Radar Resource Scheduling Using Reinforcement Learning Agents
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional radar resource management systems face computational inefficiencies and time-consuming processes when determining task configurations due to the brute force or heuristic approaches used for resource allocation, making real-time execution challenging.
Innovation Solution
Implementing a trained agent using reinforcement learning to identify task configurations with high utility, reducing computational complexity and increasing processing speed by disregarding less useful configurations, thereby optimizing resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional brute force or heuristic approaches are used to determine task configurations for every task, then all possible configurations are evaluated to ensure optimal resource allocation, but the computational complexity increases significantly and processing time becomes too long for real-time operation
Solution Approach 1:
The patent extracts and removes low-utility task configurations from the evaluation set. By using a trained agent to pre-filter configurations and keep only those with high utility values, the system eliminates unnecessary computational evaluations of configurations that would not contribute to optimal resource allocation, thus resolving the contradiction between evaluating all configurations and maintaining real-time processing speed
Solution Approach 2:
The patent applies preliminary action by training an agent beforehand to evaluate and rank task configurations based on utility. This pre-computed knowledge is then used during real-time operation to quickly identify high-utility configurations without performing exhaustive searches, enabling the system to maintain both optimality and real-time processing capability
2Measurement precision
If multiple task configurations are evaluated for each task to ensure good candidates are chosen, then the quality of resource allocation improves, but the computational complexity and time consumption increase
Solution Approach 1:
The system extracts only the high-utility task configurations from the complete set of possible configurations. The trained agent identifies and retains configurations with utility values above a certain threshold, discarding the rest. This extraction process maintains measurement precision by keeping only quality configurations while significantly reducing computational complexity by eliminating evaluations of low-value configurations
Solution Approach 2:
The patent changes the parameter of configuration evaluation from exhaustive enumeration to utility-based filtering. By introducing a utility parameter that quantifies the quality of each configuration and using a trained agent to evaluate this parameter, the system transforms the complex combinatorial optimization problem into a more manageable filtering problem, reducing computational complexity while maintaining selection quality
Data Source
Figure 1a~1c
Figure 1d~2a
Figure 2b~3a
AI summary
A radar resource management system comprises a processing circuit comprising a trained agent, which is configured to determine a task configuration having a high utility and to set the task configuration as a record on a job list.