Techniques for detecting unwanted outgoing calls

Abstract

Method for identifying unwanted calls originating from an end device of a subscriber line in a communications network, comprising the following steps: • Receiving call information data that characterizes the call made, by a call monitoring unit of the communications network; • Analysis of call information data on an identification server, wherein an algorithm based on self-learning artificial intelligence for identifying unwanted calls is implemented on the identification server, the algorithm performing the following steps: ◯ Creating a call pattern based on call information data; ◯ Comparison of the generated call pattern with defined criteria stored on the identification server for unwanted calls; ◯ Determine whether the call is an unwanted call, characterized by the fact that the call information data includes the address of the destination number and / or the age of the destination subscriber.

Description

[0001] The present invention relates to the technical field of analyzing outgoing calls from a subscriber line. Specifically, it concerns techniques for analyzing and classifying outgoing calls and assigning them to a specific group, for example, as desired or undesired calls. The invention specifically relates to a method for detecting undesired calls, a corresponding communication network, and an identification server. Within the scope of this invention, the detection of an undesired outgoing call is also referred to as identification. Specifically, the identification server can perform the corresponding analysis of the outgoing calls.

[0002] Fraudulent phone calls are becoming increasingly popular. These fraudulent calls may also benefit from new automation techniques that can make calls autonomously or assist people in doing so. Furthermore, the increasing prevalence of flat-rate phone plans makes such calls more lucrative for fraudsters. Not only the elderly, but virtually anyone can fall victim to these scams and potentially lose large sums of money.

[0003] Currently, there is no effective way to defend against fraudulent phone calls, other than for the recipient to recognize the attempted scam. However, this is often not the case. The consequences can include not only financial but also psychological harm, as the person has become the victim of a crime. The now well-known "grandchild scam," in which perpetrators impersonate a relative on the phone, claiming to be in distress and urgently needing a large sum of money from grandma or grandpa, is also becoming increasingly popular among organized gangs.

[0004] In general, other calls, which may involve contracts of any kind, are also a popular means of defrauding people and sometimes swindling them out of large sums of money.

[0005] Techniques for detecting unwanted calls are known from EP 3 629 567 A1 and US 2020 / 0193051 A1.

[0006] The present invention therefore aims to provide techniques that make it possible to efficiently protect a recipient from fraudulent calls from a caller.

[0007] This problem is solved by the method according to claim 1, the communication network according to claim 6 and the identification server according to claim 8.

[0008] According to the invention, a method for identifying unwanted calls originating from an end device of a subscriber line in a communication network is specified, wherein the method comprises the following steps: • Receiving call information data that characterizes the call made by a call monitoring unit of the communication network; in particular, the call monitoring unit of the communication network can be an identification server as described below, since the identification server already performs an analysis of the call information data according to the invention, this data does not need to be additionally transmitted to the identification server in this case. However, if the call monitoring unit is an external unit different from the identification server, the call information data is transmitted to the identification server via a communication channel. The call monitoring unit thus has technical access to data that characterizes the call.Since the network operator of the communication network needs to be informed of a large amount of call information data in order to route the call and establish the call connection, this data can be easily extracted by the call monitoring unit, especially on the part of the network operator. • Analysis of call information data on an identification server, wherein an algorithm for identifying unwanted calls is implemented on the identification server, the algorithm performing the following steps: o Creating a call pattern based on the call information data; o Comparison of the generated call pattern with defined criteria stored on the identification server for unwanted calls; o Determine whether the call is an unwanted call. • According to the invention, the call information data includes the address of the destination number and / or the age of the destination subscriber.

[0009] Typically, call information data is collected from the caller over an extended period, which is what makes it possible to create a call pattern in the first place. This is based on the idea that it's not always possible to conclude that a single call is unwanted or spam. Usually, only a cluster of calls with certain characteristics can be used to identify an unwanted call. However, in individual cases, criteria can be established that indicate an unwanted call upon the first call to a subscriber's number (for example, a blacklist of phone numbers). The collected call information data is then processed by the algorithm to create a call pattern, or rather, a model of the caller.This call pattern can be regularly compared with predetermined criteria or predefined call patterns, so that a match can indicate a potentially unwanted call. This efficiently enables the identification of subscriber lines making unwanted calls, i.e., spam calls, and allows for the implementation of appropriate technical measures. Advantageously, only call information data already known to the system is used for this purpose. However, in one embodiment, it is also possible to generate additional call information data. Such call information data can include: parameters of a SIP Invite header, which signals the call: the From and To fields, as well as the analysis of the P-Asserted-Identity field, which indicates whether someone has spoofed their number. Furthermore, parameters of the called line or subscriber are also evaluated, such as...Where is the connection located (location, address), age / date of birth, type of connection: business / private).

