DETAILED ACTION
This communication is in response to the Application No. 18/473,512 filed September 25, 2023
in which Claims 1 - 20 are presented for examination.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claim 1-20 are rejected under 35 U.S.C. 101 because these claimed inventions are directed to an
abstract idea without significantly more.
Regarding Claim 1:
Step 1: Claim 1 is a method type claim. Therefore, Claims 1-9 fall within one of the four statutory categories (i.e., process, machine, manufacture, or composition of matter).
2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance
of the limitation in the mind but for the recitation of generic computer components, then it falls within
the “Mental Processes” grouping of abstract ideas. If a claim limitation, under its broadest reasonable
interpretation, covers performance of the limitation by mathematical calculation but for the recitation
of generic computer components, then it falls within the “Mathematical Concepts” grouping of abstract
ideas.
Performing […] a task […] to obtain an output, wherein the output is configured to be used for performance of a network operation of the communication network (mental process – performing a task to obtain an output may be performed mentally by a user observing/analyzing the task and accordingly using judgment/evaluation to obtain an output based on said analysis)
and determining, based on the feedback data, to perform at least one of the following: (mental process – determining to perform may be performed mentally by a user observing/analyzing the feedback data and using judgment/evaluation to decide which listed action should be performed)
updating at least one parameter of a non-machine learning algorithm associated with performance of the task […] (mental process – updating the parameter may be performed mentally/manually by a user using judgment/evaluation to adjust the parameter associated with performance of the task)
refraining from […](mental process – refraining from an action may be performed mentally/manually by a user using judgment/evaluation to decide not to perform the action)
or refraining from updating the at least one parameter of the non-machine learning algorithm (mental process – refraining from updating the parameter may be performed mentally/manually by a user using judgment/evaluation to decide not to change the parameter of the non-machine learning algorithm)
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
[…] by a device associated with a communication network […] (recited at a high-level of generality (i.e., a device associated with a communication network, a generic processor, computer-readable storage medium, and memory) such that it amounts to no more than mere instructions to apply the exception using generic computer components)
[…] with a machine learning model […] (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model without significantly more)
receiving, from an access node of the communication network, feedback data indicative of a cause of a failure of the network operation (Adding insignificant extra-solution activity to the judicial exception - see MPEP 2106.05(g))
re-training the machine learning model for performing the task (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
[…] with the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model without significantly more)
[…] re-training the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
Step 2B: The claim does not include additional elements considered individually and in combination that
are sufficient to amount to significantly more than the judicial exception.
[…] by a device associated with a communication network […] (recited at a high-level of generality (i.e., a device associated with a communication network, a generic processor, computer-readable storage medium, and memory) such that it amounts to no more than mere instructions to apply the exception using generic computer components)
[…] with a machine learning model […] (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model without significantly more)
receiving, from an access node of the communication network, feedback data indicative of a cause of a failure of the network operation (MPEP 2106.05(d)(II) indicates that merely “Receiving or transmitting data over a network” is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer)
re-training the machine learning model for performing the task (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
[…] with the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model without significantly more)
[…] re-training the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
For the reasons above, Claim 1 is rejected as being directed to an abstract idea without
significantly more. This rejection applies equally to dependent claims 2 - 9. The additional limitations of the dependent claims are addressed below.
Regarding Claim 2:
Step 2A Prong 1: See the rejection of Claim 1 above, which Claim 2 depends on.
[…] or updating the at least one parameter of the non-machine learning model, in response to determining that the source of the failure is the device (mental process – updating the parameter may be performed mentally/manually by a user using judgment/evaluation after determining that the source of the failure is the device)
refraining from […] from updating the at least one parameter of the non-machine learning model, in response to determining that the source of the failure is not the device (mental process – refraining from updating the parameter may be performed mentally/manually by a user using judgment/evaluation after determining that the source of the failure is not the device)
Step 2A Prong 2 & Step 2B:
re-training the machine learning model for the task […] (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
[…] re-training the machine learning model […] (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 1. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 3:
Step 2A Prong 1: See the rejection of Claim 2 above, which Claim 3 depends on.
suspending, based on the feedback data, inference […] (mental process – suspending inference may be performed mentally/manually by a user observing/analyzing the feedback data and using judgment/evaluation to decide not to continue the inference)
and performing the task with a second non-machine learning algorithm […] (mental process – performing the task with a second non-machine learning algorithm may be performed mentally/manually by a user using judgment/evaluation to apply a different non-machine learning algorithm to the task)
Step 2A Prong 2 & Step 2B:
[…] with the machine learning model during the re-training of the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation applying/using a machine learning model without significantly more)
[…] during the re- training of the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 2. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 4:
Step 2A Prong 1: See the rejection of Claim 3 above, which Claim 4 depends on.
Step 2A Prong 2 & Step 2B:
wherein the inference with the machine learning model is suspended, in response to receiving a predetermined number of feedback messages indicative of the device as the source of the failure (Field of Use – limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the exception itself, and cannot integrate a judicial exception into a practical application; in this case specifying that the inference with the machine learning model is suspended, in response to receiving a predetermined number of feedback messages indicative of the device as the source of the failure does not integrate the exception into a practical application nor amount to significantly more – See MPEP 2106.05(h))
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 3. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 5:
Step 2A Prong 1: See the rejection of Claim 1 above, which Claim 5 depends on.
Step 2A Prong 2 & Step 2B:
transmitting a request for the feedback data to the access node (Adding insignificant extra-solution activity to the judicial exception & MPEP 2106.05(d)(II) indicates that merely “Receiving or transmitting data over a network” is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer)
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 1. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 6:
Step 2A Prong 1: See the rejection of Claim 1 above, which Claim 6 depends on.
Step 2A Prong 2 & Step 2B:
wherein the network operation comprises a handover of the device (Field of Use – limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the exception itself, and cannot integrate a judicial exception into a practical application; in this case specifying that the network operation comprises a handover of the device does not integrate the exception into a practical application nor amount to significantly more – See MPEP 2106.05(h))
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 1. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 7:
Step 2A Prong 1: See the rejection of Claim 6 above, which Claim 7 depends on.
Step 2A Prong 2 & Step 2B:
wherein the feedback data is received from a target access node of the handover (Field of Use – limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the exception itself, and cannot integrate a judicial exception into a practical application; in this case specifying that the feedback data is received from a target access node of the handover does not integrate the exception into a practical application nor amount to significantly more – See MPEP 2106.05(h))
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 6. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 8:
Step 2A Prong 1: See the rejection of Claim 6 above, which Claim 8 depends on.
determining a time for initiating the handover (mental process – determining the time may be performed mentally/manually by a user using judgment/evaluation to decide when the handover should be initiated)
or determining an identifier of the target access node of the handover (mental process - determining the identifier may be performed mentally/manually by a user using judgment/evaluation to identify or select the target access node for the handover)
Step 2A Prong 2 & Step 2B:
Accordingly, under Step 2A Prong 2 and Step 2B, there are no additional elements that integrate the abstract idea into practical application. The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 9:
Step 2A Prong 1: See the rejection of Claim 6 above, which Claim 9 depends on.
Step 2A Prong 2 & Step 2B:
wherein the feedback data comprises an indication of at least one of: the handover having been initiated too early by the device, the handover having been initiated too late by the device, the handover having been initiated at a substantially correct time by the device, a time interval between reception of a handover measurement report by a source access node of the handover and initiation of handover preparation of the target access node by the source access node, an identifier of at least one access node that has rejected the handover of the device, a reason for the rejection of the handover by the at least one access node, a duration of a handover preparation phase, or a time interval used by the source access node for preparation of one or more target cells for the handover (Field of Use – limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the exception itself, and cannot integrate a judicial exception into a practical application; in this case specifying that the feedback data comprises handover related information, such as timing indications, handover preparation intervals, rejected access node identifiers, rejection reasons, and preparation phase durations does not integrate the exception into a practical application nor amount to significantly more – See MPEP 2106.05(h))
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 6. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 10:
Step 1: Claim 10 is a method type claim. Therefore, Claims 10-13 fall within one of the four statutory categories (i.e., process, machine, manufacture, or composition of matter).
2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance
of the limitation in the mind but for the recitation of generic computer components, then it falls within
the “Mental Processes” grouping of abstract ideas. If a claim limitation, under its broadest reasonable
interpretation, covers performance of the limitation by mathematical calculation but for the recitation
of generic computer components, then it falls within the “Mathematical Concepts” grouping of abstract
ideas.
[…] to perform a task […] (mental process - performing a task may be performed mentally/manually by a user using judgment/evaluation to carry out the task)
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
obtaining, by an access node of a communication network, feedback data indicative of a cause of a failure of a network operation of the communication network, […] (Adding insignificant extra-solution activity to the judicial exception - see MPEP 2106.05(g))
[…] wherein performance of the network operation is based on an output of a machine learning model configured […] (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model without significantly more)
[…] at a device associated with the communication network (recited at a high-level of generality (i.e., a device associated with a communication network, a generic processor, computer-readable storage medium, and memory) such that it amounts to no more than mere instructions to apply the exception using generic computer components)
transmitting, by the access node, the feedback data to the device (Adding insignificant extra-solution activity to the judicial exception - see MPEP 2106.05(g))
Step 2B: The claim does not include additional elements considered individually and in combination that
are sufficient to amount to significantly more than the judicial exception.
obtaining, by an access node of a communication network, feedback data indicative of a cause of a failure of a network operation of the communication network, […] ((MPEP 2106.05(d)(II) indicates that merely “Receiving or transmitting data over a network” is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer)
[…] wherein performance of the network operation is based on an output of a machine learning model configured […] (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model without significantly more)
[…] at a device associated with the communication network (recited at a high-level of generality (i.e., a device associated with a communication network, a generic processor, computer-readable storage medium, and memory) such that it amounts to no more than mere instructions to apply the exception using generic computer components)
transmitting, by the access node, the feedback data to the device ((MPEP 2106.05(d)(II) indicates that merely “Receiving or transmitting data over a network” is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer)
For the reasons above, Claim 10 is rejected as being directed to an abstract idea without
significantly more. This rejection applies equally to dependent claims 11-13. The additional limitations of the dependent claims are addressed below.
Regarding Claim 11:
Step 2A Prong 1: See the rejection of Claim 10 above, which Claim 11 depends on.
Step 2A Prong 2 & Step 2B:
wherein the network operation comprises a handover of the device (Field of Use – limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the exception itself, and cannot integrate a judicial exception into a practical application; in this case specifying that the network operation comprises a handover of the device does not integrate the exception into a practical application nor amount to significantly more – See MPEP 2106.05(h))
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 10. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 12:
Step 2A Prong 1: See the rejection of Claim 11 above, which Claim 12 depends on.
