Over-the-Air Aggregation Lets Non-Connected UEs Join Federated Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless communication systems face challenges in enabling non-connected user equipment (UEs) to participate in federated learning procedures due to the need for Radio Resource Control (RRC) connected mode, which increases overhead and delays.
Innovation Solution
Implementing over-the-air (OTA) aggregation techniques that allow non-connected UEs to participate in federated learning without switching to RRC connected mode by using power control instructions based on distance or path loss, enabling gradient value transmission with consistent power levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If non-connected UEs participate in federated learning using traditional RRC connected mode, then reliable communication is established, but overhead and delay increase significantly
Solution Approach 1:
The patent extracts the essential function of federated learning participation from the RRC connected mode, allowing non-connected UEs to contribute gradient values directly through OTA aggregation without establishing full RRC connections. This separation removes the overhead of connection establishment while preserving the core functionality of model training contribution.
Solution Approach 2:
The patent introduces an intermediary mechanism where the network node collects gradient values from non-connected UEs through downlink signals and performs OTA aggregation, then transmits aggregated results to the server. This intermediary approach enables reliable data collection from non-connected devices without requiring direct RRC connections to the server.
2Ease of manufacture
If non-connected UEs transmit gradient values without power control, then transmission simplicity is maintained, but power level inconsistencies cause aggregation errors
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting transmit power levels based on distance or path loss measurements. Each non-connected UE modifies its transmission parameter (power level) according to its specific channel conditions, enabling consistent gradient aggregation at the network node while maintaining the simplicity of non-connected operation mode.
3Reliability
If all UEs switch to RRC connected mode for federated learning, then communication reliability is improved, but system overhead and resource consumption increase
Solution Approach 1:
The patent segments the UE population into connected and non-connected groups, applying different participation mechanisms to each segment. Connected UEs use traditional RRC mode when available, while non-connected UEs use the optimized OTA aggregation mechanism. This segmentation allows the system to maintain reliability where needed while avoiding unnecessary overhead for devices that can participate in non-connected mode.
Data Source
AI summary
Methods, systems, and devices for wireless communications are described. A user equipment (UE) may receive a first message requesting the UE to perform a federated learning procedure. The first message may indicate a machine learning model and a configuration for the federated learning procedure. The UE may perform, in response to the first message, a training procedure using the machine learning model to obtain one or more model parameters based on the configuration for the federated learning procedure. The UE may transmit a second message indicating one or more gradient values for the one or more model parameters via one or more resources configured for over-the-air (OTA) aggregation based on the configuration for the federated learning procedure.