[0010] The algorithm is based on a self-learning artificial intelligence. This intelligence is particularly efficient at identifying call patterns and, due to its self-learning capabilities, continuously improves its results. This offers the advantage that the longer the method is used, the more reliably unwanted calls can be identified. The AI ​​algorithm processes the parameters of the target number's location / address and / or the target subscriber's age / date of birth. Furthermore, the algorithm can sequentially process the following parameters to calculate a spam / scam probability: i) origin of the call; ii) target number of the call over time (frequency); iii) P-Asserted Identity; and / or iv) subscriber classification: business / personal.This is followed by a processing and comparison of all incoming calls within a specific time window based on the parameters mentioned above, in order to take not only the individual call, but all calls (within a time window and based on a source number or number range (A-participant call frequency)) to calculate a spam / scam probability emanating from that source number.

[0011] In one implementation, the call information data is temporarily stored on the identification server. This allows the algorithm to generate the corresponding call patterns. Another possibility is to use the call information data to continuously adapt the call pattern model, so that the actual call information data can be deleted, with only the adjustments to the corresponding model parameters remaining in the algorithm. This is a way to use storage space more efficiently and also to comply with data protection regulations, if applicable.

[0012] According to the invention, the call information data includes the address of the destination number and / or the age of the destination subscriber. With regard to the destination number, it can be determined whether the called party belongs to a specific group of people. Possible groups include elderly people, nursing homes, public institutions, and / or companies.

[0013] The call information data of the subscriber's connection should also conveniently include the following parameters: • a number of calls to one or more phone numbers, The higher the number of outgoing calls and the more different phone numbers are dialed, the higher the probability that it is an unwanted call. • Target areas of the calls, • Call target areas include, for example, calls to a specific region, a particular geographic location, and / or a specific area code. The more diverse this call pattern, the higher the probability that it is an unwanted call. The further the target areas are from the subscriber's connection, the higher the probability that it is an unwanted call. In the case of landline / VoIP, the geographic location is determined by the address of the called party. • Call duration and / or ◯ The shorter the call duration, the more likely it is to be an unwanted call. • Characteristics of the telephone numbers. ◯ This allows us to determine whether the caller's number is suppressed and / or from which region or country the call is being made.

[0014] Using a combination or individual elements of the call information described above, the algorithm, specifically artificial intelligence (AI), can calculate the probability that a call is unwanted. The call information is weighted differently as a parameter. This weighting is dynamic and can change depending on other parameters. For example, a P-Asserted Identity that does not match the caller's phone number and therefore potentially represents spoofing will be weighted more heavily if the call frequency and destination addresses suggest a spam call, such as all destination addresses being in the same location / region or all recipients being within a specific age quartile. While the P-Asserted Identity parameter may be a significant factor, it is not mandatory.Important: A spam / scam likelihood is calculated based on the parameters mentioned in the comment above, including the call frequency of the A-subscriber number. This calculation triggers technical measures (call rejection, etc.) if a likelihood threshold is exceeded. If this probability exceeds a predefined threshold, an unwanted call is detected, and a corresponding warning can be played to the called party. Another option is to completely or temporarily block the unwanted caller's number for a specific group of people and / or a specific region.

[0015] Preferably, the call monitoring unit is located on a communication link between the caller's and the called party's subscriber lines. This allows for the efficient extraction of call information data without requiring new hardware connections. If the call monitoring unit is located, for example, in the telephone exchange and / or a central location in the communication network (such as the identification server), the communication link will inevitably pass through this central location. However, it can be advantageous to locate such a call monitoring unit as close as possible to the respective subscriber lines or at suitable network nodes.This leads, on the one hand, to an efficient load distribution within the network, and on the other hand, the call monitoring unit can also be equipped with an interruption function that can effectively interrupt the subscriber connection when identifying unwanted calls.

[0016] According to a second aspect of the invention, a communication network for identifying unwanted signals originating from an end device of a subscriber connection in a communication network is specified, comprising: • a terminal device capable of making calls on a subscriber line; ◯ also typically a telephone, especially a smartphone; • an identification server; ◯ with a computing unit for executing algorithms; • Communication links between devices of the communication network, in particular between the calling terminal and the identification server; wherein the identification server is set up to carry out the procedure according to one of the claims described above.

