ML-Based RRM Analytics in O-RAN Networks

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing 5G networks face challenges in optimizing Radio Resource Management (RRM) due to increasing demand for mobile data and heterogeneous Quality of Service (QoS) requirements, which are not effectively addressed by conventional RRM methods.

Innovation Solution

The implementation of a machine learning-based RRM analytics module within the Open Radio Access Network (O-RAN) architecture, which utilizes reinforcement learning to optimize radio resource allocation by analyzing performance measurements such as buffer occupancy, channel state information, and packet error rates.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If conventional RRM methods are used in 5G networks, then network coverage and basic connectivity are maintained, but the system cannot effectively handle heterogeneous QoS requirements and increasing mobile data demand

Engineering Contradiction:

ImproveQoS requirement handling capabilityVSAvoidmobile data throughput

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent replaces conventional deterministic RRM algorithms with machine learning-based stochastic optimization. The RL agent learns optimal scheduling policies by interacting with the network environment, enabling adaptive handling of heterogeneous QoS requirements while maximizing mobile data throughput through data-driven decision making rather than fixed mechanical rules

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system dynamically adjusts RRM parameters such as scheduling priority, resource allocation ratios, and QoS thresholds based on real-time network conditions and learned patterns. The RL agent modifies these parameters adaptively to balance competing QoS requirements and optimize throughput under varying load conditions

Inventive Principle:

Principle #35Parameter changes

2Productivity

If machine learning-based RRM analytics module is implemented, then near-optimal radio resource allocation is achieved, but system complexity increases

Engineering Contradiction:

Improveradio resource allocation efficiencyVSAvoidRRM analytics module complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The RRM analytics module is segmented into distinct functional components: data collection subsystem, feature engineering module, RL training system, and real-time decision execution unit. This segmentation allows each component to be optimized independently and simplifies the overall complex ML-based system while maintaining high allocation efficiency

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary RL agent that acts as a mediator between the complex network state and the RRM decision-making process. This intermediary layer abstracts the complexity of the underlying network conditions and translates them into actionable scheduling decisions, reducing the operational complexity at execution time

Inventive Principle:

Principle #24Intermediary (Mediator)

3Speed

If reinforcement learning is used to optimize radio resource allocation, then throughput is enhanced and latency is reduced, but computational resources and processing time are increased

Engineering Contradiction:

Improvedata transmission speedVSAvoidcomputational energy consumption

Core Design Contradiction:

SpeedVSUse of energy by moving object

Solution Approach 1:

The RL agent performs preliminary learning and policy optimization during training phases using historical network data and simulations. Once trained, the agent can execute pre-learned optimal policies in near-real-time without intensive computation, enabling fast throughput while reducing online computational energy consumption through offline pre-training

Inventive Principle:

Principle #10Preliminary action

4Reliability

If performance measurements are collected and analyzed for each UE, then QoS management is improved, but measurement and analysis overhead increases

Engineering Contradiction:

ImproveQoS fulfillment reliabilityVSAvoidmeasurement and analysis time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent extracts and selects only the most critical performance measurement parameters (such as buffer occupancy, channel state information, and packet error rates) rather than collecting all possible metrics. This selective extraction maintains QoS fulfillment reliability while significantly reducing measurement and analysis overhead time

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS20250150897A1Optimized radio resource management using machine learning approaches in o-ran networks

Publication Date: 2025.05.08 GLAS USA LLC

AI summary

A method for performing optimized radio resource management (RRM) in an O-RAN network, includes: performing, at the RRM analytics module, a machine learning-based analysis to determine an optimal resource allocation policy based on at least one performance measurement performed at a distributed unit (DU); receiving, by the DU, the optimal resource allocation policy determined by the RRM; and utilizing, by the DU, the optimal resource allocation policy to one of schedule or not schedule user equipments (UEs). In a given slot and for a given state of the base station, candidate UEs to serve in the slot are selected based on a chosen policy. If the number of selected candidate UEs is higher than the maximum number the base station can serve in that slot, the radio resource management (RRM) module selects, using the chosen policy, the maximum number of UEs, and the selected UEs are allocated resources.