Step 2A Prong 2 & Step 2B:
obtaining the feedback data based on reception of a user equipment context of the device from a source access node of the handover, […](Adding insignificant extra-solution activity to the judicial exception & MPEP 2106.05(d)(II) indicates that merely “Receiving or transmitting data over a network” is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer)
[…] wherein the access node comprises a target access node of the handover (Field of Use – limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the exception itself, and cannot integrate a judicial exception into a practical application; in this case specifying the access node comprises a target access node of the handover does not integrate the exception into a practical application nor amount to significantly more – See MPEP 2106.05(h))
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 11. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 13:
Step 2A Prong 1: See the rejection of Claim 10 above, which Claim 13 depends on.
Step 2A Prong 2 & Step 2B:
obtaining the feedback data based on reception of a user equipment context of the device from a last serving access node of the device, in response to connection re-establishment of the device with the access node (Adding insignificant extra-solution activity to the judicial exception & MPEP 2106.05(d)(II) indicates that merely “Receiving or transmitting data over a network” is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer)
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 10. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 14:
Step 1: Claim 14 is a method type claim. Therefore, Claims 14-20 fall within one of the four statutory categories (i.e., process, machine, manufacture, or composition of matter).
2A Prong 1: If a claim limitation, under its broadest reasonable interpretation, covers performance
of the limitation in the mind but for the recitation of generic computer components, then it falls within
the “Mental Processes” grouping of abstract ideas. If a claim limitation, under its broadest reasonable
interpretation, covers performance of the limitation by mathematical calculation but for the recitation
of generic computer components, then it falls within the “Mathematical Concepts” grouping of abstract
ideas.
Performing […] a task […] to obtain an output, wherein the output is configured to be used for performance of a network operation of the communication network (mental process – performing a task to obtain an output may be performed mentally by a user observing/analyzing the task and accordingly using judgment/evaluation to obtain an output based on said analysis)
and determining, based on the feedback data, to perform at least one of the following: (mental process – determining to perform may be performed mentally by a user observing/analyzing the feedback data and using judgment/evaluation to decide which listed action should be performed)
updating at least one parameter of a non-machine learning algorithm associated with performance of the task […] (mental process – updating the parameter may be performed mentally/manually by a user using judgment/evaluation to adjust the parameter associated with performance of the task)
refraining from […](mental process – refraining from an action may be performed mentally/manually by a user using judgment/evaluation to decide not to perform the action)
or refraining from updating the at least one parameter of the non-machine learning algorithm (mental process – refraining from updating the parameter may be performed mentally/manually by a user using judgment/evaluation to decide not to change the parameter of the non-machine learning algorithm)
Step 2A Prong 2: This judicial exception is not integrated into a practical application.
At least one processor; at least one memory […] (recited at a high-level of generality (i.e., a device associated with a communication network, a generic processor, computer-readable storage medium, and memory) such that it amounts to no more than mere instructions to apply the exception using generic computer components)
[…] by a device associated with a communication network […] (recited at a high level of generality (i.e., a device associated with a communication network, a generic processor, computer-readable storage medium, and memory) such that it amounts to no more than mere instructions to apply the exception using generic computer components)
[…] with a machine learning model […] (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model without significantly more)
receiving, from an access node of the communication network, feedback data indicative of a cause of a failure of the network operation (Adding insignificant extra-solution activity to the judicial exception - see MPEP 2106.05(g))
re-training the machine learning model for performing the task (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
[…] with the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model without significantly more)
[…] re-training the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
Step 2B: The claim does not include additional elements considered individually and in combination that
are sufficient to amount to significantly more than the judicial exception.
[…] by a device associated with a communication network […] (recited at a high-level of generality (i.e., a device associated with a communication network, a generic processor, computer-readable storage medium, and memory) such that it amounts to no more than mere instructions to apply the exception using generic computer components)
[…] with a machine learning model […] (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model without significantly more)
receiving, from an access node of the communication network, feedback data indicative of a cause of a failure of the network operation (MPEP 2106.05(d)(II) indicates that merely “Receiving or transmitting data over a network” is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer)
re-training the machine learning model for performing the task (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
[…] with the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of applying/using a machine learning model without significantly more)
[…] re-training the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
For the reasons above, Claim 14 is rejected as being directed to an abstract idea without
significantly more. This rejection applies equally to dependent claims 15 - 20. The additional limitations of the dependent claims are addressed below.
Regarding Claim 15:
Step 2A Prong 1: See the rejection of Claim 14 above, which Claim 15 depends on.
[…] or updating the at least one parameter of the non-machine learning model, in response to determining that the source of the failure is the device (mental process – updating the parameter may be performed mentally/manually by a user using judgment/evaluation after determining that the source of the failure is the device)
refraining from […] from updating the at least one parameter of the non-machine learning model, in response to determining that the source of the failure is not the device (mental process – refraining from updating the parameter may be performed mentally/manually by a user using judgment/evaluation after determining that the source of the failure is not the device)
Step 2A Prong 2 & Step 2B:
re-training the machine learning model for the task […] (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
[…] re-training the machine learning model […] (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 14. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 16:
Step 2A Prong 1: See the rejection of Claim 15 above, which Claim 16 depends on.
suspending, based on the feedback data, inference […] (mental process – suspending inference may be performed mentally/manually by a user observing/analyzing the feedback data and using judgment/evaluation to decide not to continue the inference)
and performing the task with a second non-machine learning algorithm […] (mental process – performing the task with a second non-machine learning algorithm may be performed mentally/manually by a user using judgment/evaluation to apply a different non-machine learning algorithm to the task)
Step 2A Prong 2 & Step 2B:
[…] with the machine learning model during the re-training of the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation applying/using a machine learning model without significantly more)
[…] during the re- training of the machine learning model (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f) – Examiner’s note: high level recitation of re-training a machine learning model without significantly more)
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 15. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 17:
Step 2A Prong 1: See the rejection of Claim 16 above, which Claim 17 depends on.
Step 2A Prong 2 & Step 2B:
wherein the inference with the machine learning model is suspended, in response to receiving a predetermined number of feedback messages indicative of the device as the source of the failure (Field of Use – limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the exception itself, and cannot integrate a judicial exception into a practical application; in this case specifying that the inference with the machine learning model is suspended, in response to receiving a predetermined number of feedback messages indicative of the device as the source of the failure does not integrate the exception into a practical application nor amount to significantly more – See MPEP 2106.05(h))
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 16. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 18:
Step 2A Prong 1: See the rejection of Claim 14 above, which Claim 18 depends on.
Step 2A Prong 2 & Step 2B:
transmitting a request for the feedback data to the access node (Adding insignificant extra-solution activity to the judicial exception & MPEP 2106.05(d)(II) indicates that merely “Receiving or transmitting data over a network” is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is in the present claim). Thereby, a conclusion that the claimed limitation is well-understood, routine, conventional activity is supported under Berkheimer)
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 14. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 19:
Step 2A Prong 1: See the rejection of Claim 14 above, which Claim 19 depends on.
Step 2A Prong 2 & Step 2B:
wherein the network operation comprises a handover of the device (Field of Use – limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the exception itself, and cannot integrate a judicial exception into a practical application; in this case specifying that the network operation comprises a handover of the device does not integrate the exception into a practical application nor amount to significantly more – See MPEP 2106.05(h))
Accordingly, under Step 2A Prong 2 and Step 2B, this additional element does not integrate the
abstract idea into practical application because it does not impose any meaningful limits on practicing
the abstract idea, as discussed above in the rejection of claim 14. The claim does not include additional
elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Regarding Claim 20:
Step 2A Prong 1: See the rejection of Claim 19 above, which Claim 20 depends on.
determining a time for initiating the handover (mental process – determining the time may be performed mentally/manually by a user using judgment/evaluation to decide when the handover should be initiated)
or determining an identifier of the target access node of the handover (mental process - determining the identifier may be performed mentally/manually by a user using judgment/evaluation to identify or select the target access node for the handover)
Step 2A Prong 2 & Step 2B:
Accordingly, under Step 2A Prong 2 and Step 2B, there are no additional elements that integrate the abstract idea into practical application. The claim does not include additional elements considered individually and in combination that are sufficient to amount to significantly more than the judicial exception.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 5-14, and 18-20 are rejected under 35 U.S.C. 103 as being unpatentable over Yuxin
et al. (hereafter Yuxin) (WO 2021123285) in view of Stanislav et al. (hereinafter Stanislav) (EP 4319302).
Regarding Claim 1, Yuxin teaches a method (Yuxin, Page 3 – line 2, “a method”, thus a method is disclosed), comprising:
performing, by a device associated with a communication network, a task with a machine learning model to obtain an output, wherein the output is configured to be used for performance of a network operation of the communication network (Yuxin, Page 1 – Abstract, “A method of transmitting or receiving data by a communications device in a wireless communications network, the method comprising: establishing a connection for transmitting or receiving the data in a first cell of the wireless communications network, determining a value of one or more input parameters, using the value of the one or more input parameters as inputs to a model trained using machine learning, determining, based on an output of the model, that the communications device should perform a handover to establish a connection in a second cell, and responsive to determining that the communications device should establish a connection in the second cell, transmitting a handover message to request the establishment of a connection in a second cell”, thus performing, by a device associated with a communication network, a task with a machine learning model to obtain an output, wherein the output is configured to be used for performance of a network operation of the communication network is disclosed, because Yuxin teaches a communications device operating in a wireless communications network that uses input parameters as inputs to a model trained using machine learning. The communications device corresponds to the device associated with a communication network, and the wireless communications network corresponds to the communication network. The use of input parameters with the machine-learning model corresponds to performing the task with the machine learning model. The model’s output is used to determine whether the communications device should perform a handover, so the handover determination corresponds to the output. The handover to establish a connection in a second cell corresponds to the network operation, and the transmission of a handover message shows that the output is used to perform that network operation)
[…] the communication network […] the network operation (Yuxin, Page 3 – line 4, “receiving the data in a first cell of the wireless communications network, determining a value of one or more input parameters, using the value of the one or more input parameters as inputs to a model trained using machine learning, determining, based on an output of the model, that the communications device should perform a handover to establish a connection in a second cell”, thus […] the communication network […] the network operation is disclosed, because Yuxin teaches that the communications device receives data in a first cell of a wireless communications network and uses input parameters with a machine-learning model to determine, based on the model output, that the communications device should perform a handover to establish a connection in a second cell. The wireless communications network corresponds to the communication network. The handover from the first cell to the second cell corresponds to the network operation. Yuxin teaches that the model output is used for the network operation because the output determines whether the communications device should perform the handover)
and determining […] to perform at least one of the following: re-training the machine learning model for performing the task, updating at least one parameter of a non-machine learning algorithm associated with performance of the task with the machine learning model, refraining from re-training the machine learning model, or refraining from updating the at least one parameter of the non-machine learning algorithm (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus determining […] to perform one of the following: re-training the machine learning model for performing the task, updating at least one parameter of a non-machine learning algorithm associated with performance of the task with the machine learning model, refraining from re-training the machine learning model, or refraining from updating the at least one parameter of the non-machine learning algorithm is disclosed, because Yuxin teaches that the communications device may dynamically update the machine-learning model based on previous handover attempts. The communications device stores the input parameters that triggered each handover and an output value based on the outcome of the handover attempt. Those stored parameter values and outcome-based output values are then used to update the model by adjusting one or more weights to reduce the loss function. Therefore, Yuxin teaches determining to update or re-train the machine-learning model based on handover-result information, where the model being updated corresponds to the machine learning model for performing the task, and the handover attempt outcome corresponds to information used to determine whether the model should be updated)
Yuxin does not explicitly teach receiving from an access node of […], feedback data indicative of a cause of a failure of […] and […determining…], based on the feedback data, […to perform…].
