form said vector.

[0034] In particular, each category is transformed into a single subvector.

[0035] According to one embodiment of the invention, the subset of the neural network based on recurrent neurons may be a recurrent network of the LSTM (“Long Short Term Memory”) or GR.U (“Gated Recurrent Unit”) type. Other types of approaches may be envisaged, such as attention-based layers, of the transformer type.

[0036] According to one embodiment of the invention, for a given piece of equipment, the statistical model may comprise a triplet cost function (“Triplet loss” in English) making it possible to generate a vector closer to the vectors of equipment identical to said given piece of equipment and further from the vectors of equipment different from said given piece of equipment.

[0037] Such an implementation consists of analyzing not only the likelihoods but also the differences.

[0038] According to another embodiment of the invention, for a given piece of equipment, the statistical model may comprise a cost function per contrast loss (“Contrastive loss” in English) making it possible to generate a vector closer to the vectors of equipment identical to said given piece of equipment and further from the vectors of equipment different from said given piece of equipment. According to the invention, the pre-recorded digital fingerprints may be predetermined by the neural network from the following data of the known pieces of equipment:

[0039] - DHCP protocol identifiers including hostname, options, vendor class, list of options in a request packet,

[0040] - the first three bytes of the MAC address (OUI),

[0041] - names of mDNS announcement services, for example ptrs fields of mDNS announcements,

[0042] - WiFi data,

[0043] - TLS client and server fingerprints, e.g. in JA3 format,

[0044] - the list of domain names contacted,

[0045] - the number of different domain names contacted,

[0046] - the list of network ports used (TCP and UDP),

[0047] - the list of open network ports (TCP and UDP),

[0048] - network communication timing information including the frequency of connection to Wi-Fi and DHCP server, and / or the frequency of network access by domain name, and

[0049] - the type of network connection: wifi or ethernet.

[0050] In addition to the above, Wi-Fi data may include:

[0051] - HT / VHT / HE capacities,

[0052] - the first three bytes present in the vendor-specific label,

[0053] - the number of antennas,

[0054] - the list of supported MCS (“Modulation and Coding Scheme”),

[0055] - the maximum bandwidth supported,

[0056] - the capacities of UNII (“Unlicensed National Information Infrastructure”) tapes,

[0057] - spatial flow: maximum rx / tx supported,

[0058] - supported standards,

[0059] - the supported radio standards. According to an advantageous embodiment of the invention, the identification data comprises at least one of the following data:

[0060] - a user agent in an HTTP or QUIC protocol,

[0061] - DHCP protocol identifiers including hostname, vendor class, user class and vendor specific information,

[0062] - names of mDNS advertisement services, for example “_printer._tcp”, and

[0063] - UPnP protocol data including: manufacturer, familiar name, model, description, model number.

[0064] With at least one of these data, it is possible to identify a piece of equipment.

[0065] According to a preferred embodiment of the invention, before using the statistical model, the identification data can first feed an expert system capable of identifying the equipment or transmitting the identification data to the statistical model if the identification fails; the expert system comprising an equipment recognition algorithm based on regular expression-based rules.

[0066] With such an embodiment, the present invention constitutes a hybrid solution composed of an expert system and a machine learning solution. A first test is carried out with the expert system, if the equipment is identified, the algorithm stops, otherwise the identification data is sent to the statistical model for identification. If the statistical model fails to identify the equipment, human intervention can be provided, or not, to develop a digital fingerprint of this equipment and record it in the reference database.

[0067] The expert system is composed of rules based on regular expressions to recognize certain characteristic patterns. These rules use identification data to identify a piece of equipment.

[0068] Advantageously, the reference base may comprise digital fingerprints obtained from data from information collections and fingerprints obtained from data synthesized from a generator.

[0069] The collected data comes from retrieving information from listed equipment or from queries that have been made. Otherwise, the data collected consists of a sample of queries that have been made. This sampling is done in such a way as to have a sufficient quantity of examples for each equipment to be recognized. It is important in this sampling to balance the number of samples for each type of equipment.

[0070] For equipment that is not widely available, the data generator can generate a sufficient number of examples to add to the reference database.

[0071] The generator used for the reference base can also be used during the model training phase. In this case, it is used to rebalance the data distribution. Balancing the distribution of classes during model training is essential for its subsequent good performance.

[0072] Thus, the reliability of the system's prediction depends on the amount and type of information provided to it as input. To do this, the data used is balanced as much as possible:

[0073] - by deleting data from the most popular equipment, and

[0074] - using synthetic data for rarer equipment.

