Cloudified MAC scheduler
Some embodiments provide a method for performing radio access network (RAN) functions in a cloud at a medium access control (MAC) scheduler application that executes on a machine deployed on a host computer in the cloud. The method receives data, via a RAN intelligent controller (RIC), from a first RAN component. The method uses the received data to generate a MAC scheduling output. The method provides the MAC scheduling output to a second RAN component via the RIC.
1. A method for performing radio access network (RAN) functions in a cloud, the method comprising:
at a medium access control (MAC) scheduler application that executes on a machine deployed on a host computer in the cloud, the MAC scheduler application operating on top of a distributed RAN intelligent controller (dRIC) that executes across the host computer as well as a plurality of additional host computers in the cloud:
receiving data, via the dRIC, from a first RAN component;
using the received data to generate a MAC scheduling output, thereby offloading the generation of said MAC scheduling output from the first RAN component; and
providing the MAC scheduling output to a second RAN component via the dRIC.
2. The method of claim 1 , wherein the first RAN component and the second RAN component are the same RAN component.
3. The method of claim 2 , wherein the RAN component is an Open RAN Distributed Unit (O-DU).
4. The method of claim 1 , wherein the first RAN component and the second RAN component are different RAN components.
5. The method of claim 1 , wherein the MAC scheduler application is a user equipment (UE)-specific beamforming application.
6. The method of claim 5 , wherein:
the data received at the UE-specific beamforming application from the first RAN component via the dRIC comprises information related to uplink sounding reference signal (UL SRS); and
the MAC scheduling output provided to the second RAN component via the dRIC comprises information related to UE-specific beamforming.
7. The method of claim 6 , wherein:
the information related to UL SRS that is received via the dRIC from the first RAN component at the UE-specific beamforming application comprises at least one of phase and quadrature (IQ) data and a channel response matrix; and
the information related to UE-specific beamforming provided to the second RAN component via the dRIC comprises at least one of a beamforming weight matrix and a beam index.
8. The method of claim 1 , wherein the MAC scheduler application is a UE radio frequency (RF) condition prediction application.
9. The method of claim 8 , wherein:
the data received at the UE RF condition prediction application from the first RAN component via the dRIC comprises a downlink channel condition report; and
the MAC scheduling output provided to the second RAN component via the dRIC comprises a predicted downlink channel condition for an upcoming scheduling window.
10. The method of claim 1 , wherein the MAC scheduler application is a multi-user multi-input multi-output (MU-MIMO) pairing suggestion application.
11. The method of claim 10 , wherein:
the data received at the MU-MIMO pairing suggestion application from the first RAN component via the dRIC comprises UE-specific beamforming information for a set of one or more UEs; and
the MAC scheduling output provided to the second RAN component via the dRIC comprises a UE pairing suggestion.
12. The method of claim 1 further comprising, prior to receiving the data:
sending a subscription request for the data to the first RAN component via the dRIC; and
receiving a subscription acknowledgment from the first RAN component via the dRIC.
13. The method of claim 12 , wherein:
the dRIC receives the subscription request from the MAC scheduler application through a set of APIs of the dRIC; and
the dRIC converts the subscription request received from the MAC scheduler application into a subscription request in a format recognized by the first RAN component.
14. The method of claim 13 , wherein the dRIC receives the data from the first RAN component and directs the data to the MAC scheduler application via the set of APIs based on stored information regarding the subscription of the MAC scheduler application to the data.
15. The method of claim 1 , wherein using the received data to generate a MAC scheduling output comprises applying a machine learning algorithm to the received data.
16. A non-transitory machine-readable medium storing a medium access control (MAC) scheduler application which when executed by at least one processing unit of a host computer performs radio access network (RAN) functions in a cloud, the MAC scheduler application executing on a machine deployed on a host computer, the MAC scheduler application comprising sets of instructions for:
receiving data, via a RAN intelligent controller (RIC), from a first RAN component;
using the received data to generate a MAC scheduling output; and
providing the MAC scheduling output to a second RAN component via the RIC.
17. The non-transitory machine-readable medium of claim 16 , wherein the first RAN component and the second RAN component are a same Open RAN Distributed Unit (O-DU).
18. The non-transitory machine-readable medium of claim 16 , wherein:
the MAC scheduler application is a user equipment (UE)-specific beamforming application;
the data received from the first RAN component comprises information related to uplink sounding reference signal (UL SRS); and
the MAC scheduling output comprises information related to UE-specific beamforming.
19. The non-transitory machine-readable medium of claim 16 , wherein:
the MAC scheduler application is a UE radio frequency (RF) condition prediction application;
the data received from the first RAN component comprises a downlink channel condition report; and
the MAC scheduling output comprises a predicted downlink channel condition for an upcoming scheduling window.
20. The non-transitory machine-readable medium of claim 16 , wherein:
the MAC scheduler application is a multi-user multi-input multi-output (MU-MIMO) pairing suggestion application;
the data received from the first RAN component comprises UE-specific beamforming information for a set of one or more UEs; and
the MAC scheduling output comprises a UE pairing suggestion.
21. The non-transitory machine-readable medium of claim 16 , wherein the MAC scheduler application further comprises sets of instructions for, prior to receiving the data:
sending a subscription request for the data to the first RAN component via the RIC; and
receiving a subscription acknowledgment from the first RAN component via the RIC,
wherein the RIC receives the subscription request from the MAC scheduler application through a set of APIs of the RIC,
wherein the RIC converts the subscription request received from the MAC scheduler application into a subscription request in a format recognized by the first RAN component.
22. The non-transitory machine-readable medium of claim 21 , wherein the RIC receives the data from the first RAN component and directs the data to the MAC scheduler application via the set of APIs based on stored information regarding the subscription of the MAC scheduler application to the data.
23. The non-transitory machine-readable medium of claim 16 , wherein the set of instructions for using the received data to generate a MAC scheduling output comprises a set of instructions for applying a machine learning algorithm to the received data.
24. The non-transitory machine-readable medium of claim 16 , wherein the MAC scheduler executes on the host computer on top of a distributed RAN intelligent controller dRIC) that executes across the host computer as well as a plurality of additional host computers in the cloud.