Determining open loop power control parameters

1. An apparatus comprising means for:

obtaining, from one or more first base stations, a first set of key performance indicators associated with a first set of open loop power control parameters;

determining a statistical model based at least partly on the first set of key performance indicators and the first set of open loop power control parameters;

determining a second set of open loop power control parameters based at least partly on the statistical model;

updating a set of hyperparameters associated with the statistical model by maximizing a likelihood of the first set of key performance indicators, wherein the set of hyperparameters define at least a kernel function and a noise variance associated with the statistical model;

wherein the statistical model comprises a Gaussian process model.

2. An apparatus according to claim 1 , wherein the apparatus further comprises means for transmitting the second set of open loop power control parameters to the one or more first base stations.

3. An apparatus according to claim 1 , wherein the apparatus further comprises means for obtaining contextual information from the one or more first base stations, wherein the second set of open loop power control parameters is determined based at least partly on the contextual information.

4. An apparatus according to claim 3 , wherein the contextual information comprises one or more of the following: a radio cell load, a hardware cell load, a backhaul cell load, a percentage of terminal devices at a cell edge, a satellite image, a two-dimensional map, a three-dimensional building model, a spatio-temporal distribution of the terminal devices, weather information, a date, a timestamp, and/or a random access channel failure rate.

5. An apparatus according to any claim 1 , wherein the apparatus further comprises means for obtaining historical information indicating a relationship between one or more third sets of open loop power control parameters and one or more second sets of key performance indicators associated with one or more second base stations, wherein the statistical model is determined based at least partly on the historical information.

6. An apparatus according to claim 1 , wherein the first set of key performance indicators comprises an average achieved throughput over a time window, an average signal-to-interference-plus-noise ratio associated with a plurality of terminal devices, and/or a number of lost packets.

7. An apparatus according to claim 1 , wherein the second set of open loop power control parameters is determined by maximizing an acquisition function that is a function of a mean and a variance associated with the Gaussian process model.

8. An apparatus according to claim 1 , wherein the second set of open loop power control parameters is determined by solving an optimization problem using at least the statistical model as input.

9. An apparatus according to claim 1 , wherein the second set of open loop power control parameters is determined by using Bayesian optimization.

10. An apparatus according to claim 1 , wherein the apparatus is comprised in a centralized self-organizing network server.

11. A method comprising:

obtaining, from one or more base stations, a first set of key performance indicators associated with a first set of open loop power control parameters;

determining a statistical model based at least partly on the first set of key performance indicators and the first set of open loop power control parameters;

determining a second set of open loop power control parameters based at least partly on the statistical model;

updating a set of hyperparameters associated with the statistical model by maximizing a likelihood of the first set of key performance indicators, wherein the set of hyperparameters define at least a kernel function and a noise variance associated with the statistical model;

wherein the statistical model comprises a Gaussian process model.

12. A non-transitory computer program comprising instructions for causing an apparatus to perform at least the following:

obtain, from one or more base stations, a first set of key performance indicators associated with a first set of open loop power control parameters;

determine a statistical model based at least partly on the first set of key performance indicators and the first set of open loop power control parameters;

determine a second set of open loop power control parameters based at least partly on the statistical model;

update a set of hyperparameters associated with the statistical model by maximizing a likelihood of the first set of key performance indicators, wherein the set of hyperparameters define at least a kernel function and a noise variance associated with the statistical model;

wherein the statistical model comprises a Gaussian process model.

13. A system comprising at least a first apparatus and one or more base stations:

wherein the one or more base stations are configured to:

collect a first set of key performance indicators associated with a first set of open loop power control parameters;

transmit the first set of key performance indicators to the first apparatus;

wherein the first apparatus is configured to:

receive the first set of key performance indicators from the one or more base stations;

determine a statistical model based at least partly on the first set of key performance indicators and the first set of open loop power control parameters;

determine a second set of open loop power control parameters based at least partly on the statistical model;

update a set of hyperparameters associated with the statistical model by maximizing a likelihood of the first set of key performance indicators, wherein the set of hyperparameters define at least a kernel function and a noise variance associated with the statistical model;

transmit the second set of open loop power control parameters to the one or more base stations;

wherein the one or more base stations are further configured to:

receive the second set of open loop power control parameters from the first apparatus; and

apply the second set of open loop power control parameters;

wherein the statistical model comprises a Gaussian process model.