[0075] The present invention also relates to a computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the method described above.

[0076] There is also provided a data processing system comprising a processor adapted to the method described above.

[0077] Other advantages and characteristics of the invention will appear on examining the detailed description of a non-limiting embodiment, and the appended drawings, in which:

[0078] [Fig. 1] Figure 1 is a schematic overall view of an operating system; [Fig. 2] Figure 2 is a schematic view of different steps of identifying equipment according to the invention; and

[0079] [Fig. 3] Figure 3 is a schematic view of an example of the architecture of a neural network according to the invention.

[0080] The embodiments which will be described below are in no way limiting; it will be possible in particular to implement variants of the invention comprising only a selection of characteristics described below isolated from the other characteristics described, if this selection of characteristics is sufficient to confer a technical advantage or to differentiate the invention compared to the state of the prior art. This selection comprises at least one preferably functional characteristic without structural details, or with only a part of the structural details if this part alone is sufficient to confer a technical advantage or to differentiate the invention compared to the state of the prior art.

[0081] We will first describe, with reference to Figure 1, a wireless home local area network 100. The center of the network is a home gateway 1 distributing an internet connection to the different stations 2 connected in the home. The home gateway 1 is placed in the home, in the living room for example. The home gateway 1 is the central point of all flows: images, music, videos etc. Each station 2 is connected by WIFI or by an Ethernet cable to the gateway. The home gateway 1 is connected, on one side to the local area network 4, in English "Local Area Network" or LAN of the home and on the other side, to the external network 5, in English "Wide Area Network" or WAN. The local area network 4 designates any interconnected computer installation or station 2 within the user's home such as for example, a television, a tablet, a mobile phone or even a game console.A remote server 3, such as a server of a telecom operator, is connected to the local network 4 via the external network 5. The home gateway 1 is for example a device of said telecom operator.

[0082] The purpose of the present invention is to identify any equipment connecting to the telecom operator's devices. To do this, an algorithm according to the invention is implemented in the remote server 3.

[0083] The context of the home local area network is exemplary. The method described by the present invention is applicable to other types of wireless communication networks, for example to a university network, to corporate networks, etc.

[0084] Figure 2 is an overall view of a process according to the invention.

[0085] A device such as a mobile phone 6 connects to the server 3. The latter will implement the method according to the invention to identify this mobile phone, that is to say to identify the following elements:

[0086] - the type of equipment, for example a computer, a camera, a television...

[0087] - the seller of the equipment,

[0088] - the product line of the equipment,

[0089] - the model of the equipment,

[0090] - the operating system used by the equipment, and

[0091] - the version of the operating system used by the equipment.

[0092] The minimum to identify a device is, for example, to find its operating system and its type of device.

[0093] Once the equipment is identified, the result obtained can be:

[0094] - displayed on a display device either for the owner of the equipment or for a customer support service to help resolve problems,

[0095] - used in other services such as quality of service (adapting the network configuration to each device), IT security (detecting that a device requires a software update, isolating a device, etc.), physical security (detecting abnormal behavior that is a sign of a possible burglary), etc.

[0096] The result obtained is therefore data that can be transmitted to another processing device.

[0097] The server 3 comprises hardware and software components for implementing the invention. However, these components can be integrated into a single device or distributed across several nearby or distant devices.

[0098] The server 3 comprises an expert system 7, a statistical model 8 based on a neural network, a search algorithm 9, a reference base 10 and outputs 11 and 12. When the telephone 6 connects to the telecom operator's network, identification data for this telephone are transmitted to the server 3.

[0099] Expert system 7 applies its rules.

[0100] If phone 6 is already known to the expert system, then this phone is identified as 11.

[0101] If phone 6 connects for the first time, the expert system deduces that this phone is not known. The identification data is then sent to the statistical model 8 which generates a digital fingerprint 13.

[0102] The knn search algorithm 9 is used to search the reference database 10 for a pre-recorded digital fingerprint that would be close to the digital fingerprint 13. A distance between the digital fingerprint 13 and each pre-recorded digital fingerprint is tested. When this distance is less than a predetermined threshold, then the digital fingerprint 13 is considered to be the pre-recorded digital fingerprint allowing it to go below the threshold. In this case, the identification is confirmed at 11. The reference 11 can be a visual or audio output device, an entry in a database, or other. For example, information can be sent to the recognized equipment.

[0103] When the phone is identified, services can be unlocked or information exchanged between the recognized equipment and the server 3 or any other device.

[0104] If the phone is not recognized in the reference database, the digital fingerprint 13 is placed in a lObis prediction database. Human intervention can then manually identify the device and place its signature in the reference database by associating it with the phone 6.