However, Stanislav teaches:
receiving, from an access node of […], feedback data indicative of a cause of a failure of […] (Stanislav, Page 22 – line 33, “The RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate the unintended event that has occurred during the handover. Further, the RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate that a normal HO has been performed. The unintended event may include a Too Late Handover, a Too Early Handover, and a Handover to wrong cell”, thus receiving, from an access node of […], feedback data indicative of a cause of a failure of […] is disclosed, because Stanislav teaches that RAN node 3 transmits a HANDOVER REPORT message to RAN node 2 to indicate an unintended event that occurred during handover. RAN node 3 corresponds to the access node. The HANDOVER REPORT message corresponds to the feedback data. The unintended event corresponds to the cause of the failure, and the Too Late Handover, Too Early Handover, and Handover to wrong cell events correspond to specific causes of failure)
[…determining…], based on the feedback data, […to perform…] (Stanislav, Page 35 – line 41, “The RAN node 2 receives, from another RAN node, information related to the HO containing feedback from an unintentional HO event (i.e., a negative HO event) and a normal (intended) HO event (i.e., a positive HO event). Therefore, in the steps S1001 to S1006, the RAN node 2 updates the AI model by making it perform reinforcement learning based on the information received from the other apparatus”, thus […determining…], based on the feedback data, […to perform…] is disclosed, because Stanislav teaches that RAN node 2 receives handover-related information containing feedback from an unintentional handover event and a normal handover event, and updates the AI model by performing reinforcement learning based on that received information. The handover-related information containing feedback corresponds to the feedback data. Updating the AI model based on the received information corresponds to determining, based on the feedback data, to perform the update action)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to combine Yuxin’s teaching of a communications device using a machine learning model to determine handover with Stanislav’s teaching of handover feedback and AI based handover optimization. Yuxin teaches using a model trained using machine learning to determine, based on an output of the model, that the communications device should perform a handover, and further teaches dynamically updating the model based on previous handover attempts. Stanislav teaches receiving handover related feedback from unintended and normal handover events and updating an AI model based on that feedback. Therefore, a POSITA would have been motivated to incorporate Stanislav’s handover feedback and AI updating techniques into Yuxin’s ML based handover system to optimize handover configuration information, identify cells having a high probability of handover success or failure, and improve handover performance (Stanislav, Page 32 – line 3, “Further, the RAN node 2 can transmit assistance information related to the handover to the RAN node 3 and the UE 6 by performing the above described procedure. The RAN node 3 and the UE 6 can update HO configuration information based on the HO configuration information received from the RAN node 2, and therefore can optimize the HO configuration information. Further, the RAN node 3 and the UE 6 can determine a cell having a high probability of succeeding in a handover or a cell having a high probability of failing in a handover based on the assistance information received from the RAN node 2, and therefore can improve the handover performance”)
Regarding Claim 5, Yuxin combined with Stanislav teaches all the limitations of claim 1 as cited above and Stanislav further teaches:
transmitting a request for the feedback data to the access node (Stanislav, Page 25 – line 1, “In a step S121, the RAN node 2 transmits an HO REPORT REQUEST message requesting the transmission of the information related to the HO from the RAN node 3 to the RAN node 3. The step S121 is performed after the management relation described in the fifth example embodiment is established between the RAN node 2, which is the AI-enhanced RAN node, and the RAN node 3. In a step S122, when the aforementioned management relation has already been established between the RAN nodes 2 and 3, the RAN node 3 accepts the request from the RAN node 2 and transmits an HO REPORT RESPONSE message to the RAN node 2”, thus transmitting a request for the feedback data to the access node is disclosed, because Stanislav teaches that RAN node 2 transmits an HO REPORT REQUEST message to RAN node 3 requesting transmission of information related to the handover. The HO REPORT REQUEST message corresponds to the request for the feedback data. RAN node 3 corresponds to the access node. The information related to the handover corresponds to the feedback data)
Regarding Claim 6, Yuxin combined with Stanislav teaches all the limitations of claim 1 as cited above and Yuxin further teaches:
wherein the network operation comprises a handover of the device (Yuxin, Page 1 – Abstract, “using the value of the one or more input parameters as inputs to a model trained using machine learning, determining, based on an output of the model, that the communications device should perform a handover to establish a connection in a second cell”, thus wherein the network operation comprises a handover of the device is disclosed, because Yuxin teaches that the communications device determines, based on an output of the model, that the communications device should perform a handover to establish a connection in a second cell. The communications device corresponds to the device. The handover to establish a connection in the second cell corresponds to the network operation comprising a handover of the device)
Regarding Claim 7, Yuxin combined with Stanislav teaches all the limitations of claim 6 as cited above and Stanislav further teaches:
wherein the feedback data is received from a target access node of the handover (Stanislav, Page 23 – line 34, “As shown on the right side of the sequence diagram shown in Fig. 17, the RAN node 4, which is the source RAN node, the RAN node 3, which is the target RAN node, and the UE 6 performs a Handover preparation procedure. After that, the RAN node 4, the RAN node 3, and the UE 6 perform a Handover Execution procedure, and lastly performs a Handover Completion procedure”, thus wherein the feedback data is received from a target access node of the handover is disclosed, because Stanislav teaches that RAN node 3 is the target RAN node in the handover involving UE 6. RAN node 3 corresponds to the target access node of the handover. The HANDOVER REPORT message transmitted by RAN node 3 corresponds to the feedback data received from the target access node)
Regarding Claim 8, Yuxin combined with Stanislav teaches all the limitations of claim 6 as cited above and Yuxin further teaches:
determining a time for initiating the handover, or determining an identifier of the target access node of the handover (Yuxin, Page 12 – line 15, “The output may comprise an indication of a single target cell. In some embodiments, the output may comprise an indication of one or more candidate cells and an indication of respective associated priorities and/or respective associated probabilities. The associated priorities may correspond to a ranking of the cells, such that a cell with higher ranking is to be preferred for selection as a target cell”, & Page 12 – line 25, “Based on the output, a determination is made as to whether a handover should be performed, and in some embodiments, which of a plurality of candidate cells should be the target cell in which a connection is to be established following (or as part of) the handover procedure”, thus determining a time for initiating the handover, or determining an identifier of the target access node of the handover is disclosed, because Yuxin teaches that the model output may indicate a single target cell or one or more candidate cells with associated priorities or probabilities. Yuxin further teaches that, based on the output, a determination is made as to whether a handover should be performed, and which candidate cell should be the target cell. The target cell corresponds to the target access node of the handover. The indication of the single target cell or candidate target cell corresponds to determining an identifier of the target access node of the handover)
Regarding Claim 9, Yuxin combined with Stanislav teaches all the limitations of claim 6 as cited above and Stanislav further teaches:
wherein the feedback data comprises an indication of at least one of: the handover having been initiated too early by the device, the handover having been initiated too late by the device, the handover having been initiated at a substantially correct time by the device, a time interval between reception of a handover measurement report by a source access node of the handover and initiation of handover preparation of the target access node by the source access node, an identifier of at least one access node that has rejected the handover of the device, a reason for the rejection of the handover by the at least one access node, a duration of a handover preparation phase, or a time interval used by the source access node for preparation of one or more target cells for the handover (Stanislav, Page 24 – line 31, “The RAN node 3 may determine whether the handover performed by the UE 6 is a normal handover (a normal HO) or a handover based on an unintended event that has occurred during the HO, and set a result of the determination in the first IE. Note that similarly to the first operation example, when information indicating that the handover is a handover based on an unintended event is set in the first IE, the first IE may indicate which of a Too Late Handover, a Too Early Handover, and a Handover to wrong cell the unintended event corresponds to”, thus wherein the feedback data comprises an indication of at least one of: the handover having been initiated too early by the device, the handover having been initiated too late by the device, the handover having been initiated at a substantially correct time by the device, a time interval between reception of a handover measurement report by a source access node of the handover and initiation of handover preparation of the target access node by the source access node, an identifier of at least one access node that has rejected the handover of the device, a reason for the rejection of the handover by the at least one access node, a duration of a handover preparation phase, or a time interval used by the source access node for preparation of one or more target cells for the handover is disclosed, because Stanislav teaches that RAN node 3 determines whether the handover performed by UE 6 is a normal handover or a handover based on an unintended event. The information indicating whether the handover is normal or based on an unintended event corresponds to the feedback data. The Too Late Handover indication corresponds to the handover having been initiated too late by the device. The Too Early Handover indication corresponds to the handover having been initiated too early by the device. The normal handover indication corresponds to the handover having been initiated at a substantially correct time by the device)
Regarding Claim 10, Yuxin teaches a method (Yuxin, Page 3 – line 2, “a method”, thus a method is disclosed), comprising:
[…] a communication network[…] a network operation of the communication network, wherein performance of the network operation is based on an output of a machine learning model configured to perform a task at a device associated with the communication network (Yuxin, Page 1 – Abstract, “A method of transmitting or receiving data by a communications device in a wireless communications network, the method comprising: establishing a connection for transmitting or receiving the data in a first cell of the wireless communications network, determining a value of one or more input parameters, using the value of the one or more input parameters as inputs to a model trained using machine learning, determining, based on an output of the model, that the communications device should perform a handover to establish a connection in a second cell, and responsive to determining that the communications device should establish a connection in the second cell, transmitting a handover message to request the establishment of a connection in a second cell”, thus […] a communication network […] a network operation of the communication network, wherein performance of the network operation is based on an output of a machine learning model configured to perform a task at a device associated with the communication network is disclosed, because Yuxin teaches a communications device in a wireless communications network using input parameters as inputs to a model trained using machine learning. Yuxin’s communications device corresponds to the device associated with the communication network, and the wireless communications network corresponds to the communication network. The model trained using machine learning corresponds to the machine learning model configured to perform the task at the device. The model output is used to determine that the communications device should perform a handover to establish a connection in a second cell. Therefore, the handover corresponds to the network operation, and performance of the handover is based on the output of the machine learning model)
Yuxin does not explicitly teach obtaining, by an access node of […] , feedback data indicative of a cause of a failure of […] and transmitting, by the access node, the feedback data to the device.
