# US 12,114,292 B2 · Method, device and radio access network for adjusting resources of radio access network

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,114,292 B2
- Document ID: US12114292B2
- Status: Granted
- Kind: B2
- Application number: 17/829,370
- Filed: 2022-06-01
- Published: 2023-09-14
- Granted: 2024-10-08
- Earliest claimed: 2022-03-09
- Assignees: Wistron Corporation
- Applicants: Wistron Corporation
- Inventors: Yuan-Fu Huang
- Web page: https://patentgrep.com/patent/US12114292B2

## CPC

- G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
- H04W 72/04: Wireless resource allocation
- G06N 3/063: using electronic means
- G06N 3/08: Learning methods
- H04W 16/22: Traffic simulation tools or models
- H04W 24/02: Arrangements for optimising operational condition
- H04W 88/085: {Access point devices with remote components}
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- H04W: WIRELESS COMMUNICATION NETWORKS (broadcast communication H04H; communication systems using wireless links for non-selective communication, e.g. wireless extensions H04M1/72)

## Abstract

A resource adjustment method for a radio access network includes obtaining a plurality of radio access network training information at a first time point; predicting a radio access network usage condition of a second time point according to the plurality of radio access network training information; and pre-adjusting the radio access network resource allocation of the second time point according to the radio access network usage condition at a third time point, so as to allocate a plurality of user equipments to a plurality of radio units and adjust an arrangement of computing resources of a distributed unit and a central unit; wherein the first time point is earlier than the third time point, and the third time point is earlier than the second time point.

## Claims

### Claim 1

A resource adjustment method for a radio access network (RAN), wherein the RAN complies with a specification of an open radio access network (O-RAN) of a 5th generation mobile communication system (5G) and comprises a plurality of radio units (RU), a distributed unit (DU) and a central unit (CU), the resource adjustment method comprising:
obtaining a plurality of RAN training information at a first time point;
predicting a PAN usage condition of a second time point according to the plurality of RAN training information; and
pre-adjusting a PAN resource allocation of the second time point according to the PAN usage condition at a third time point, so as to allocate a plurality of user equipments (UE) to the plurality of RUs and adjust an arrangement of computing resources of the DU and the CU;
wherein the first time point is earlier than the third time point, and the third time point is earlier than the second time point.

### Claim 2 (depends on 1)

The resource adjustment method of claim 1, wherein the plurality of RAN training information comprise one or more information of a radio resource utilization, a UE throughput, a CPU utilization and an accelerator utilization.

### Claim 3 (depends on 1)

The resource adjustment method of claim 1, wherein the step of predicting the RAN usage condition of the second time point according to the plurality of RAN training information comprises:
predicting a RAN predicting information of the second time point according to a first AI model; and
predicting the RAN usage condition of the second time point according to a second AI model.

### Claim 4 (depends on 3)

The resource adjustment method of claim 3, further comprising:
training the first AI model with the plurality of RAN training information by a first deep learning method; and
training the second AI model with the RAN predicting information by a second deep learning method.

### Claim 5 (depends on 3)

The resource adjustment method of claim 3, wherein the first AI model is a recurrent neural network (RNN) or a long short-term memory (LSTM).

### Claim 6 (depends on 1)

The resource adjustment method of claim 1, wherein the RAN usage condition of the second time point comprises a first feature value, a second feature value and a third feature value, the first feature value is related to a predicted value of an average IP throughput between the user equipments and the RU, the second feature value is related to a predicted value of an average downlink physical resource block (PRB) usage rate, and the third feature value is related to a predicted value of a CPU utility of the DU or the CU.

### Claim 7 (depends on 6)

The resource adjustment method of claim 6, wherein the step of pre-adjusting the RAN resource allocation of the second time point according to the RAN usage condition at the third time point comprises:
in response to the first feature value being smaller than a first threshold and the second feature value being larger than a second threshold, equally distributing the plurality of user equipments among the plurality of RUs;
in response to the first feature value being larger than the first threshold, respectively adjusting a computing resource of the CU or the DU according to a ratio of the third feature value to the CPU utility of the third time point; and
in response to the second feature value being smaller than the second threshold, respectively adjusting a computing resource of the CU or the DU according to a ratio of the third feature value to the CPU utility of the third time point.

### Claim 8

A resource adjustment device for a radio access network (RAN), wherein the RAN complies with a specification of an open radio access network (O-RAN) of a 5th generation mobile communication system (5G) and comprises a plurality of radio units (RU), a distributed unit (DU) and a central unit (CU), the resource adjustment device comprising:
a processing unit, configured to execute a program code; and
a storage unit, coupled to the processing unit to store the program code, wherein the program code is configured to instruct the processing unit to execute a resource adjustment method, the resource adjustment method comprises:
obtaining a plurality of RAN training information at a first time point;
predicting a PAN usage condition of a second time point according to the plurality of RAN training information; and
pre-adjusting a PAN resource allocation of the second time point according to the RAN usage condition at a third time point, so as to allocate a plurality of user equipments (UE) to the plurality of RUs and adjust an arrangement of computing resources of the DU and the CU;
wherein the first time point is earlier than the third time point, and the third time point is earlier than the second time point.