[0105] Indeed, when a piece of equipment is not recognized, its digital fingerprint is only put into the reference database after manual identification of the equipment. In the meantime, the data associated with this equipment is saved in a second database called the lObis prediction database, which includes the list of predictions already made.

[0106] The present invention advantageously uses machine learning techniques not to predict and identify the device, but to generate a digital fingerprint. This saves time because a database is used to quickly search for a similar digital fingerprint. Given the large number of devices to be identified for a given operator, using machine learning to directly identify the device would be too costly in terms of computation time.

[0107] The statistical model can be updated as well as the reference base based on fingerprints received from the same device. A fingerprint from the same device can change, for example, following equipment updates.

[0108] For example, all identification requests are saved in the lObis prediction database to establish statistical data and improve the reference database and the neural network.

[0109] The neural network uses the following information:

[0110] - DHCP protocol identifiers including hostname, options, vendor class, list of options in a request packet,

[0111] - the first three bytes of the MAC address (OUI),

[0112] - names of mDNS announcement services,

[0113] - WiFi data:

[0114] - HT / VHT / HE capacities,

[0115] - the first three bytes present in the vendor-specific label,

[0116] - the number of antennas,

[0117] - the list of supported MCS (“Modulation and Coding Scheme”),

[0118] - the maximum bandwidth supported,

[0119] - the capacities of UNII (“Unlicensed National Information Infrastructure”) tapes,

[0120] - spatial flow: maximum rx / tx supported,

[0121] - supported standards,

[0122] - supported radio standards,

[0123] - TLS client and server fingerprints,

[0124] - the list of domain names contacted,

[0125] - the number of different domain names contacted,

[0126] - the list of network ports used (TCP and UDP),

[0127] - the list of open network ports (TCP and UDP), - network communication timing information including the frequency of connection to Wi-Fi and the DHCP server, and / or the frequency of network access by domain name, and

[0128] - the type of network connection: wifi or ethernet.

[0129] Other sources of information can be added to improve the statistical model.

[0130] The neural network takes the information listed above as input and converts it into a vector. The input information is used either directly or encoded.

[0131] Figure 3 is a schematic view of a simplified architecture of a neural network according to the invention.

[0132] The textual information 14 such as the "hostname" is used as is, and processed by a subset 15 of the neural network based on "recurrent" neurons, for example of the LSTM type, or GR.U. Other approaches for processing text, such as "attention", can be used for this type of data. A first sub-vector is formed. The subset 15 makes it possible to identify patterns characteristic of certain equipment, for example from their "network" name (hostname dhcp, name mdns, friendly name upnp). For example, during training, the model will learn that the text "iphone"® is common to the iphone® product line. On the other hand, it will learn to ignore patterns that are common in names but not significant for the identification of the equipment.For example, in a network name like "iphone-de-romain", the model will have learned that the "romain" part is not significant because it is present in several different devices.

[0133] The categorical information 16, for example the “TLS fingerprint”, is first encoded by ordinal encoding 17 in the form of a unique number per category. These categories are then used in an embedding layer 18 which makes it possible to associate each category with a second sub-vector which will be learned during the training of the statistical model.

[0134] The first sub-vector and the second sub-vector feed several fully connected layers 19 (“dense layers”) to combine sub-vectors and form said vector 20. The neural network is trained with the objective of generating vectors that are close for identical devices, and distant for different devices. This can be done in different ways such as for example by using a cost function per triplet (“Triplet loss” in English) or a cost function per contrast loss (“Contrastive loss” or “contrastive learning” in English).

[0135] The cost per triplet tells the model which examples are identical and which are different. This learning is done by triplets of examples: two examples from the same class and one from another class. The model is then explicitly trained to strongly separate the different examples and group similar examples together as much as possible.

[0136] Contrastive learning does not require different examples, only examples from the same class. The model is then trained to group similar examples together. The separation of different examples is done implicitly.

[0137] The reference base is composed of two different sources of information:

[0138] - data from the collection of information on requests that have been previously made,

[0139] - data synthesized from a generator.

[0140] The data collected consists of a sample of the queries that were made. This sampling is done in such a way as to have a sufficient quantity of examples for each device to be recognized. It is preferable in this sampling to balance the number of samples for each type of device.

[0141] For equipment that is not widely available, the data generator can generate a sufficient number of examples to add to the reference database.

[0142] The generator used for the baseline can also be used during the model training phase. In this case, it is used to rebalance the data distribution. Balancing the class distribution when training a model helps achieve good performance.