However, Stanislav teaches:
obtaining, by an access node of […] , feedback data indicative of a cause of a failure of […] , (Stanislav, Page 11 – line 18, “The UE 6 may transmit a Measurement Report containing the cell quality information to the RAN node 3, which serves the cell 5-1, after the UE 6 has performed the handover from the cell 5-2 to the cell 5-1 “, & Page 22 – line 33, “The RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate the unintended event that has occurred during the handover. Further, the RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate that a normal HO has been performed. The unintended event may include a Too Late Handover, a Too Early Handover, and a Handover to wrong cell”, obtaining, by an access node of […], feedback data indicative of a cause of a failure of […] is disclosed, because Stanislav teaches that UE 6 transmits a Measurement Report to RAN node 3 after handover, and RAN node 3 transmits a HANDOVER REPORT message to RAN node 2 indicating an unintended event that occurred during handover. RAN node 3 corresponds to the access node. The Measurement Report and HANDOVER REPORT message correspond to the feedback data. The unintended event corresponds to the cause of the failure, and the Too Late Handover, Too Early Handover, and Handover to wrong cell events correspond to specific causes of failure)
and transmitting, by the access node, the feedback data to the device (Stanislav, Page 25 – line 1, “In a step S121, the RAN node 2 transmits an HO REPORT REQUEST message requesting the transmission of the information related to the HO from the RAN node 3 to the RAN node 3. The step S121 is performed after the management relation described in the fifth example embodiment is established between the RAN node 2, which is the AI-enhanced RAN node, and the RAN node 3. In a step S122, when the aforementioned management relation has already been established between the RAN nodes 2 and 3, the RAN node 3 accepts the request from the RAN node 2 and transmits an HO REPORT RESPONSE message to the RAN node 2”, & Page 37 – line 17, “Specifically, in a step S1009, the RAN node 2 transmits at least one of HO configuration information containing an updated value(s) of an HO parameter(s) and assistance information output by the AI model to the RAN node 3 and the UE 6”, thus transmitting, by the access node, the feedback data to the device is disclosed, because Stanislav teaches that RAN node 3 transmits an HO REPORT RESPONSE message containing information related to the handover to RAN node 2, and then RAN node 2 transmits HO configuration information and assistance information output by the AI model to UE 6. The HO REPORT RESPONSE message and information related to the handover correspond to the feedback data. RAN node 2 corresponds to the access node that transmits the handover related feedback/assistance information. UE 6 corresponds to the device. Therefore, Stanislav discloses transmitting handover related feedback information by an access node to the device)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to combine Yuxin’s teaching of a communications device using a machine learning model to determine handover with Stanislav’s teaching of handover feedback and AI based handover optimization. Yuxin teaches using a model trained using machine learning to determine, based on an output of the model, that the communications device should perform a handover. Stanislav teaches obtaining handover related feedback, including feedback indicating unintended handover events, and using that information in an AI based handover optimization process. Stanislav further teaches transmitting HO configuration information and assistance information output by the AI model to the UE. Therefore, a POSITA would have been motivated to incorporate Stanislav’s handover feedback and assistance information transmission into Yuxin’s ML based handover system so that the device may receive handover related feedback or assistance information based on handover results. This would allow the device to update or adjust handover operation using information identifying handover success or failure conditions, optimize HO configuration information, identify cells having a high probability of handover success or failure, and improve handover performance (Stanislav, Page 32 – line 3, “Further, the RAN node 2 can transmit assistance information related to the handover to the RAN node 3 and the UE 6 by performing the above described procedure. The RAN node 3 and the UE 6 can update HO configuration information based on the HO configuration information received from the RAN node 2, and therefore can optimize the HO configuration information. Further, the RAN node 3 and the UE 6 can determine a cell having a high probability of succeeding in a handover or a cell having a high probability of failing in a handover based on the assistance information received from the RAN node 2, and therefore can improve the handover performance”)
Regarding Claim 11, Yuxin combined with Stanislav teaches all the limitations of claim 10 as cited above and Yuxin further teaches:
wherein the network operation comprises a handover of the device (Yuxin, Page 1 – Abstract, “using the value of the one or more input parameters as inputs to a model trained using machine learning, determining, based on an output of the model, that the communications device should perform a handover to establish a connection in a second cell”, thus wherein the network operation comprises a handover of the device is disclosed, because Yuxin teaches that the communications device determines, based on an output of the model, that the communications device should perform a handover to establish a connection in a second cell. The communications device corresponds to the device. The handover to establish a connection in the second cell corresponds to the network operation comprising a handover of the device)
Regarding Claim 12, Yuxin combined with Stanislav teaches all the limitations of claim 11 as cited above and Stanislav further teaches:
obtaining the feedback data […] wherein the access node comprises a target access node of […] (Stanislav, Page 10 – line 30, “The UE 6 performs a handover from the cell 5-2 served by the RAN node 4 to the cell 5-1 served by the RAN node 3. The cell 5-2 may be referred to as a source cell, and the cell 5-1 may be referred to as a target cell. The RAN node 4 may be referred to as a source RAN node, and the RAN node 3 may be referred to as a target RAN node. The cell 5-1 is a serving cell after the handover, and the cell 5-2 may be referred to as a last serving cell”, & Page 22 – line 33, “The RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate the unintended event that has occurred during the handover. Further, the RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate that a normal HO has been performed. The unintended event may include a Too Late Handover, a Too Early Handover, and a Handover to wrong cell”, thus obtaining the feedback data […] wherein the access node comprises a target access node of […] is disclosed, because Stanislav teaches that UE 6 performs a handover from cell 5-2 served by RAN node 4 to cell 5-1 served by RAN node 3, where RAN node 4 is the source RAN node and RAN node 3 is the target RAN node. RAN node 3 corresponds to the target access node. Stanislav further teaches that RAN node 3 transmits a HANDOVER REPORT message to RAN node 2 to indicate an unintended event that occurred during handover. The HANDOVER REPORT message corresponds to the feedback data, and the unintended event corresponds to the failure-related handover information obtained by the target access node)
Yuxin further teaches […] based on reception of a user equipment context of the device from a source access node of the handover […] the handover (Yuxin, Page 16 – line 24, “Context information associated with the communications device 270 may be transmitted by the source infrastructure equipment 272 to the target infrastructure equipment 372. This may be as part of step S704 or may be earlier or later”, thus […] based on reception of a user equipment context of the device from a source access node of the handover […] the handover is disclosed, because Yuxin teaches that context information associated with the communications device may be transmitted from the source infrastructure equipment to the target infrastructure equipment. The context information associated with the communications device corresponds to the user equipment context of the device. The source infrastructure equipment corresponds to the source access node of the handover. The target infrastructure equipment corresponds to the target access node that receives the user equipment context during the handover procedure)
Regarding Claim 13, Yuxin combined with Stanislav teaches all the limitations of claim 10 as cited above and Stanislav further teaches:
obtaining the feedback data […] from a last serving access node […], in response to connection re-establishment […] with the access node (Stanislav, Page 10 – line 30, “The cell 5-2 may be referred to as a source cell, and the cell 5-1 may be referred to as a target cell. The RAN node 4 may be referred to as a source RAN node, and the RAN node 3 may be referred to as a target RAN node. The cell 5-1 is a serving cell after the handover, and the cell 5-2 may be referred to as a last serving cell”, & Page 21 – line 16, “Note that the measurement report that the UE 6 transmits after the handover thereof may also be referred to as a UE HO report. Note that the UE 6 may transmit the cell quality information (the cell reference signal measurement information) to the RAN node 3 by using one or more measurement reports” & Page 32 – line 28, “An event in which a radio link failure (RLF) occurs after a UE has stayed for a long period of time in the cell, and the UE attempts to re-establish the radio link connection in a different cell. Intra-system Too Early Handover: An event in which an RLF occurs shortly after a successful handover from a source cell to a target cell or a handover failure occurs during the handover procedure, and the UE attempts to re-establish the radio link connection in the source cell. Intra-system Handover to Wrong Cell”, thus obtaining the feedback data […] from a last serving access node […], in response to connection re-establishment […] with the access node is disclosed, because Stanislav teaches that cell 5-2 is the last serving cell and RAN node 4 is the source RAN node. Stanislav further teaches that UE 6 transmits a Measurement Report, also referred to as a UE HO report, after handover, and that the UE attempts to re-establish the radio link connection after an RLF or handover failure. The last serving cell/source RAN node corresponds to the last serving access node. The Measurement Report or UE HO report corresponds to the feedback data. The UE attempting to re-establish the radio link connection corresponds to connection re-establishment with the access node)
Yuxin further teaches […] based on reception of a user equipment context of the device […] of the device […] of the device […] (Yuxin, Page 16 – line 24, “Context information associated with the communications device 270 may be transmitted by the source infrastructure equipment 272 to the target infrastructure equipment 372. This may be as part of step S704 or may be earlier or later”, thus […] based on reception of a user equipment context of the device […] of the device […] of the device […] is disclosed, because Yuxin teaches that context information associated with the communications device may be transmitted from the source infrastructure equipment to the target infrastructure equipment. The context information associated with the communications device corresponds to the user equipment context of the device. The communications device corresponds to the device. The source infrastructure equipment corresponds to the access node from which the user equipment context is received, and the target infrastructure equipment corresponds to the access node receiving the user equipment context)
Regarding Claim 14, Yuxin teaches an apparatus comprising:
at least one processor (Yuxin, Page 28 – line 6, “a processor”, thus a processor is disclosed); and at least one memory (Yuxin, Page 19 – line 4, “non-volatile memory”, thus at least one memory is disclosed) storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform:
performing a task with a machine learning model to obtain an output, wherein the output is configured to be used for performance of a network operation of the communication network (Yuxin, Page 1 – Abstract, “A method of transmitting or receiving data by a communications device in a wireless communications network, the method comprising: establishing a connection for transmitting or receiving the data in a first cell of the wireless communications network, determining a value of one or more input parameters, using the value of the one or more input parameters as inputs to a model trained using machine learning, determining, based on an output of the model, that the communications device should perform a handover to establish a connection in a second cell, and responsive to determining that the communications device should establish a connection in the second cell, transmitting a handover message to request the establishment of a connection in a second cell”, thus performing, by a device associated with a communication network, a task with a machine learning model to obtain an output, wherein the output is configured to be used for performance of a network operation of the communication network is disclosed, because Yuxin teaches a communications device operating in a wireless communications network that uses input parameters as inputs to a model trained using machine learning. The wireless communications network corresponds to the communication network. The use of input parameters with the machine-learning model corresponds to performing the task with the machine learning model. The model’s output is used to determine whether the communications device should perform a handover, so the handover determination corresponds to the output. The handover to establish a connection in a second cell corresponds to the network operation, and the transmission of a handover message shows that the output is used to perform that network operation)
[…] the communication network […] the network operation (Yuxin, Page 3 – line 4, “receiving the data in a first cell of the wireless communications network, determining a value of one or more input parameters, using the value of the one or more input parameters as inputs to a model trained using machine learning, determining, based on an output of the model, that the communications device should perform a handover to establish a connection in a second cell”, thus […] the communication network […] the network operation is disclosed, because Yuxin teaches that the communications device receives data in a first cell of a wireless communications network and uses input parameters with a machine-learning model to determine, based on the model output, that the communications device should perform a handover to establish a connection in a second cell. The wireless communications network corresponds to the communication network. The handover from the first cell to the second cell corresponds to the network operation. Yuxin teaches that the model output is used for the network operation because the output determines whether the communications device should perform the handover)
and determining […] to perform at least one of the following: re-training the machine learning model for performing the task, updating at least one parameter of a non-machine learning algorithm associated with performance of the task with the machine learning model, refraining from re-training the machine learning model, or refraining from updating the at least one parameter of the non-machine learning algorithm (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus determining […] to perform one of the following: re-training the machine learning model for performing the task, updating at least one parameter of a non-machine learning algorithm associated with performance of the task with the machine learning model, refraining from re-training the machine learning model, or refraining from updating the at least one parameter of the non-machine learning algorithm is disclosed, because Yuxin teaches that the communications device may dynamically update the machine-learning model based on previous handover attempts. The communications device stores the input parameters that triggered each handover and an output value based on the outcome of the handover attempt. Those stored parameter values and outcome-based output values are then used to update the model by adjusting one or more weights to reduce the loss function. Therefore, Yuxin teaches determining to update or re-train the machine-learning model based on handover-result information, where the model being updated corresponds to the machine learning model for performing the task, and the handover attempt outcome corresponds to information used to determine whether the model should be updated)
Yuxin does not explicitly teach receiving from an access node of […], feedback data indicative of a cause of a failure of […] and […determining…], based on the feedback data, […to perform…].
