Systems and methods for utilizing machine learning models to conserve energy in network devices

1 . A method, comprising:

receiving, by a device, network data identifying reference signal data for a radio access network (RAN), control signal data for the RAN, and network key performance indicators (KPIs) associated with the RAN;

receiving, by the device, energy consumption data identifying energy consumption by the RAN;

processing, by the device, the network data and the energy consumption data, with one or more machine learning models, to identify actions that reduce energy consumption at a radio unit (RU), a distributed unit (DU), or a control unit (CU) of the RAN and that control and minimize control signal and reference signal utilization at the RAN; and

causing, by the device, the actions to be implemented by the RU, the DU, the CU, or the RAN to save energy at the RAN, wherein the actions comprise causing the RAN to reduce traffic on a control channel of the RAN.

2 . The method of claim 1 , further comprising:

processing the network data and the energy consumption data, with the one or more machine learning models, to identify additional actions that load balance a shared channel of the RAN, reduce mobility of a user equipment (UE), and reduce data on the control channel of the RAN; and

causing the additional actions to be implemented by the RAN or the UE to save energy at the RAN or the UE.

3 . The method of claim 2 , wherein the additional actions prevent the UE from moving from the RAN to another RAN geographically located adjacent to the RAN.

4 . The method of claim 1 , further comprising:

processing the network data and the energy consumption data, with the one or more machine learning models, to identify additional actions that reduce and shut down traffic channel power at the RAN while maintaining the control signal and reference signal utilization at the RAN; and

causing the additional actions to be implemented by the RAN to save energy at the RAN.

5 . The method of claim 1 , further comprising:

processing the network data and the energy consumption data, with the one or more machine learning models, to identify additional actions that utilize sleep modes, connected mode discontinuous reception (CDRX) cycle settings, physical downlink control channel (PDCCH) monitoring, synchronization signal block (SSB) periodicity settings, or selective switching on and off radio frequency (RF) channels at the RAN; and

causing the additional actions to be implemented by the RAN to save energy at the RAN.

6 . The method of claim 1 , further comprising:

processing the network data and the energy consumption data, with the one or more machine learning models, to identify additional actions that utilize scaling resources of the RAN, switching the RAN on and off, scaling resources of the CU or the DU of the RAN, and switching the CU or the DU on and off; and

causing the additional actions to be implemented by the RAN to save energy at the RAN, the CU, or the DU.

7 . The method of claim 1 , wherein the energy consumption data is received from one or more energy meter counters associated with the RAN.

8 . A device, comprising:

one or more processors configured to:

receive network data identifying reference signal data for a radio access network (RAN), control signal data for the RAN, and network key performance indicators (KPIs) associated with the RAN;

receive energy consumption data identifying energy consumption by the RAN;

process the network data and the energy consumption data, with one or more machine learning models, to identify actions that reduce energy consumption at a radio unit (RU), a distributed unit (DU), or a control unit (CU) of the RAN and that control and minimize control signal and reference signal utilization at the RAN;

cause the actions to be implemented by the RU, the DU, the CU, or the RAN to save energy at the RAN, wherein the actions comprise causing the RAN to reduce traffic on a control channel of the RAN;

process the network data and the energy consumption data, with the one or more machine learning models, to identify additional actions that load balance a shared channel of the RAN, reduce mobility of a user equipment (UE), and reduce data on the control channel of the RAN; and

cause the additional actions to be implemented by the RAN or the UE to save the energy at the RAN or the UE.

9 . The device of claim 8 , wherein the reference signal data includes data identifying levels of reference signals associated with the RAN, and the control signal data includes data identifying levels of control signals associated with the RAN.

10 . The device of claim 8 , wherein the network data further identifies a frequency domain resource allocation associated with the RAN.

11 . The device of claim 8 , wherein the actions maintain the network KPIs while controlling and minimizing the control signal and reference signal utilization at the RAN.

12 . The device of claim 8 , wherein the network KPIs provide a measure of one or more of:

user throughput at the RAN,

latency at the RAN, or

reliability of the RAN.

13 . The device of claim 8 , wherein the one or more processors are further configured to:

train the one or more machine learning models with historical traffic load data for the RAN, historical UE location distribution data for the RAN, historical UE mobility data for the RAN, and historical energy consumption by the RAN.

14 . The device of claim 8 , wherein the one or more processors are further configured to:

retrain the one or more machine learning models based on the actions.

15 . A non-transitory computer-readable medium storing a set of instructions, the set of instructions comprising:

one or more instructions that, when executed by one or more processors of a device, cause the device to:

receive network data identifying reference signal data for a radio access network (RAN), control signal data for the RAN, and network key performance indicators (KPIs) associated with the RAN;

receive energy consumption data identifying energy consumption by the RAN,

wherein the energy consumption data is received from one or more energy meter counters associated with the RAN;

process the network data and the energy consumption data, with one or more machine learning models, to identify actions that reduce energy consumption at a radio unit (RU), a distributed unit (DU), or a control unit (CU) of the RAN and that control and minimize control signal and reference signal utilization at the RAN; and

cause the actions to be implemented by the RU, the DU, the CU, or the RAN to save energy at the RAN, wherein the actions comprise causing the RAN to reduce traffic on a control channel of the RAN.

16 . The non-transitory computer-readable medium of claim 15 , wherein the one or more instructions further cause the device to:

process the network data and the energy consumption data, with the one or more machine learning models, to identify additional actions that load balance a shared channel of the RAN, reduce mobility of a user equipment (UE), and reduce data on the control channel of the RAN; and

cause the additional actions to be implemented by the RAN or the UE to save energy at the RAN or the UE.

17 . The non-transitory computer-readable medium of claim 15 , wherein the one or more instructions further cause the device to:

process the network data and the energy consumption data, with the one or more machine learning models, to identify additional actions that reduce traffic channel power at the RAN while maintaining a control signal and a reference signal at the RAN; and

cause the additional actions to be implemented by the RAN to save energy at the RAN.

18 . The non-transitory computer-readable medium of claim 15 , wherein the one or more instructions further cause the device to:

process the network data and the energy consumption data, with the one or more machine learning models, to identify additional actions that utilize sleep modes, connected mode discontinuous reception (CDRX) cycle settings, physical downlink control channel (PDCCH) monitoring, synchronization signal block (SSB) periodicity settings, or selective switching on and off radio frequency (RF) channels at the RAN; and

cause the additional actions to be implemented by the RAN to save energy at the RAN.

19 . The non-transitory computer-readable medium of claim 15 , wherein the one or more instructions further cause the device to:

process the network data and the energy consumption data, with the one or more machine learning models, to identify additional actions that utilize scaling resources of the RAN, switching the RAN on and off, scaling resources of the CU or the DU of the RAN, and switching the CU or the DU on and off; and

cause the additional actions to be implemented by the RAN to save energy at the RAN, the CU, or the DU.

20 . The non-transitory computer-readable medium of claim 15 , wherein the one or more instructions further cause the device to:

train the one or more machine learning models with historical traffic load data for the RAN, historical user equipment (UE) location distribution data for the RAN, historical UE mobility data for the RAN, and historical energy consumption by the RAN; and

retrain the one or more machine learning models based on the actions.