Managing faults in a telecommunications network
A method performed by a node in a telecommunications network for managing faults includes obtaining predictions of faults in the telecommunications network and time intervals in which the faults are predicted to occur. The method then includes determining possible actions that could be performed to address the predicted faults and associated resource usages to perform the possible actions, and selecting actions to perform, from the possible actions, in order to address the predicted faults, based on the predicted time intervals and the determined resource usages.
1 . A method performed by a node in a telecommunications network for managing faults, the method comprising:
obtaining predictions of faults in the telecommunications network and time intervals in which the faults are predicted to occur;
determining possible actions that could be performed to address the predicted faults and associated resource usages to perform the possible actions, wherein the resource usages are determined based on historical accuracy values of models used to predict the faults and/or confidence values with which the respective faults were predicted, and wherein determining possible actions comprises determining a plurality of action categories based on estimated time to perform actions; and
selecting actions to perform, from the possible actions, in order to address the predicted faults, based on the predicted time intervals and the determined resource usages, wherein selecting actions comprises selecting from the plurality of action categories based on the predicted time intervals in which the faults are predicted to occur.
2 . A method as in claim 1 wherein the predictions of faults are based on a plurality of network performance measures.
3 . A method as in claim 1 wherein the predictions of faults are based on one or more of:
information relating to previous faults;
information relating to actions performed in response to previous faults; and/or
information relating to configuration of the telecommunications network.
4 . A method as in claim 1 wherein the step of selecting actions to perform, comprises:
determining an order in which the selected actions should be performed.
5 . A method as in claim 4 wherein the order is determined so as to optimise resource usage and address the respective predicted faults before the time intervals in which the respective predicted faults are predicted to occur.
6 . A method as in claim 1 wherein the step of selecting actions to perform comprises selecting an action from the possible actions if:
the action may be performed in a timeframe that is less than the predicted time interval in which the respective fault is predicted to occur;
the corresponding determined resource usage is less than a resource available to perform the action and/or
the corresponding determined resource usage is less than a resource usage that would be needed to fix the fault if the fault were to left to occur.
7 . A method as in claim 1 wherein the faults are predicted to occur at a plurality of different sites in the telecommunications network and wherein the step of selecting actions to perform comprises:
selecting actions that minimise a total resource usage across the plurality of different sites.
8 . A method as in claim 7 wherein the step of selecting actions that minimise a total resource usage across the plurality of different sites, comprises selecting actions which minimise the expression:
∑
s
=
0
N
s
i
t
e
s
R
s
wherein N sites comprises a number of sites in the plurality of sites, and Rs comprises a total resource usage associated with a site S.
9 . A method as in claim 8 wherein Rs is determined according to:
R s =R A +R SiteFailure ,
wherein RA comprises a total resource usage in performing selected actions for the site, S, and R SiteFailure comprises a resource usage associated with failure of site S.
10 . A method as in claim 9 wherein R SiteFailure for the site, S, is determined using a second machine learning model that takes as input a state of the respective site and/or characteristics of the respective site.
11 . A method as in claim 8 wherein
∑
s
=
0
N
s
i
t
e
s
R
s
is minimised using a Hungarian optimisation method.
12 . A method as in claim 8 wherein
∑
s
=
0
N
s
i
t
e
s
R
s
is minimised according to one or more of the following constraints:
a total resource usage to perform the selected actions being less than a total resource available for performing actions;
a time associated with performing actions at each site being less than a predetermined time requirement for performing actions at the respective site; and
the selected actions comprising fewer actions than a maximum number of actions that may be performed at any given time.
13 . A method as in claim 1 further comprising initiating performance of the selected actions in order to proactively address the respective predicted faults.
14 . A method as in claim 1 wherein the resource usage relates to an amount of human resource needed by an engineer to perform the respective action.
15 . A method as in claim 1 wherein the resource usage relates to a cost associated with performing the respective action.
16 . A method as in claim 15 wherein the cost comprises a network cost associated with a change in a key performance indicator in the telecommunications network.
17 . A node in a telecommunications network for managing faults, the node comprising:
a memory comprising instruction data representing a set of instructions; and
a processor configured to communicate with the memory and to execute the set of instructions, wherein the set of instructions, when executed by the processor, cause the processor to:
obtain predictions of faults in the telecommunications network and time intervals in which the faults are predicted to occur;
determine possible actions that could be performed to address the predicted faults and associated resource usages to perform the possible actions, wherein the resource usages are determined based on historical accuracy values of models used to predict the faults and/or confidence values with which the respective faults were predicted, and wherein determining possible actions comprises determining a plurality of action categories based on estimated time to perform actions; and
select actions to perform, from the possible actions, in order to address the predicted faults, based on the predicted time intervals and the determined resource usages, wherein selecting actions comprises selecting from the plurality of action categories based on the predicted time intervals in which the faults are predicted to occur.
18 . A node as in claim 17 wherein the predictions of faults are based on a plurality of network performance measures.
19 . A telecommunications system comprising:
a first node; and
a plurality of other nodes,
wherein the first node is configured to:
obtain predictions of faults in the other nodes and time intervals in which the faults are predicted to occur;
determine possible actions that could be performed to address the predicted faults and associated resource usages to perform the possible actions;
select actions to perform, from the possible actions, in order to address the predicted faults, based on the predicted time intervals and the determined resource usages, wherein the resource usages are determined based on historical accuracy values of models used to predict the faults and/or confidence values with which the respective faults were predicted, and wherein determining possible actions comprises determining a plurality of action categories based on estimated time to perform actions; and
initiate performance of the selected actions at the plurality of other nodes in order to proactively address the associated predicted faults, wherein selecting actions comprises selecting from the plurality of action categories based on the predicted time intervals in which the faults are predicted to occur.