However, Stanislav teaches:
receiving, from an access node of […], feedback data indicative of a cause of a failure of […] (Stanislav, Page 22 – line 33, “The RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate the unintended event that has occurred during the handover. Further, the RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate that a normal HO has been performed. The unintended event may include a Too Late Handover, a Too Early Handover, and a Handover to wrong cell”, thus receiving, from an access node of […], feedback data indicative of a cause of a failure of […] is disclosed, because Stanislav teaches that RAN node 3 transmits a HANDOVER REPORT message to RAN node 2 to indicate an unintended event that occurred during handover. RAN node 3 corresponds to the access node. The HANDOVER REPORT message corresponds to the feedback data. The unintended event corresponds to the cause of the failure, and the Too Late Handover, Too Early Handover, and Handover to wrong cell events correspond to specific causes of failure)
[…determining…], based on the feedback data, […to perform…] (Stanislav, Page 35 – line 41, “The RAN node 2 receives, from another RAN node, information related to the HO containing feedback from an unintentional HO event (i.e., a negative HO event) and a normal (intended) HO event (i.e., a positive HO event). Therefore, in the steps S1001 to S1006, the RAN node 2 updates the AI model by making it perform reinforcement learning based on the information received from the other apparatus”, thus […determining…], based on the feedback data, […to perform…] is disclosed, because Stanislav teaches that RAN node 2 receives handover-related information containing feedback from an unintentional handover event and a normal handover event, and updates the AI model by performing reinforcement learning based on that received information. The handover-related information containing feedback corresponds to the feedback data. Updating the AI model based on the received information corresponds to determining, based on the feedback data, to perform the update action)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to combine Yuxin’s teaching of a communications device using a machine learning model to determine handover with Stanislav’s teaching of handover feedback and AI based handover optimization. Yuxin teaches using a model trained using machine learning to determine, based on an output of the model, that the communications device should perform a handover, and further teaches dynamically updating the model based on previous handover attempts. Stanislav teaches receiving handover related feedback from unintended and normal handover events and updating an AI model based on that feedback. Therefore, a POSITA would have been motivated to incorporate Stanislav’s handover feedback and AI updating techniques into Yuxin’s ML based handover system to optimize handover configuration information, identify cells having a high probability of handover success or failure, and improve handover performance (Stanislav, Page 32 – line 3, “Further, the RAN node 2 can transmit assistance information related to the handover to the RAN node 3 and the UE 6 by performing the above described procedure. The RAN node 3 and the UE 6 can update HO configuration information based on the HO configuration information received from the RAN node 2, and therefore can optimize the HO configuration information. Further, the RAN node 3 and the UE 6 can determine a cell having a high probability of succeeding in a handover or a cell having a high probability of failing in a handover based on the assistance information received from the RAN node 2, and therefore can improve the handover performance”)
Regarding Claim 18, Yuxin combined with Stanislav teaches all the limitations of claim 14 as cited above and Stanislav further teaches:
transmitting a request for the feedback data to the access node (Stanislav, Page 25 – line 1, “In a step S121, the RAN node 2 transmits an HO REPORT REQUEST message requesting the transmission of the information related to the HO from the RAN node 3 to the RAN node 3. The step S121 is performed after the management relation described in the fifth example embodiment is established between the RAN node 2, which is the AI-enhanced RAN node, and the RAN node 3. In a step S122, when the aforementioned management relation has already been established between the RAN nodes 2 and 3, the RAN node 3 accepts the request from the RAN node 2 and transmits an HO REPORT RESPONSE message to the RAN node 2”, thus transmitting a request for the feedback data to the access node is disclosed, because Stanislav teaches that RAN node 2 transmits an HO REPORT REQUEST message to RAN node 3 requesting transmission of information related to the handover. The HO REPORT REQUEST message corresponds to the request for the feedback data. RAN node 3 corresponds to the access node. The information related to the handover corresponds to the feedback data)
Regarding Claim 19, Yuxin combined with Stanislav teaches all the limitations of claim 14 as cited above and Yuxin further teaches:
wherein the network operation comprises a handover of the device (Yuxin, Page 1 – Abstract, “using the value of the one or more input parameters as inputs to a model trained using machine learning, determining, based on an output of the model, that the communications device should perform a handover to establish a connection in a second cell”, thus wherein the network operation comprises a handover of the device is disclosed, because Yuxin teaches that the communications device determines, based on an output of the model, that the communications device should perform a handover to establish a connection in a second cell. The communications device corresponds to the device. The handover to establish a connection in the second cell corresponds to the network operation comprising a handover of the device)
Regarding Claim 20, Yuxin combined with Stanislav teaches all the limitations of claim 19 as cited above and Yuxin further teaches:
determining a time for initiating the handover, or determining an identifier of the target access node of the handover (Yuxin, Page 12 – line 15, “The output may comprise an indication of a single target cell. In some embodiments, the output may comprise an indication of one or more candidate cells and an indication of respective associated priorities and/or respective associated probabilities. The associated priorities may correspond to a ranking of the cells, such that a cell with higher ranking is to be preferred for selection as a target cell”, & Page 12 – line 25, “Based on the output, a determination is made as to whether a handover should be performed, and in some embodiments, which of a plurality of candidate cells should be the target cell in which a connection is to be established following (or as part of) the handover procedure”, thus determining a time for initiating the handover, or determining an identifier of the target access node of the handover is disclosed, because Yuxin teaches that the model output may indicate a single target cell or one or more candidate cells with associated priorities or probabilities. Yuxin further teaches that, based on the output, a determination is made as to whether a handover should be performed, and which candidate cell should be the target cell. The target cell corresponds to the target access node of the handover. The indication of the single target cell or candidate target cell corresponds to determining an identifier of the target access node of the handover)
Claims 2 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Yuxin et al. (hereafter Yuxin) (WO 2021123285) in view of Stanislav et al. (hereinafter Stanislav) (EP 4319302), and further in view of Adolfo et al. (hereinafter Adolfo) (WO 2017025773) and Fuchs et al. (hereinafter Fuchs) (US 11176677).
Regarding Claim 2, Yuxin combined with Stanislav teaches all the limitations of claim 1 as cited above and Yuxin further teaches:
re-training the machine learning model for the task or updating the at least one parameter of the non-machine learning model, […] the device (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus re-training the machine learning model for the task or updating the at least one parameter of the non-machine learning model, […] is disclosed, because Yuxin teaches that the communications device may dynamically update the model based on previous handover attempts. Yuxin’s model corresponds to the machine learning model for performing the task. The previous handover attempts correspond to prior performance of the network operation. The stored input parameters and outcome-based output values are used to update the model by adjusting weights to reduce the loss function, which corresponds to re-training or updating the machine learning model for the task)
[…retraining…] the machine learning model […or updating…] the at least one parameter of the non-machine learning model, […] the device (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus […retraining…] the machine learning model […or updating…] the at least one parameter of the non-machine learning model, […] the device is disclosed, because Yuxin teaches that the communications device dynamically updates the model based on previous handover attempts. Yuxin’s communications device corresponds to the device. Yuxin’s model corresponds to the machine learning model. The stored input parameters and output value based on the outcome of each handover attempt are used to update the model by adjusting one or more weights to reduce the loss function. Adjusting the model weights based on prior handover-attempt outcomes corresponds to retraining or updating the machine learning model)
Yuxin combined with Stanislav does not explicitly teach […] in response to determining that the source of the failure is [… the device…], refraining from re-training […a machine learning model…] or from updating […], and in response to determining that the source of the failure is not [….the device…].
However, Adolfo teaches:
[…] in response to determining that the source of the failure is […device…] (Adolfo, Page 4 – line 19, “an individual call's radio channel signaling and KPIs from transmitters and receivers, and is used by RCA Diagnosis unit 320 to accurately determine a root cause of the failure using the FSM methodology described above. Advanced signaling call flow 315 may also be used to determine whether the failure is in the DL or UL direction. Thus, the methodology described herein allows the detection of the exact time when every call failed from the radio point of view, what the radio resource management procedure state was at the time of the failure, the UE and eNB Call KPIs at the exact time when every call failed, the eNB Cell KPIs at the exact time when every call failed, information provided by the internal and external events before the call failed. With this information, the RCA diagnosis algorithm may provide”, thus […] in response to determining that the source of the failure is […device…] is disclosed, because Adolfo teaches using radio channel signaling and KPIs from transmitters and receivers to determine a root cause of the failure. Adolfo further teaches determining whether the failure is in the DL or UL direction and using UE Call KPIs, eNB Call KPIs, and eNB Cell KPIs at the time of the failure. The root cause of the failure corresponds to the source of the failure. The UE Call KPIs correspond to information associated with the device. Therefore, Adolfo discloses determining whether the source of the failure is associated with the device)
in response to determining that the source of the failure is not […. the device…] (Adolfo, Page 4 – line 19, “an individual call's radio channel signaling and KPIs from transmitters and receivers, and is used by RCA Diagnosis unit 320 to accurately determine a root cause of the failure using the FSM methodology described above. Advanced signaling call flow 315 may also be used to determine whether the failure is in the DL or UL direction. Thus, the methodology described herein allows the detection of the exact time when every call failed from the radio point of view, what the radio resource management procedure state was at the time of the failure, the UE and eNB Call KPIs at the exact time when every call failed, the eNB Cell KPIs at the exact time when every call failed, information provided by the internal and external events before the call failed. With this information, the RCA diagnosis algorithm may provide”, thus in response to determining that the source of the failure is not [… the device …] is disclosed, because Adolfo teaches determining the root cause of a failure using radio channel signaling, KPIs from transmitters and receivers, and advanced signaling call flow. Adolfo further teaches using UE Call KPIs, eNB Call KPIs, and eNB Cell KPIs at the time of failure. The root cause of the failure corresponds to the source of the failure. The eNB Call KPIs, eNB Cell KPIs, DL/UL direction, and radio resource management procedure state corresponds to information used to determine that the source of the failure is not the device)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to further combine Yuxin and Stanislav with Adolfo. Yuxin teaches dynamically updating a machine learning handover model based on previous handover attempts. Stanislav teaches using handover feedback from unintended and normal handover events to update an AI model and improve handover performance. Adolfo teaches that root cause analysis can be applied in mobile communication systems, including RAN optimization activities, and that improved root cause determination is needed for problematic calls in mobile communication systems. Therefore, a POSITA would have been motivated to incorporate Adolfo’s root cause analysis into the ML based handover optimization system of Yuxin and Stanislav to determine whether the source of a handover or call failure is the device or another network side condition before deciding whether to retrain or update the model. This would improve the accuracy of failure diagnosis, support RAN optimization, reduce unnecessary tuning and optimization costs, and help network providers maintain service performance (Adolfo, Page 4 – line 19, “As another example, the embodiments described herein may reduce initial tuning and optimization services cost, while allowing network providers to maintain excellence in the service. As yet another example, the various embodiments described herein may allow for optimization services using industrial calls trace analysis, and thereby may advantageously reduce or eliminate the need for probes, which can be expensive and generate large and hardly manageable amounts of data for customers”, & Page 5 – line 17, “The concept of RCA can be applied in mobile communication systems. One such case involves the application of RCA in Radio Access Network (RAN) optimization activities. Optimization activities may be of particular concern in certain types of mobile communications, such as, for example, Long Term Evolution (LTE) and/or Voice over LTE (VoLTE) networks. As described above, there is a need for an improved method of determining the root cause of problematic calls in a mobile communications system”)
Yuxin and Stanislav combined with Adolfo does not explicitly teach refraining from, re-training […a machine learning model…] or from updating […].
