Dynamic Frequency Resource Allocation via Reinforcement Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current cellular network technologies, such as FFR, face challenges in adapting to varying user densities due to fixed frequency resource allocation, leading to insufficient spectrum usage in high-demand areas and inefficient spectral efficiency and interference management.
Innovation Solution
A method employing a reinforcement learning algorithm to dynamically allocate frequency resources across regions of a cell, allowing access to a shared set of frequency resources and updating resource values based on communication performance, enabling flexible resource prioritization and interference management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If FFR technique is used to limit interference at cell edges, then interference between adjacent cells is reduced, but spectral efficiency deteriorates due to fixed frequency allocation that cannot adapt to varying user densities
Solution Approach 1:
The patent applies dynamics by transitioning from static frequency allocation (FFR with fixed re-use factors) to dynamic frequency resource allocation. The system continuously monitors communication quality metrics and user density, then adjusts frequency resource assignment in real-time. This allows the network to adapt frequency allocation to varying user densities and interference conditions, resolving the contradiction between interference reduction and spectral efficiency.
Solution Approach 2:
The patent implements feedback mechanisms where the system monitors communication quality (such as SINR measurements) and user distribution patterns, then uses this information to adjust frequency resource allocation. The feedback loop enables the system to learn from past allocations and optimize future assignments, allowing simultaneous achievement of interference reduction and high spectral efficiency through adaptive resource management.
2Object-affected harmful factors
If frequency resources are partitioned into central and peripheral areas with fixed re-use factors, then interference is managed, but adaptability to varying user densities deteriorates
Solution Approach 1:
The patent makes the frequency resource allocation dynamic by allowing the system to adjust which frequencies are allocated to central versus peripheral areas based on real-time conditions. Instead of fixed re-use factors, the system can change frequency assignments adaptively when user density patterns change, maintaining both interference management and adaptability to varying traffic conditions.
Solution Approach 2:
The patent changes the parameter of frequency resource allocation from fixed to variable. The system monitors communication quality parameters and user distribution, then modifies frequency assignment parameters dynamically. This allows the re-use factor and frequency allocation to be adjusted according to actual network conditions, resolving the contradiction between structured interference management and adaptability.
3Adaptability or versatility
If the same set of frequency resources is made accessible to all regions, then adaptability to user density variations improves, but interference between adjacent cells worsens
Solution Approach 1:
The patent uses feedback control to monitor communication quality when the same frequency set is allocated to all regions. When interference levels exceed thresholds, the system adjusts frequency assignments for specific regions or users. This feedback mechanism allows the system to maintain flexible resource allocation while actively managing interference through quality-based adjustments.
Solution Approach 2:
The patent applies local quality by allowing different regions or users to have different frequency allocations based on their specific conditions. Even though the same frequency set is initially available to all regions, the system assigns frequencies locally based on interference measurements and user density, enabling each location to use the most appropriate frequencies for its conditions.
4Productivity
If reinforcement learning algorithm is used to dynamically allocate frequencies, then spectral efficiency and adaptability improve, but device complexity increases
Solution Approach 1:
The patent applies self-service by implementing a reinforcement learning algorithm that enables the system to automatically optimize frequency allocation without requiring complex manual configuration or external control. The algorithm learns optimal allocation strategies through interaction with the network environment, making the system self-adjusting and reducing the need for complex external management while achieving high spectral efficiency.
Data Source
AI summary
A method is described for allocating a frequency resource, from a set E of frequency resources, to at least one terminal positioned in a region of a cell belonging to a cellular network. The method includes, after the terminal sends a request for allocation of a frequency resource, identifying the cell and the region which are associated with said at least one terminal, verifying the availability of at least one frequency resource in a table comprising values respectively associated with the resources of the set E, and if at least one frequency resource is available, executing a reinforcement learning algorithm based on the value or values associated with said at least one available frequency resource, so as to select a resource for said at least one terminal.