### Claim 9 (depends on 8)

The resource adjustment device of claim 8, wherein the plurality of RAN training information comprise one or more information of a radio resource utilization, a UE throughput, a CPU utilization and an accelerator utilization.

### Claim 10 (depends on 8)

The resource adjustment device of claim 8, wherein the step of predicting the RAN usage condition of the second time point according to the plurality of RAN training information comprises:
predicting a RAN predicting information of the second time point according to a first AI model; and
predicting the RAN usage condition of the second time point according to a second AI model.

### Claim 11 (depends on 10)

The resource adjustment device of claim 10, wherein the resource adjustment method further comprises:
training the first AI model with the plurality of RAN training information by a first deep learning method; and
training the second AI model with the RAN predicting information by a second deep learning method.

### Claim 12 (depends on 10)

The resource adjustment device of claim 10, wherein the first AI model is a recurrent neural network (RNN) or a long short-term memory (LSTM).

### Claim 13 (depends on 8)

The resource adjustment device of claim 8, wherein the RAN usage condition of the second time point comprises a first feature value, a second feature value and a third feature value, the first feature value is related to a predicted value of an average IP throughput between the user equipments and the RU, the second feature value is related to a predicted value of an average downlink physical resource block (PRB) usage rate, and the third feature value is related to a predicted value of a CPU utility of the DU or the CU.

### Claim 14 (depends on 13)

The resource adjustment device of claim 13, wherein the step of pre-adjusting the RAN resource allocation of the second time point according to the RAN usage condition at the third time point comprises:
in response to the first feature value being smaller than a first threshold and the second feature value being larger than a second threshold, equally distributing the plurality of user equipments among the plurality of RUs; and
in response to the first feature value being larger than the first threshold or the second feature value being smaller than the second threshold, respectively adjusting a computing resource of the CU or the DU according to a ratio of the third feature value to the CPU utility of the third time point.

### Claim 15

A radio access network (RAN), wherein the RAN complies with a specification of an open radio access network (O-RAN) of a 5th generation mobile communication system (5G), and comprises:
a plurality of radio units (RUs);
a distributed unit (DU), running on a first O-Cloud platform;
a central unit (CU), running on a second O-Cloud platform; and
a resource adjustment device, configured to execute a resource adjustment method for the radio access network, the resource adjustment method comprises:
obtaining a plurality of RAN training information at a first time point;
predicting a PAN usage condition of a second time point according to the plurality of RAN training information; and
pre-adjusting a PAN resource allocation of the second time point according to the PAN usage condition at a third time point, so as to allocate a plurality of user equipments (UE) to the plurality of RUs and adjust an arrangement of computing resources of the DU and the CU;
wherein the first time point is earlier than the third time point, and the third time point is earlier than the second time point.

### Claim 16 (depends on 15)

The radio access network of claim 15, wherein the plurality of RAN training information comprise one or more information of a radio resource utilization, a UE throughput, a CPU utilization and an accelerator utilization.

### Claim 17 (depends on 15)

The radio access network of claim 15, wherein the step of predicting the RAN usage condition of the second time point according to the plurality of RAN training information comprises:
training a first AI model with the plurality of RAN training information by a first deep learning method;
predicting a RAN predicting information of the second time point according to the first AI model;
training a second AI model with the RAN predicting information by a second deep learning method; and
predicting the RAN usage condition of the second time point according to the second AI model.

### Claim 18 (depends on 17)

The radio access network of claim 17, wherein the first AI model is a recurrent neural network (RNN) or a long short-term memory (LSTM).

### Claim 19 (depends on 15)

The radio access network of claim 15, wherein the RAN usage condition of the second time point comprises a first feature value, a second feature value and a third feature value, the first feature value is related to a predicted value of an average IP throughput between the user equipments and the RU, the second feature value is related to a predicted value of an average downlink physical resource block (PRB) usage rate, and the third feature value is related to a predicted value of a CPU utility of the DU or the CU.

### Claim 20 (depends on 19)

The radio access network of claim 19, wherein the step of pre-adjusting the RAN resource allocation of the second time point according to the RAN usage condition at the third time point comprises:
in response to the first feature value being smaller than a first threshold and the second feature value being larger than a second threshold, equally distributing the plurality of user equipments among the plurality of RUs;
in response to the first feature value being larger than the first threshold, respectively adjusting a computing resource of the CU or the DU according to a ratio of the third feature value to the CPU utility of the third time point; and
in response to the second feature value being smaller than the second threshold, respectively adjusting a computing resource of the CU or the DU according to a ratio of the third feature value to the CPU utility of the third time point.