However, Fuchs teaches:
and refraining from, re-training […a machine learning model…] or from updating […] (Fuchs, Par. [0068], “When the indicator indicates that the image segmentation model 504 is not to be retrained, the feedback handler 508 in conjunction with the model trainer 506 may refrain from retraining of the image segmentation model 504. In some embodiments, the feedback handler 508 may also maintain the parameters of the image segmentation model 504. In some embodiments, the feedback handler 508 in conjunction with the model trainer 506 may also determine to terminate retraining of the image segmentation model”, thus and refraining from re-training […a machine learning model…] or from updating […] is disclosed, because Fuchs teaches that when an indicator shows the image segmentation model is not to be retrained, the feedback handler and model trainer may refrain from retraining the image segmentation model. Fuchs further teaches maintaining the parameters of the image segmentation model and terminating retraining. The image segmentation model corresponds to the machine learning model. Refraining from retraining the image segmentation model corresponds to refraining from re-training the machine learning model)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to further combine Yuxin, Stanislav, and Adolfo with Fuchs. Yuxin teaches dynamically updating a machine learning handover model based on previous handover attempts. Stanislav teaches using handover feedback from unintended and normal handover events to update an AI model and improve handover performance. Adolfo teaches determining the root cause of mobile communication failures so the system can identify whether the failure source is the device or another network condition. Fuchs further teaches refraining from retraining a machine learning model and maintaining model parameters when feedback indicates that retraining is not needed. Therefore, a POSITA would have been motivated to incorporate Fuchs’s refraining from retraining technique into the ML based handover optimization system of Yuxin, Stanislav, and Adolfo so that, after Adolfo’s root cause analysis determines that the failure source is not the device, the system would avoid unnecessary retraining or parameter updating. This would reduce unnecessary model training, shorten training time, reduce the amount of additional training data or feedback needed, and improve system efficiency while maintaining the existing model parameters when retraining is not beneficial. This is supported by Fuchs’s teaching that DIaL minimizes annotation time by iteratively correcting only mislabeled regions and that using fewer sample images allows the model to be trained over a shorter time period with fewer annotations and less time spent preparing annotations (Fuchs, Par. [0028], “Deep Interactive Learning (DIaL) may be used to minimize pathologists' annotation time by iteratively annotating mislabeled regions to improve a model. DIaL may be used with a pretrained model from a different cancer type to reduce manual training annotation on pancreatic pathology images”, & Par. [0070], “Because less sample images 518 are used, the image segmentation model 504 may be trained over a shorter time period than using a larger training dataset to train. In addition, with less sample images 518, the number of annotations 522 may be lessened and the time spent in preparing the annotations 522 may be reduced”)
Regarding Claim 15, Yuxin combined with Stanislav teaches all the limitations of claim 14 as cited above and Yuxin further teaches:
re-training the machine learning model for the task or updating the at least one parameter of the non-machine learning model, […] the device (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus re-training the machine learning model for the task or updating the at least one parameter of the non-machine learning model, […] is disclosed, because Yuxin teaches that the communications device may dynamically update the model based on previous handover attempts. Yuxin’s model corresponds to the machine learning model for performing the task. The previous handover attempts correspond to prior performance of the network operation. The stored input parameters and outcome-based output values are used to update the model by adjusting weights to reduce the loss function, which corresponds to re-training or updating the machine learning model for the task)
[…retraining…] the machine learning model […or updating…] the at least one parameter of the non-machine learning model, […] the device (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus […retraining…] the machine learning model […or updating…] the at least one parameter of the non-machine learning model, […] the device is disclosed, because Yuxin teaches that the communications device dynamically updates the model based on previous handover attempts. Yuxin’s communications device corresponds to the device. Yuxin’s model corresponds to the machine learning model. The stored input parameters and output value based on the outcome of each handover attempt are used to update the model by adjusting one or more weights to reduce the loss function. Adjusting the model weights based on prior handover-attempt outcomes corresponds to retraining or updating the machine learning model)
Yuxin combined with Stanislav does not explicitly teach […] in response to determining that the source of the failure is [… the device…], refraining from re-training […a machine learning model…] or from updating […], and in response to determining that the source of the failure is not [….the device…].
However, Adolfo teaches:
[…] in response to determining that the source of the failure is […device…] (Adolfo, Page 4 – line 19, “an individual call's radio channel signaling and KPIs from transmitters and receivers, and is used by RCA Diagnosis unit 320 to accurately determine a root cause of the failure using the FSM methodology described above. Advanced signaling call flow 315 may also be used to determine whether the failure is in the DL or UL direction. Thus, the methodology described herein allows the detection of the exact time when every call failed from the radio point of view, what the radio resource management procedure state was at the time of the failure, the UE and eNB Call KPIs at the exact time when every call failed, the eNB Cell KPIs at the exact time when every call failed, information provided by the internal and external events before the call failed. With this information, the RCA diagnosis algorithm may provide”, thus […] in response to determining that the source of the failure is […device…] is disclosed, because Adolfo teaches using radio channel signaling and KPIs from transmitters and receivers to determine a root cause of the failure. Adolfo further teaches determining whether the failure is in the DL or UL direction and using UE Call KPIs, eNB Call KPIs, and eNB Cell KPIs at the time of the failure. The root cause of the failure corresponds to the source of the failure. The UE Call KPIs correspond to information associated with the device. Therefore, Adolfo discloses determining whether the source of the failure is associated with the device)
in response to determining that the source of the failure is not […. the device…] (Adolfo, Page 4 – line 19, “an individual call's radio channel signaling and KPIs from transmitters and receivers, and is used by RCA Diagnosis unit 320 to accurately determine a root cause of the failure using the FSM methodology described above. Advanced signaling call flow 315 may also be used to determine whether the failure is in the DL or UL direction. Thus, the methodology described herein allows the detection of the exact time when every call failed from the radio point of view, what the radio resource management procedure state was at the time of the failure, the UE and eNB Call KPIs at the exact time when every call failed, the eNB Cell KPIs at the exact time when every call failed, information provided by the internal and external events before the call failed. With this information, the RCA diagnosis algorithm may provide”, thus in response to determining that the source of the failure is not [… the device …] is disclosed, because Adolfo teaches determining the root cause of a failure using radio channel signaling, KPIs from transmitters and receivers, and advanced signaling call flow. Adolfo further teaches using UE Call KPIs, eNB Call KPIs, and eNB Cell KPIs at the time of failure. The root cause of the failure corresponds to the source of the failure. The eNB Call KPIs, eNB Cell KPIs, DL/UL direction, and radio resource management procedure state corresponds to information used to determine that the source of the failure is not the device)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to further combine Yuxin and Stanislav with Adolfo. Yuxin teaches dynamically updating a machine learning handover model based on previous handover attempts. Stanislav teaches using handover feedback from unintended and normal handover events to update an AI model and improve handover performance. Adolfo teaches that root cause analysis can be applied in mobile communication systems, including RAN optimization activities, and that improved root cause determination is needed for problematic calls in mobile communication systems. Therefore, a POSITA would have been motivated to incorporate Adolfo’s root cause analysis into the ML based handover optimization system of Yuxin and Stanislav to determine whether the source of a handover or call failure is the device or another network side condition before deciding whether to retrain or update the model. This would improve the accuracy of failure diagnosis, support RAN optimization, reduce unnecessary tuning and optimization costs, and help network providers maintain service performance (Adolfo, Page 4 – line 19, “As another example, the embodiments described herein may reduce initial tuning and optimization services cost, while allowing network providers to maintain excellence in the service. As yet another example, the various embodiments described herein may allow for optimization services using industrial calls trace analysis, and thereby may advantageously reduce or eliminate the need for probes, which can be expensive and generate large and hardly manageable amounts of data for customers”, & Page 5 – line 17, “The concept of RCA can be applied in mobile communication systems. One such case involves the application of RCA in Radio Access Network (RAN) optimization activities. Optimization activities may be of particular concern in certain types of mobile communications, such as, for example, Long Term Evolution (LTE) and/or Voice over LTE (VoLTE) networks. As described above, there is a need for an improved method of determining the root cause of problematic calls in a mobile communications system”)
Yuxin and Stanislav combined with Adolfo does not explicitly teach refraining from, re-training […a machine learning model…] or from updating […].