[0017] The communication network thus enables automated detection of unwanted calls without the subscriber, i.e., the called party, having to actively recognize that it is an unwanted call.

[0018] In one embodiment, the communication network includes an interruption unit for selectively interrupting a communication connection. This offers the advantage that a communication connection assigned to a potential fraudster can be selectively interrupted in order to effectively protect the other subscribers / connections from this individual.

[0019] According to a third aspect of the invention, an identification server is specified which is configured to carry out the method described above.

[0020] Further advantageous design features of the present invention are defined in the patent claims.

[0021] Preferred embodiments of the present invention are explained below with reference to the accompanying figures: Fig. Figure 1 shows a communication network according to the invention for detecting unwanted calls.

[0022] Numerous features of the present invention are explained in detail below with reference to preferred embodiments. The present disclosure is not limited to the specific combinations of features mentioned. Rather, the features mentioned here can be combined arbitrarily to form embodiments according to the invention, unless expressly excluded below.

[0023] Fig. Figure 1 shows a communication network 10 for detecting unwanted calls.

[0024] The communication network 10 comprises: an end device 12 of a subscriber line that makes potentially unwanted calls, wherein the end device 12 is connected via communication links 14 to an interface 16 of the communication network 10. This interface 16 of the communication network 10 can, in particular, be provided by a switching center 16. An identification server 18 can also be provided in the switching center 16 or is at least in communication communication with the switching interface 16 via a communication link 14. The identification server 18 can also be provided at a location within the communication network 10 such that it is provided in a communication link 14 between the end device 12 and the other end device 20 of a called party.In such an arrangement, the identification server 18 can intercept call information data between the terminal device 12 of the potential spammer and the other terminal device 20 particularly efficiently.

[0025] An algorithm for detecting unwanted calls is implemented on identification server 18, with the algorithm performing the following steps.

[0026] In step 22, the identification server 18 receives call information data and can evaluate it to determine if it is an unwanted call. This call information data is typically transmitted by terminal device 12 and must be used to establish a connection anyway – for example, the phone number of the other terminal device 20.

[0027] Step 22: Receiving the call information data that characterizes the call made, by a call monitoring unit 34 of the communication network, whereby this function may also be implemented in the identification server 18.

[0028] Step 24: Creating a call pattern based on the call information data using the algorithm;

[0029] Step 26: Comparison of the generated call pattern with defined criteria stored on the identification server for unwanted calls;

[0030] Step 28: Determine if the call is unwanted.

[0031] If the algorithm determines that the call is unwanted with a probability exceeding a predefined threshold, appropriate countermeasures can be initiated in step 30.

[0032] The invention according to the method thus makes it possible to effectively warn and protect subscribers from unwanted calls.

Claims

[1] Method for identifying unwanted calls originating from an end device of a subscriber line in a communications network, comprising the following steps: • Receiving call information data that characterizes the call made, by a call monitoring unit of the communications network; • Analysis of call information data on an identification server, wherein an algorithm based on self-learning artificial intelligence for identifying unwanted calls is implemented on the identification server, the algorithm performing the following steps: ◯ Creating a call pattern based on call information data; ◯ Comparison of the generated call pattern with defined criteria stored on the identification server for unwanted calls; ◯ Determine whether the call is an unwanted call,characterized by that the call information data includes the address of the destination number and / or the age of the destination subscriber. [2] Method according to the preceding claim, characterized by that the call information data is temporarily stored on the identification server for a certain period of time. [3] Method according to any one of the preceding claims, characterized by , that the call information data of the subscriber line includes the following parameters: • a number of calls to one or more phone numbers, • Target areas of the calls, • Call duration and / or • Characteristics of the telephone numbers. [4] Method according to any one of the preceding claims, characterized by that the call monitoring unit is provided on a communication link between the subscriber line of the caller and the called party. [5] Method according to any one of the preceding claims, characterized by, that the algorithm detects an unwanted call when a defined probability threshold has been exceeded. [6] Communication network for identifying unwanted communications originating from an end device of a subscriber connection in a communication network, comprising: • a terminal device capable of making calls on a subscriber line; • an identification server; • Communication links between devices of the communication network, in particular between the calling terminal and the identification server; characterized by that the identification server is set up to carry out the method according to one of claims 1-5. [7] Communication network according to claim 6, characterized by that the communication network has a break unit for interrupting a selective communication link. [8] Identification server set up to identify unwanted calls and to perform the steps of the method according to any one of claims 1-5.

Images