RL Traffic Load Management for Dynamic UL/DL Allocation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing mobile communication networks face challenges in dynamically managing the imbalance between uplink (UL) and downlink (DL) traffic loads due to the evolution of services like mobile broadband, ultra-reliable low latency communications, and massive machine-type communications, which are not adequately addressed by current manual or predictive methods.

Innovation Solution

Implementing a reinforcement learning (RL) process to automatically optimize UL/DL bandwidth allocation using an RL agent with an Upper Confidence Bound (UCB) algorithm, enabling dynamic adjustments and potential handovers to manage traffic loads effectively.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If manual configuration or predictive methods are used for UL/DL bandwidth allocation, then device complexity is reduced, but adaptability to dynamic traffic patterns deteriorates

Engineering Contradiction:
Improveadaptability to dynamic traffic patternsVSAvoiddevice complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system employs self-service by enabling the network to automatically learn from historical data and make intelligent decisions about bandwidth allocation without manual intervention. The machine learning model processes historical traffic data, identifies patterns, and autonomously adjusts UL/DL bandwidth ratios to optimize network performance under varying traffic conditions.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The invention applies parameter changes by dynamically adjusting the bandwidth allocation ratio between uplink and downlink based on predicted traffic patterns. The system changes the bandwidth ratio parameter from fixed manual configuration to dynamic values derived from machine learning predictions, allowing optimal resource distribution across different time periods and network conditions.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If equal bandwidth is assigned to UL and DL traffic, then device complexity is minimized, but productivity deteriorates due to inability to handle asymmetric traffic loads

Engineering Contradiction:
ImproveproductivityVSAvoiddevice complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The system implements dynamics by transitioning from static equal bandwidth allocation to dynamic asymmetric allocation. The machine learning model continuously predicts future traffic patterns and adjusts bandwidth ratios in real-time, enabling the network to adapt to changing traffic conditions, handle peak loads, and optimize resource utilization across different time periods.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The invention applies preliminary action by using historical data to predict future traffic patterns before actual traffic occurs. The machine learning model processes past traffic information to forecast future conditions, allowing the system to proactively adjust bandwidth allocation in anticipation of changing traffic demands, thereby optimizing productivity before peaks or imbalances occur.

Inventive Principle:
Principle #10Preliminary action

3Productivity

If dynamic UL/DL bandwidth allocation optimization is implemented, then productivity is improved, but device complexity increases

Engineering Contradiction:
ImproveproductivityVSAvoiddevice complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The system implements feedback by using historical traffic data to train machine learning models that predict future patterns. The model continuously learns from actual traffic behavior, compares predictions with real outcomes, and refines its accuracy over time. This feedback mechanism enables increasingly accurate predictions and optimizations, improving productivity while keeping the complexity manageable through automated learning processes.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4179772B1Traffic load management
Publication Date: 2026.04.01 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
  • EP4179772B1 patent drawingFigure 1
  • EP4179772B1 patent drawingFigure 2
  • EP4179772B1 patent drawingFigure 3

AI summary

A reinforcement learning (RL) process is used to allocate UL/DL resources and is also used to offload traffic from a first access point (e.g., macro access point) to a second access point (e.g., a micro access point) when the load on the first access point is too high. The RL process is able to handle the dynamic nature of UL/DL imbalances and is therefore able to maximize usage of resources without compromising quality of service.