However, Fuchs teaches:
and refraining from, re-training […a machine learning model…] or from updating […] (Fuchs, Par. [0068], “When the indicator indicates that the image segmentation model 504 is not to be retrained, the feedback handler 508 in conjunction with the model trainer 506 may refrain from retraining of the image segmentation model 504. In some embodiments, the feedback handler 508 may also maintain the parameters of the image segmentation model 504. In some embodiments, the feedback handler 508 in conjunction with the model trainer 506 may also determine to terminate retraining of the image segmentation model”, thus and refraining from re-training […a machine learning model…] or from updating […] is disclosed, because Fuchs teaches that when an indicator shows the image segmentation model is not to be retrained, the feedback handler and model trainer may refrain from retraining the image segmentation model. Fuchs further teaches maintaining the parameters of the image segmentation model and terminating retraining. The image segmentation model corresponds to the machine learning model. Refraining from retraining the image segmentation model corresponds to refraining from re-training the machine learning model)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to further combine Yuxin, Stanislav, and Adolfo with Fuchs. Yuxin teaches dynamically updating a machine learning handover model based on previous handover attempts. Stanislav teaches using handover feedback from unintended and normal handover events to update an AI model and improve handover performance. Adolfo teaches determining the root cause of mobile communication failures so the system can identify whether the failure source is the device or another network condition. Fuchs further teaches refraining from retraining a machine learning model and maintaining model parameters when feedback indicates that retraining is not needed. Therefore, a POSITA would have been motivated to incorporate Fuchs’s refraining from retraining technique into the ML based handover optimization system of Yuxin, Stanislav, and Adolfo so that, after Adolfo’s root cause analysis determines that the failure source is not the device, the system would avoid unnecessary retraining or parameter updating. This would reduce unnecessary model training, shorten training time, reduce the amount of additional training data or feedback needed, and improve system efficiency while maintaining the existing model parameters when retraining is not beneficial. This is supported by Fuchs’s teaching that DIaL minimizes annotation time by iteratively correcting only mislabeled regions and that using fewer sample images allows the model to be trained over a shorter time period with fewer annotations and less time spent preparing annotations (Fuchs, Par. [0028], “Deep Interactive Learning (DIaL) may be used to minimize pathologists' annotation time by iteratively annotating mislabeled regions to improve a model. DIaL may be used with a pretrained model from a different cancer type to reduce manual training annotation on pancreatic pathology images”, & Par. [0070], “Because less sample images 518 are used, the image segmentation model 504 may be trained over a shorter time period than using a larger training dataset to train. In addition, with less sample images 518, the number of annotations 522 may be lessened and the time spent in preparing the annotations 522 may be reduced”)
Claims 3-4 and 16-17 are rejected under 35 U.S.C. 103 as being unpatentable over
Yuxin et al. (hereafter Yuxin) (WO 2021123285) in view of Stanislav et al. (hereinafter Stanislav) (EP 4319302), and further in view of Adolfo et al. (hereinafter Adolfo) (WO 2017025773), Fuchs et al. (hereinafter Fuchs) (US 11176677), and Phan et al. (hereinafter Phan), a non-patent literature reference titled “Neural Simplex Architecture”).
Regarding Claim 3, Yuxin, Stanislav, and Adolfo combined with Fuchs teaches all the limitations of claim 2 as cited above and Yuxin further teaches:
[...] the machine learning model during the re-training of the machine learning model (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus […] the machine learning model during the re-training of the machine learning model is disclosed, because Yuxin teaches that the communications device updates the model based on previous handover attempts. Yuxin’s model corresponds to the machine learning model. The stored input parameters and output value based on the outcome of each handover attempt are used to update the model by adjusting one or more weights to reduce the loss function. Adjusting the model weights based on handover-attempt outcomes corresponds to retraining the machine learning model)
performing the task [….] the re-training of the machine learning model (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus performing the task [….] the re-training of the machine learning model is disclosed, because Yuxin teaches that the communications device dynamically updates the model based on previous handover attempts. Yuxin’s communications device performs the handover-related task using the machine-learning model. Yuxin’s model corresponds to the machine learning model. The stored input parameters and outcome-based output values from previous handover attempts are used to update the model by adjusting one or more weights to reduce the loss function. Updating the model weights based on handover attempt outcomes corresponds to re-training the machine learning model for performing the task)
Stanislav further teaches […] based on the feedback data […] (Stanislav, Page 22 – line 33, “The RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate the unintended event that has occurred during the handover. Further, the RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate that a normal HO has been performed. The unintended event may include a Too Late Handover, a Too Early Handover, and a Handover to wrong cell”, thus […] based on the feedback data […] is disclosed, because Stanislav teaches that RAN node 3 transmits a HANDOVER REPORT message to RAN node 2 to indicate an unintended event that occurred during handover. The HANDOVER REPORT message corresponds to the feedback data. The unintended event corresponds to the information used as the basis for the later action, and the Too Late Handover, Too Early Handover, and Handover to wrong cell events correspond to specific feedback information identifying the handover event)
Yuxin, Stanislav, and Adolfo combined with Fuchs does not explicitly teach suspending […] inference with […. a machine learning model during retraining…].
However, Phan teaches:
suspending […] inference with […. a machine learning model during retraining…] (Phan, Page 2 – Section 1, “We address two limitations of the traditional Simplex approach, namely lack of established guidelines for switching control back to the AC so that mission completion can be attained; and lack of techniques for correcting the AC’s behavior after a failover to the BC, so that reverse switching makes sense in the first place”, & Page 5 – Section 3, “Adaptation and Retraining. The AM is used to retrain the NC in an online manner while the BC is in control of the plant (due to NC-to-BC failover). The main purpose of this retraining is to make the NC less likely to trigger the FSC, thereby allowing it to remain in control for longer periods of time, thereby improving overall system performance”, thus suspending […] inference with [… a machine learning model during retraining…] is disclosed, because Phan teaches correcting the neural controller’s behavior after failover to the baseline controller and retraining the neural controller while the baseline controller is in control. The neural controller corresponds to the machine learning model. The baseline controller being in control due to neural controller to baseline controller failover corresponds to suspending use of the machine learning model. Retraining the neural controller online corresponds to retraining the machine learning model. Therefore, Phan discloses suspending inference with the machine learning model during retraining by switching control from the neural controller to the baseline controller while the neural controller is retrained)
and […] with a second non-machine learning algorithm during [… a retraining of a machine learning model…] (Phan, Page 5 – Section 3, “The AM is used to retrain the NC in an online manner while the BC is in control of the plant (due to NC-to-BC failover). The main purpose of this retraining is to make the NC less likely to trigger the FSC, thereby allowing it to remain in control for longer periods of time, thereby improving overall system performance. Techniques that we consider for online retraining of the NC include supervised learning and reinforcement learning. In supervised learning, state-action pairs of the form (s,a) are required for training purposes. The training algorithm uses these examples to teach the NC safe behavior. The control inputs produced by the BC can be used as training samples, although this will train the NC to imitate BC’s behavior, which may lead to a loss in performance”, thus and […] with a second non-machine learning algorithm during [… a retraining of a machine learning model…] is disclosed, because Phan teaches that the adaptation module retrains the neural controller online while the baseline controller is in control of the plant due to neural controller to baseline controller failover. The neural controller corresponds to the machine learning model. The online retraining of the neural controller corresponds to the retraining of the machine learning model. The baseline controller corresponds to the second non-machine learning algorithm)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to further combine Yuxin, Stanislav, Adolfo, and Fuchs with Phan. Yuxin teaches retraining or updating the machine learning model based on previous handover attempts. Stanislav teaches using feedback data from handover events as a basis for updating an AI model. Adolfo teaches determining the source of a communication failure, and Fuchs teaches refraining from retraining when retraining is not needed. Phan further teaches suspending use of a neural controller during retraining and performing the controlled task with a baseline controller while the neural controller is being retrained. Therefore, a POSITA would have been motivated to incorporate Phan’s failover and retraining technique into the ML-based handover system of Yuxin, Stanislav, Adolfo, and Fuchs so that, when the feedback data indicates that the machine learning model should be retrained, the system can suspend inference with the machine learning model during the re-training of the machine learning model and continue performing the task with a second non-machine learning algorithm during the re-training of the machine learning model. This would allow the network operation to continue using a reliable fallback algorithm while the machine learning model is corrected, would make future failover less likely, and would improve overall system performance (Phan, Page 5 – Section 3, “Adaptation and Retraining. The AM is used to retrain the NC in an online manner while the BC is in control of the plant (due to NC-to-BC failover). The main purpose of this retraining is to make the NC less likely to trigger the FSC, thereby allowing it to remain in control for longer periods of time, thereby improving overall system performance)
Regarding Claim 4, Yuxin, Stanislav, Adolfo, and Fuchs combined with Phan teaches all the limitations of claim 3 as cited above and Phan further teaches:
wherein the inference with […a machine learning…] is suspended, […] (Phan, Page 5 – Section 3, “Adaptation and Retraining. The AM is used to retrain the NC in an online manner while the BC is in control of the plant (due to NC-to-BC failover). The main purpose of this retraining is to make the NC less likely to trigger the FSC, thereby allowing it to remain in control for longer periods of time, thereby improving overall system performance”, thus wherein the inference with […a machine learning…] is suspended, […] is disclosed, because Phan teaches that the adaptation module retrains the neural controller online while the baseline controller is in control of the plant due to neural controller to baseline controller failover. The neural controller corresponds to the machine learning model. The neural-controller-to-baseline-controller failover corresponds to suspending inference or use of the machine learning model. The baseline controller while the neural controller is retrained shows that inference with the machine learning model is suspended during retraining)
Stanislav further teaches […] in response to receiving a predetermined number of feedback messages […] (Stanislav, Page 22 – line 40, “The RAN node 3 may receive measurement reports from several UEs, and then may aggregate information contained in the measurement reports of these UEs and transmit an HO REPORT message containing the aggregated information to the RAN node 2”, & Page 22 – line 20, “when the UE 6 can determine whether or not the handover from the cell 5-2 to the cell 5-1 has been normally performed, the UE 6 may set the above-described indicator contained in the measurement report in the above-described message. Further, the UE 6 may transmit the cell quality information (the cell reference signal measurement information) to the RAN node 3 via one or more RRC signaling.”, thus […] in response to receiving a predetermined number of feedback messages […] is disclosed, because Stanislav teaches that RAN node 3 may receive measurement reports from several UEs, aggregate the information contained in those measurement reports, and transmit an HO REPORT message containing the aggregated information to RAN node 2. Stanislav further teaches that the UE may transmit cell quality information to RAN node 3 via one or more RRC signaling, and that the measurement report may include an indicator showing whether the handover was normally performed. The measurement reports and one or more RRC signaling correspond to feedback messages. The aggregation of reports from several UEs corresponds to processing multiple feedback messages before transmitting the HO REPORT message. Therefore, Stanislav suggests receiving a number of feedback messages)
Adolfo further teaches […] indicative of […the device…] as the source of the failure (Adolfo, Page 4 – line 19, “an individual call's radio channel signaling and KPIs from transmitters and receivers, and is used by RCA Diagnosis unit 320 to accurately determine a root cause of the failure using the FSM methodology described above. Advanced signaling call flow 315 may also be used to determine whether the failure is in the DL or UL direction. Thus, the methodology described herein allows the detection of the exact time when every call failed from the radio point of view, what the radio resource management procedure state was at the time of the failure, the UE and eNB Call KPIs at the exact time when every call failed, the eNB Cell KPIs at the exact time when every call failed, information provided by the internal and external events before the call failed. With this information, the RCA diagnosis algorithm may provide”, thus […] indicative of […the device…] as the source of the failure is disclosed, because Adolfo teaches that radio channel signaling and KPIs from transmitters and receivers are used to accurately determine a root cause of the failure. The root cause of the failure corresponds to the source of the failure. Adolfo further teaches using UE Call KPIs at the exact time of the failure. The UE Call KPIs corresponds to information associated with the device. Therefore, Adolfo discloses information indicative of whether the device is the source of the failure)
Yuxin further teaches […] the machine learning model […] the device […] (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus Yuxin further teaches […] the machine learning model […] the device […], because Yuxin teaches that the communications device may dynamically update the model based on previous handover attempts. Yuxin’s communications device corresponds to the device. Yuxin’s model corresponds to the machine learning model)
Regarding Claim 16, Yuxin, Stanislav, and Adolfo combined with Fuchs teaches all the limitations of claim 15 as cited above and Yuxin further teaches:
[...] the machine learning model during the re-training of the machine learning model (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus […] the machine learning model during the re-training of the machine learning model is disclosed, because Yuxin teaches that the communications device updates the model based on previous handover attempts. Yuxin’s model corresponds to the machine learning model. The stored input parameters and output value based on the outcome of each handover attempt are used to update the model by adjusting one or more weights to reduce the loss function. Adjusting the model weights based on handover-attempt outcomes corresponds to retraining the machine learning model)
performing the task [….] the re-training of the machine learning model (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus performing the task [….] the re-training of the machine learning model is disclosed, because Yuxin teaches that the communications device dynamically updates the model based on previous handover attempts. Yuxin’s communications device performs the handover-related task using the machine-learning model. Yuxin’s model corresponds to the machine learning model. The stored input parameters and outcome-based output values from previous handover attempts are used to update the model by adjusting one or more weights to reduce the loss function. Updating the model weights based on handover attempt outcomes corresponds to re-training the machine learning model for performing the task)
Stanislav further teaches […] based on the feedback data […] (Stanislav, Page 22 – line 33, “The RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate the unintended event that has occurred during the handover. Further, the RAN node 3 transmits the HANDOVER REPORT message to the RAN node 2 in order to indicate that a normal HO has been performed. The unintended event may include a Too Late Handover, a Too Early Handover, and a Handover to wrong cell”, thus […] based on the feedback data […] is disclosed, because Stanislav teaches that RAN node 3 transmits a HANDOVER REPORT message to RAN node 2 to indicate an unintended event that occurred during handover. The HANDOVER REPORT message corresponds to the feedback data. The unintended event corresponds to the information used as the basis for the later action, and the Too Late Handover, Too Early Handover, and Handover to wrong cell events correspond to specific feedback information identifying the handover event)
Yuxin, Stanislav, and Adolfo combined with Fuchs does not explicitly teach suspending […] inference with […. a machine learning model during retraining…].
