Device and method for machine learning in a telecommunications network based on radio cells

1 . A method for machine learning in a telecommunications network comprising radio cells, the method comprising:

recording a series of observations of a property of at least one signal received over a period of time by a mobile terminal in the telecommunications network;

recording a series of observations of at least one connection handover signal, transmitted over the period of time by a network device in the telecommunications network;

determining, a machine-learned model as a function of both of the recorded series of observations;

determining, using the determined machine-learned model, an estimated value of a handover parameter; and

predicting, based on the estimated value of the handover timing parameter, a time of a connection handover in the telecommunications network in which the mobile terminal switches from one radio cell of the telecommunications network to another radio cell of the telecommunications network during an ongoing connection without interrupting the connection, wherein the connection is a call connection or a data connection.

2 . The method according to claim 1 , wherein a further series of observations of a property of a signal received by a further mobile terminal in the telecommunications network is recorded, wherein the estimated value for the parameter is determined as a function of the further series of observations.

3 . The method according to claim 1 , wherein the model includes a set of parameters, wherein the set of parameters is learned as a function of the series of observations, wherein the estimated value is determined as a function of the set of parameters.

4 . The method according to claim 1 , wherein the estimated value and the parameter characterize a threshold value for a difference between a quality of the connection from the mobile terminal to one of the radio cells and a quality of the connection from the mobile terminal to the other of the radio cells, including a hysteresis margin.

5 . The method according to claim 1 , wherein the estimated value and the parameter characterize a threshold value for a counter for securing a successful connection handover, including a handover failure timer.

6 . The method according to claim 1 , further comprising carrying out earlier instances of the connection handover during the period of time over which both of the series of observations are recorded.

7 . The method according to claim 6 , further comprising modifying the machine-learned model based on further observations of the property of the at least one signal received by the mobile terminal and of the at least one connection handover signal that are recorded after carrying out the instance of the connection handover, over a further period of time.

8 . A method for machine learning in a telecommunications network based on radio cells, the method comprising:

recording a series of observations of a property of a signal received by a mobile terminal in the telecommunications network;

recording a series of observations of a connection handover signal transmitted by a network device in the telecommunications network;

determining a model, wherein the determining includes learning a set of model parameters of the model as a function of both of the series of observations;

estimating an estimated value of a handover parameter using the model as a function of the learned set of model parameters;

predicting, as a function of (i) an observation of the property and (ii) the estimated value, a time of a connection handover in the telecommunications network in which the mobile terminal switches from one radio cell of the telecommunications network to another radio cell of the telecommunications network during a connection without interrupting the connection, wherein the connection is a call connection or a data connection;

carrying out the connection handover as a function of the parameter;

recording a time at which the connection handover is, carried out; and:

modifying the set of parameters as a function of the predicted time and the recorded time.

9 . The method according to claim 8 , wherein: the estimated value and the parameter characterize: (i) a threshold value for a power of the signal including a reference signal received power (RSRP), or (ii) a threshold value for an indicator for a field strength of the signal, in particular a received signal strength indicator (RSSI), or (iii) a threshold value for a ratio value of a value for a power of the signal and an indicator for a field strength of the signal, in particular a reference signal received quality (RSRQ).

10 . A device for machine learning in a telecommunications network based on radio cells, the device comprising:

a computer; and

a computer-readable medium on which are stored instructions that are executable by the computer and that, when executed by the computer, cause the computer to perform a method, the method including the following steps:

recording a series of observations of a property of at least one signal received over a period of time by a mobile terminal in the telecommunications network;

recording a series of observations of at least one connection handover signal, transmitted over the period of time by a network device in the telecommunications network;

determining, a machine-learned model as a function of both of the recorded series of observations;

determining, using the determined machine-learned model, an estimated value of a handover parameter; and

predicting, based on the estimated value of the handover timing parameter, a time of a connection handover in the telecommunications network in which the mobile terminal switches from one radio cell of the telecommunications network to another radio cell of the telecommunications network during an ongoing connection without interrupting the connection, wherein the connection is a call connection or a data connection.

11 . The device according to claim 10 , wherein the device further comprises:

at least one interface configured to receive signals which are transmitted in the telecommunications network by the mobile terminal or the network device.

12 . A non-transitory machine readable medium on which is stored a computer program including machine-readable instructions, the instructions, when executed by a computer, causing the computer to perform the following steps:

recording a series of observations of a property of at least one signal received over a period of time by a mobile terminal in the telecommunications network;

recording a series of observations of at least one connection handover signal, transmitted over the period of time by a network device in the telecommunications network;

determining, a machine-learned model as a function of both of the recorded series of observations;

determining, using the determined machine-learned model, an estimated value of a handover parameter; and

predicting, based on the estimated value of the handover timing parameter, a time of a connection handover in the telecommunications network in which the mobile terminal switches from one radio cell of the telecommunications network to another radio cell of the telecommunications network during an ongoing connection without interrupting the connection, wherein the connection is a call connection or a data connection.

13 . A method for machine learning in a telecommunications network based on radio cells, the method comprising:

recording a series of observations of a property of a signal received by a mobile terminal in the telecommunications network;

recording a series of observations of a connection handover signal transmitted by a network device in the telecommunications network;

determining a model, wherein the determining includes learning a set of model parameters of the model as a function of both of the series of observations;

estimating an estimated value of a handover parameter using the model as a function of the learned set of model parameters;

predicting, as a function of (i) an observation of the property and (ii) the estimated value, whether there will occur a connection handover, in the telecommunications network and as a function of the handover parameter, in which the mobile terminal switches from one radio cell of the telecommunications network to another radio cell of the telecommunications network during a connection without interrupting the connection, wherein the connection is a call connection or a data connection;

recording whether the connection handover actually occurs; and

modifying the set of parameters as a function of the prediction and the record of whether the connection handover actually occurred.

14 . The method according to claim 13 , wherein: the estimated value and the parameter characterize: (i) a threshold value for a power of the signal including a reference signal received power (RSRP), or (ii) a threshold value for an indicator for a field strength of the signal, in particular a received signal strength indicator (RSSI), or (iii) a threshold value for a ratio value of a value for a power of the signal and an indicator for a field strength of the signal, in particular a reference signal received quality (RSRQ).