Federated Learning QoS Allocation in 5G Wireless Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In federated learning, the overall learning speed is determined by the slowest learning speed of participating terminals due to varying network conditions, necessitating efficient allocation of network resources to optimize the learning process.
Innovation Solution
A 5G mobile communication operator provides information on terminals that can receive the same QoS for federated learning traffic transmission through an application function (AF), utilizing the 5G network's capabilities to allocate network resources effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If terminals with varying network conditions participate in federated learning, then the number of participating terminals increases, but the overall learning speed decreases due to the slowest terminal determining the synchronization pace
Solution Approach 1:
The patent segments terminals into different groups based on their network conditions and QoS capabilities. Terminals are divided into those that can guarantee minimum QoS requirements and those that cannot, allowing the system to selectively include only capable terminals in federated learning tasks, thus maintaining learning speed while enabling broader participation.
Solution Approach 2:
The patent dynamically adjusts QoS parameters (bandwidth, latency, packet loss thresholds) based on terminal network conditions. By changing these parameters adaptively, the system can accommodate more terminals with varying network qualities while ensuring that the overall federated learning process maintains acceptable performance levels.
2Ease of operation
If network resources are allocated without QoS consideration, then resource allocation simplicity increases, but federated learning performance deteriorates due to inconsistent learning speeds across terminals
Solution Approach 1:
The patent performs preliminary QoS assessment and terminal capability evaluation before assigning federated learning tasks. By pre-evaluating terminal network conditions and QoS guarantees, the system can make informed allocation decisions that balance simplicity with performance, avoiding the need for complex real-time adjustments during the learning process.
Solution Approach 2:
The patent introduces a network controller or edge server as an intermediary that manages QoS allocation between terminals and federated learning tasks. This intermediary handles the complexity of QoS management centrally, allowing individual terminals to participate without needing to implement complex QoS negotiation, thus maintaining ease of operation while improving overall performance.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided is a 5G or 6G communication system for supporting higher data rates. Provided are a method and device for supporting federated learning in a wireless communication system. The method comprises: receiving, from an AF, a first request including a list of UE addresses, QoS reference or individual QoS parameters, and alternative service requirements; and transmitting, to the AF, a first response message including a result for the list of UE addresses in response to receiving to the first request. The result for the list of UE addresses includes information indicating whether the first request is granted or not for UE addresses in the list of UE addresses.