However, Phan teaches:
suspending […] inference with […. a machine learning model during retraining…] (Phan, Page 2 – Section 1, “We address two limitations of the traditional Simplex approach, namely lack of established guidelines for switching control back to the AC so that mission completion can be attained; and lack of techniques for correcting the AC’s behavior after a failover to the BC, so that reverse switching makes sense in the first place”, & Page 5 – Section 3, “Adaptation and Retraining. The AM is used to retrain the NC in an online manner while the BC is in control of the plant (due to NC-to-BC failover). The main purpose of this retraining is to make the NC less likely to trigger the FSC, thereby allowing it to remain in control for longer periods of time, thereby improving overall system performance”, thus suspending […] inference with [… a machine learning model during retraining…] is disclosed, because Phan teaches correcting the neural controller’s behavior after failover to the baseline controller and retraining the neural controller while the baseline controller is in control. The neural controller corresponds to the machine learning model. The baseline controller being in control due to neural controller to baseline controller failover corresponds to suspending use of the machine learning model. Retraining the neural controller online corresponds to retraining the machine learning model. Therefore, Phan discloses suspending inference with the machine learning model during retraining by switching control from the neural controller to the baseline controller while the neural controller is retrained)
and […] with a second non-machine learning algorithm during [… a retraining of a machine learning model…] (Phan, Page 5 – Section 3, “The AM is used to retrain the NC in an online manner while the BC is in control of the plant (due to NC-to-BC failover). The main purpose of this retraining is to make the NC less likely to trigger the FSC, thereby allowing it to remain in control for longer periods of time, thereby improving overall system performance. Techniques that we consider for online retraining of the NC include supervised learning and reinforcement learning. In supervised learning, state-action pairs of the form (s,a) are required for training purposes. The training algorithm uses these examples to teach the NC safe behavior. The control inputs produced by the BC can be used as training samples, although this will train the NC to imitate BC’s behavior, which may lead to a loss in performance”, thus and […] with a second non-machine learning algorithm during [… a retraining of a machine learning model…] is disclosed, because Phan teaches that the adaptation module retrains the neural controller online while the baseline controller is in control of the plant due to neural controller to baseline controller failover. The neural controller corresponds to the machine learning model. The online retraining of the neural controller corresponds to the retraining of the machine learning model. The baseline controller corresponds to the second non-machine learning algorithm)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to further combine Yuxin, Stanislav, Adolfo, and Fuchs with Phan. Yuxin teaches retraining or updating the machine learning model based on previous handover attempts. Stanislav teaches using feedback data from handover events as a basis for updating an AI model. Adolfo teaches determining the source of a communication failure, and Fuchs teaches refraining from retraining when retraining is not needed. Phan further teaches suspending use of a neural controller during retraining and performing the controlled task with a baseline controller while the neural controller is being retrained. Therefore, a POSITA would have been motivated to incorporate Phan’s failover and retraining technique into the ML-based handover system of Yuxin, Stanislav, Adolfo, and Fuchs so that, when the feedback data indicates that the machine learning model should be retrained, the system can suspend inference with the machine learning model during the re-training of the machine learning model and continue performing the task with a second non-machine learning algorithm during the re-training of the machine learning model. This would allow the network operation to continue using a reliable fallback algorithm while the machine learning model is corrected, would make future failover less likely, and would improve overall system performance (Phan, Page 5 – Section 3, “Adaptation and Retraining. The AM is used to retrain the NC in an online manner while the BC is in control of the plant (due to NC-to-BC failover). The main purpose of this retraining is to make the NC less likely to trigger the FSC, thereby allowing it to remain in control for longer periods of time, thereby improving overall system performance)
Regarding Claim 17, Yuxin, Stanislav, Adolfo, and Fuchs combined with Phan teaches all the limitations of claim 16 as cited above and Phan further teaches:
wherein the inference with […a machine learning…] is suspended, […] (Phan, Page 5 – Section 3, “Adaptation and Retraining. The AM is used to retrain the NC in an online manner while the BC is in control of the plant (due to NC-to-BC failover). The main purpose of this retraining is to make the NC less likely to trigger the FSC, thereby allowing it to remain in control for longer periods of time, thereby improving overall system performance”, thus wherein the inference with […a machine learning…] is suspended, […] is disclosed, because Phan teaches that the adaptation module retrains the neural controller online while the baseline controller is in control of the plant due to neural controller to baseline controller failover. The neural controller corresponds to the machine learning model. The neural-controller-to-baseline-controller failover corresponds to suspending inference or use of the machine learning model. The baseline controller while the neural controller is retrained shows that inference with the machine learning model is suspended during retraining)
Stanislav further teaches […] in response to receiving a predetermined number of feedback messages […] (Stanislav, Page 22 – line 40, “The RAN node 3 may receive measurement reports from several UEs, and then may aggregate information contained in the measurement reports of these UEs and transmit an HO REPORT message containing the aggregated information to the RAN node 2”, & Page 22 – line 20, “when the UE 6 can determine whether or not the handover from the cell 5-2 to the cell 5-1 has been normally performed, the UE 6 may set the above-described indicator contained in the measurement report in the above-described message. Further, the UE 6 may transmit the cell quality information (the cell reference signal measurement information) to the RAN node 3 via one or more RRC signaling.”, thus […] in response to receiving a predetermined number of feedback messages […] is disclosed, because Stanislav teaches that RAN node 3 may receive measurement reports from several UEs, aggregate the information contained in those measurement reports, and transmit an HO REPORT message containing the aggregated information to RAN node 2. Stanislav further teaches that the UE may transmit cell quality information to RAN node 3 via one or more RRC signaling, and that the measurement report may include an indicator showing whether the handover was normally performed. The measurement reports and one or more RRC signaling correspond to feedback messages. The aggregation of reports from several UEs corresponds to processing multiple feedback messages before transmitting the HO REPORT message. Therefore, Stanislav suggests receiving a number of feedback messages)
Adolfo further teaches […] indicative of […the device…] as the source of the failure (Adolfo, Page 4 – line 19, “an individual call's radio channel signaling and KPIs from transmitters and receivers, and is used by RCA Diagnosis unit 320 to accurately determine a root cause of the failure using the FSM methodology described above. Advanced signaling call flow 315 may also be used to determine whether the failure is in the DL or UL direction. Thus, the methodology described herein allows the detection of the exact time when every call failed from the radio point of view, what the radio resource management procedure state was at the time of the failure, the UE and eNB Call KPIs at the exact time when every call failed, the eNB Cell KPIs at the exact time when every call failed, information provided by the internal and external events before the call failed. With this information, the RCA diagnosis algorithm may provide”, thus […] indicative of […the device…] as the source of the failure is disclosed, because Adolfo teaches that radio channel signaling and KPIs from transmitters and receivers are used to accurately determine a root cause of the failure. The root cause of the failure corresponds to the source of the failure. Adolfo further teaches using UE Call KPIs at the exact time of the failure. The UE Call KPIs corresponds to information associated with the device. Therefore, Adolfo discloses information indicative of whether the device is the source of the failure)
Yuxin further teaches […] the machine learning model […] the device […] (Yuxin, Page 20 – line 1, “In some embodiments, the communications device 270 may dynamically update the model, based on previous handover attempts. For example, in some embodiments, the communications device 270 may store, for each handover attempted, the value of the input parameters triggering the handover and an output value based on the outcome of the handover attempt. Based on the stored parameter values and output value, the communications device 270 may update the model, for example by adjusting one or more of the weights to reduce the loss function of the model”, thus Yuxin further teaches […] the machine learning model […] the device […], because Yuxin teaches that the communications device may dynamically update the model based on previous handover attempts. Yuxin’s communications device corresponds to the device. Yuxin’s model corresponds to the machine learning model)
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant’s disclosure. US 20200259700 A1 is pertinent because it teaches AI and machine learning based network control, proactive network operations, network telemetry, alarms, performance monitoring data, feedback, and root cause analysis to identify network issues and determine remedial actions. The reference further teaches an AI driven feedback loop for adaptive control of a network, supervised learning and reinforcement learning models for network decision making, and safeguard functions that may allow, block, modify, or stop AI based actions and cause a deterministic algorithm to be used instead. Because applicant’s disclosure similarly concerns machine learning based control of network operations, feedback related to network operation failures, determining whether to update or retrain an ML model, refraining from updating or retraining when appropriate, and using a non-machine learning fallback approach during unsafe or failure conditions, the reference is relevant to the invention but is not relied upon in the rejection.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MAHLIET ADMASU whose telephone number is (571)272-0034. The examiner can normally be reached Mon-Fri, 8am-5pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Alexey Shmatov can be reached at (571)270-3428. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/M.T.A./
Examiner, Art Unit 2123
/ALEXEY SHMATOV/Supervisory Patent Examiner, Art Unit 2123
Prosecution Insights