Apparatus, method, and computer program
The disclosure relates to an apparatus comprising at least one processor and at least one memory including computer code for one or more programs, the at least one memory and the computer code configured, with the at least one processor, to cause the apparatus at least to: determine ( 700 ) whether at least one reporting criterion is met; and trigger ( 702 ) the provision of a report to a central node when the at least one reporting criterion is met, wherein the report comprises an indication that at least one reporting criterion is met and/or an indication of the at least one reporting criterion being met.
1 . A distributed node for participating in federated learning, the distributed node comprising:
at least one processor; and
at least one memory storing instructions which, when executed by the at least one processor, cause the distributed node at least to perform:
receiving, from a central node participating in the federated learning, a federated learning reporting configuration comprising data collection context criterion;
determining whether at least some training data of training data that is available at the distributed node for training a local model of a global federated learning model satisfies the data collection context criterion, wherein the determining comprises comparing the data collection context criterion to a data collection context of the training data that is available at the distributed node, wherein the data collection context defines a context in which the training data that is available was collected;
based on determining that the at least some training data of the training data that is available at the distributed node satisfies the data collection context criterion, sending a federated learning node report to the central node, wherein the federated learning node report comprises an indication indicating that the at least some training data that is available at the distributed node satisfies the data collection context criterion;
wherein the data collection criterion comprises geographical area criterion specifying a geographical area and the data collection context comprises information indicating whether the training data was collected in the geographic area; and
wherein the determining that the training data that is available at the distributed node satisfies the data collection context criterion comprises determining at least that the distributed node collected the training data from the geographical area.
2 . The distributed node of claim 1 , wherein the data collection context criterion further comprises at least one of the following:
channel load criterion specifying a channel load of a channel and the data collection context comprises information indicating whether the training data that is available was collected from a channel having the channel load;
mobility criterion specifying a mobility condition of the distributed node and the data collection context comprises information indicating whether the training data that is available was collected has the mobility condition;
environment criterion specifying an environment of the distributed node and the data collection context comprises information indicating whether the training data that is available was collected in the environment; or
time criterion specifying a time and the data collection context comprises information indicating whether the training data that is available was collected at the time.
3 . The distributed node of claim 1 , wherein the federated learning reporting configuration further comprises an available resource criterion that is to be compared to an available resource of the distributed node, and wherein the federated learning node report is sent when the available resource of the distributed node satisfies the available resource criterion, and wherein the federated learning node report comprises an indication that the available resource of the distributed node satisfies the available resource criterion.
4 . The distributed node of claim 3 , wherein the federated learning reporting configuration further comprises a computation resource criterion that is to be compared to information indicating a computation resource of the distributed node.
5 . The apparatus distributed node of claim 1 , wherein the distributed node comprises a user equipment and the central node comprises a base station.
6 . The distributed node of claim 1 , wherein the distributed node comprises a base station and the central node comprises a core network function.
7 . The distributed node of claim 1 , wherein the distributed node comprises an individual management domain apparatus and the central node comprises an end-to-end service management domain apparatus.
8 . A method performed by a distributed node participating in federated learning, the method comprising:
receiving, from a central node participating in the federated learning, a federated learning reporting configuration comprising data collection context criterion; and
determining whether at least some training data of training data that is available at the distributed node for training a local model of a global federated learning model satisfies the data collection context criterion, wherein the determining comprises comparing the data collection context criterion to a data collection context of the training data that is available at the distributed node, wherein the data collection context defines a context in which the training data that is available was collected;
based on determining that the at least some training data of the training data that is available at the distributed node satisfies the data collection context criterion, sending a federated learning node report to the central node, wherein the federated learning node report comprises an indication indicating that the at least some training data of the training data that is available at the distributed node satisfies the data collection context criterion,
wherein the data collection criterion comprises geographical area criterion specifying a geographical area and the data collection context comprises information indicating whether the training data was collected in the geographic area; and
wherein the determining that the training data that is available at the distributed node satisfies the data collection context criterion comprises determining at least that the distributed node collected the training data from the geographical area.
9 . The distributed node of claim 3 , wherein the federated learning reporting configuration further comprises a power resource criterion that is to be compared to information indicating a power resource of the distributed node.
10 . The distributed node of claim 4 , wherein the federated learning reporting configuration further comprises a power resource criterion that is to be compared to information indicating a power resource of the distributed node.
11 . The method of claim 8 , wherein the data collection context criterion further comprises at least one of the following:
channel load criterion specifying a channel load of a channel and the data collection context comprises information indicating whether the training data that is available was collected from a channel having the channel load;
mobility criterion specifying a mobility condition of the distributed node and the data collection context comprises information indicating whether the training data that is available was collected has the mobility condition;
environment criterion specifying an environment of the distributed node and the data collection context comprises information indicating whether the training data that is available was collected in the environment; or
time criterion specifying a time and the data collection context comprises information indicating whether the training data that is available was collected at the time.
12 . The method of claim 8 , wherein the federated learning reporting configuration further comprises an available resource criterion that is to be compared to an available resource of the distributed node, and wherein the federated learning node report is sent when the available resource of the distributed node satisfies the available resource criterion, and wherein the federated learning node report comprises an indication that the available resource of the distributed node satisfies the available resource criterion.
13 . The method of claim 12 , wherein the federated learning reporting configuration further comprises computation resource criterion that is to be compared to information indicating a computation resource of the distributed node.
14 . The method of claim 12 , wherein the federated learning reporting configuration further comprises a power resource criterion that is to be compared to information indicating a power resource of the distributed node.
15 . The method of claim 13 , wherein the federated learning reporting configuration further comprises a power resource criterion that is to be compared to information indicating a power resource of the distributed node.
16 . The distributed node of claim 8 , wherein the distributed node comprises a user equipment and the central node comprises a base station.
17 . The method of claim 8 , wherein the distributed node comprises a base station and the central node comprises a core network function.
18 . The method of claim 8 , wherein the distributed node comprises an individual management domain apparatus and the central node comprises an end-to-end service management domain apparatus.