Reinforcement Learning Agent for Radio Resource Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Radio Resource Management (RRM) in radio access networks, such as LTE, faces challenges including power control, radio resource scheduling, and load balancing, leading to complex signaling overhead and performance degradation due to the need for dedicated algorithms for specific tasks and interactions between RRM tasks.
Innovation Solution
An agent node is introduced to autonomously learn radio resource management strategies using environmental measurements, determining control actions for radio resource parameters through a control policy, allowing for adaptive management without manual adjustments and improving network performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If dedicated algorithms are used for specific RRM tasks, then the network operation is optimized for one specific task, but the same algorithm cannot be reused for other RRM tasks and large signalling overhead is introduced
Solution Approach 1:
The patent applies a single learning-based algorithm that can perform multiple RRM tasks (power control, scheduling, load balancing) universally, replacing multiple dedicated algorithms. The agent node learns to handle different RRM functions through a unified reinforcement learning framework, eliminating the need for separate dedicated algorithms for each task and reducing signaling overhead between network elements.
Solution Approach 2:
The patent merges multiple RRM functions (power control, scheduling, load balancing) into a single integrated learning algorithm. Instead of having separate algorithms interacting with each other, the invention combines them into one unified agent that learns to coordinate all RRM tasks simultaneously, thereby reducing the complexity and signaling overhead associated with inter-algorithm interactions.
2Reliability
If multiple dedicated RRM algorithms are deployed for different tasks, then each task can be optimized independently, but additional complexity and performance degradation occur due to interactions between RRM tasks
Solution Approach 1:
The learning algorithm is designed to be universal and handle multiple RRM tasks simultaneously through a single unified framework. The agent learns to perform power control, scheduling, and load balancing as integrated functions rather than separate tasks, maintaining task-specific optimization while eliminating the complexity of coordinating multiple independent algorithms.
3Adaptability or versatility
If traditional RRM algorithms are used, then the system follows conventional approaches, but adaptability to changing radio environment conditions is limited without manual adjustment
Solution Approach 1:
The patent implements self-service through a learning-based algorithm that automatically adapts to changing radio environment conditions without manual intervention. The agent node continuously learns from environmental feedback and autonomously adjusts RRM parameters, replacing the need for manual configuration and adaptation while enhancing environmental adaptability.
Solution Approach 2:
The invention incorporates feedback mechanisms where the learning algorithm receives continuous information about radio environment conditions and performance outcomes. This feedback loop enables the system to automatically adapt to changing conditions by learning from past experiences and adjusting its behavior accordingly, eliminating the need for manual reconfiguration.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Agent node (210, 500) and method (600) therein, for configuring a radio resource parameter of a radio cell (215) of a communication system (200). The agent node (210, 500) is configured to obtain a control policy (430); determine a feature representing a state of at least a part of the communication system (200), at a first time period; determine a control action to be performed for control in the radio cell (215) at the first time period, out of a set of available control actions associated with the radio cell (215), based on the obtained control policy (430) and the determined at least one feature; and configure an instruction of the radio cell (215) based on the determined control action. Also, a corresponding trainer node (400) and method (800) for providing the agent node (210, 500) with the control policy (430), based on received parameters is disclosed.