[0143] In general, the present invention comprises two steps: - a first step consisting of a statistical model based on a neural network which calculates a digital fingerprint of the equipment to be recognized, and

[0144] - a second step consisting of finding this fingerprint in a database of known equipment. This search is done by calculating the distance between the fingerprint of the equipment to be searched for and that of known equipment. If a known equipment is at a sufficiently small distance from the equipment to be recognized, then we consider that there is a match.

[0145] Of course, the invention is not limited to the examples just described. Many modifications can be made to these examples without departing from the scope of the present invention as described.

Claims

DEMANDS 1. A method for identifying a first piece of equipment present in a communication network, this method being implemented by a processing unit and comprising the following steps: - receiving identification data from said first piece of equipment, - use of a statistical model based on a neural network to calculate a digital fingerprint of the first device from identification data, - successive determination of distances between the calculated digital fingerprint and, respectively, the digital fingerprints pre-recorded in a reference database; these pre-recorded digital fingerprints being digital fingerprints of known equipment, - identification of the first device as being a known device when the distance between the digital fingerprint of the first device and the pre-registered digital fingerprint of said known device is less than a predetermined threshold.

2. A method according to claim 1, characterized in that the pre-recorded digital fingerprints are vectors obtained from known equipment identification data.

3. A method according to claim 2, characterized in that each vector is determined from: - on the one hand, identification data containing textual information on which processing is applied using a subset of a recurrent neural network to determine a first sub-vector, - on the other hand, identification data containing categorical information on which ordinal encoding is applied to determine a unique code per category, then embedding to associate each category with a second sub-vector, - the first and second sub-vectors are then combined by means of at least one fully connected layer ("dense layers") to form said vector.

4. Method according to claim 3, characterized in that the subset of the neural network based on recurrent neurons is a recurrent network of type LSTM (“Long Short Term Memory”) or GR.U (“Gated Recurrent Unit”).

5. Method according to claim 3 or 4, characterized in that, for a given piece of equipment, the statistical model includes a triplet cost function (“Triplet loss” in English) allowing the generation of a vector closer to the vectors of identical equipment said given equipment and further away from the vectors of different equipment said given equipment.

6. Method according to claim 3 or 4, characterized in that, for a given piece of equipment, the statistical model includes a contrast loss cost function ("Contrastive loss" in English) allowing the generation of a vector closer to the vectors of identical equipment said given equipment and further away from the vectors of different equipment said given equipment.

7. A method according to any one of the preceding claims, characterized in that the pre-recorded digital fingerprints are predetermined by the neural network from the following data from known equipment: - DHCP protocol identifiers including the hostname, options, vendor class, and the list of options in a request packet, - the first three octets of the MAC address (YES), - Names of mDNS announcement services - WiFi data, - TLS client and server fingerprinting ("Fingerprints"), - the list of domain names contacted, - the number of different domain names contacted, - the list of network ports used (TCP and UDP), - the list of open network ports (TCP and UDP), 17 - - network communication timing information including the connection frequency to the Wi-Fi and DHCP server, and / or the network access frequency per domain name, and - the type of network connection: wifi or ethernet.

8. A method according to claim 7, characterized in that the Wi-Fi data comprises: - the HT / VHT / HE capacities, - the first three bytes present in the supplier's specific label, - the number of antennas, - the list of supported MCS (“Modulation and Coding Scheme”), - the maximum supported bandwidth, - the capacities of UNII bands (“Unlicensed National Information Infrastructure”), - spatial flux: maximum supported rx / tx, - the supported standards, - Supported radio standards.

9. A method according to any one of the preceding claims, characterized in that the identification data includes at least one of the following: - a user agent in an HTTP or QUIC protocol, - DHCP protocol identifiers including the hostname, vendor class, user class, and vendor-specific information. - names of mDNS announcement services, and - data relating to the UPnP protocol including: manufacturer, common name, model, description, model number.

10. A method according to any one of the preceding claims, characterized in that, before using the statistical model, the identification data first feeds into an expert system capable of identifying the equipment or transmitting the identification data to the statistical model if identification fails; the expert system comprising a — lo _ equipment recognition algorithm based on regular expression rules.

11. A method according to any one of the preceding claims, characterized in that the reference database comprises digital fingerprints obtained from data collected through information gathering and digital fingerprints obtained from data synthesized from a generator.

12. A computer program comprising instructions which, when the program is executed by a computer, cause the computer to implement the method according to any one of the preceding claims.

13. A data processing system comprising a processor adapted to the method according to any one of claims 1 to 11.