Wireless Resource Scheduling Using Reinforcement Learning Policies
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional schedulers in wireless communication systems, such as those used in 6G networks, rely on heuristic metrics like Proportional Fairness (PF) for resource allocation, which may not optimize network performance effectively.
Innovation Solution
Implementing a policy-based reinforcement learning agent to determine frequency domain scheduling decisions based on a feature vector of devices and frequency resource units, enabling simultaneous allocation across the entire set of available units, thereby enhancing throughput, reducing latency, and improving spectral efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional heuristic metrics like Proportional Fairness are used for scheduling decisions, then the scheduling operation is simple to implement, but network performance optimization is insufficient
Solution Approach 1:
The patent replaces conventional heuristic-based scheduling mechanisms with a reinforcement learning agent that uses neural network-based policy evaluation. This substitution enables the system to learn optimal scheduling strategies through interaction with the wireless environment, achieving superior network performance while maintaining computational efficiency through parallel processing of frequency resource units.
Solution Approach 2:
The patent transforms the scheduling approach by changing from fixed heuristic metrics to dynamic parameter optimization through reinforcement learning. The system continuously adjusts scheduling parameters based on learned policies and environmental feedback, enabling adaptive optimization of network performance metrics such as throughput and spectral efficiency.
2Productivity
If reinforcement learning is used for scheduling decisions, then network performance is optimized, but the computational complexity increases
Solution Approach 1:
The patent segments the scheduling problem into independent frequency resource unit evaluations that can be processed in parallel. By dividing the overall scheduling task into separate, concurrently executable operations for each frequency resource unit, the system achieves computational efficiency while maintaining the sophistication of reinforcement learning-based optimization.
Solution Approach 2:
The patent implements partial action by evaluating and scheduling frequency resource units independently rather than requiring complete system state analysis. This allows the reinforcement learning agent to make scheduling decisions based on局部 information for each frequency resource unit, reducing computational complexity while maintaining overall optimization effectiveness.
3Productivity
If conventional scheduling methods are used, then the system is easier to operate, but spectral efficiency is reduced
Solution Approach 1:
The patent replaces traditional manual or heuristic-based scheduling operations with an automated reinforcement learning system. The neural network-based agent autonomously learns optimal scheduling strategies through environmental interaction, achieving high spectral efficiency without requiring manual configuration or complex operational intervention.
Solution Approach 2:
The reinforcement learning agent performs self-service by autonomously optimizing scheduling decisions through continuous learning from environmental feedback. The system automatically adapts to changing wireless conditions and optimizes spectral efficiency without external intervention, making the complex operation transparent to users while achieving superior performance.
Data Source
AI summary
The present subject matter relates to a method for performing for a time unit a scheduling operation comprising: receiving a time domain scheduling decision for the time unit, the time domain scheduling decision indicating a set of devices of the wireless communication system; determining a feature vector descriptive of the set of devices and an available set of frequency resource units of the wireless communication system; inputting the feature vector to a policy-based reinforcement learning agent for receiving an output, the output comprising a distribution between the set of devices and the set of frequency resource units; and using the distribution to determine a frequency domain scheduling decision.