Federated Learning Update Reporting Under Radio Interference

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Federated learning systems face challenges in efficient communication between agents and a centralized parameter server due to interference, especially when using analog modulation, which can be mitigated by dedicated agent-to-PS channels but at the cost of increased network resources and latency.

Innovation Solution

Implement a linear mapping selected based on estimated interference levels for the radio propagation channel, allowing agents to report computational results to the server, with the server applying an inverse mapping to aggregate these results efficiently.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If dedicated agent-to-PS channels are used for transmission of model updates, then communication reliability is improved, but network resources and computational resources are increased

Engineering Contradiction:

Improvecommunication reliabilityVSAvoidnetwork resources and computational resources

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent combines multiple model updates from different agents into a single aggregated transmission by computing their sum at the PS. This merging approach allows multiple updates to be transmitted over shared channels rather than requiring dedicated channels for each agent, reducing network resources while maintaining communication reliability through aggregation

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The PS is designed to handle multiple functions: it acts as both a computation node that aggregates model updates and as a communication coordinator that manages shared channel allocations. This multi-functionality allows the system to achieve reliable communication without requiring separate dedicated channels for each agent

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Device complexity

If analog modulation with over-the-air computation is used, then network resources are reduced, but communication is susceptible to interference

Engineering Contradiction:

Improvenetwork resourcesVSAvoidinterference susceptibility

Core Design Contradiction:

Device complexityVSObject-affected harmful factors

Solution Approach 1:

The patent introduces digital modulation and coding as intermediary mechanisms between the analog over-the-air computation and the final model update aggregation. These intermediary techniques process the received signals to mitigate interference effects before computing the final aggregated update, allowing the system to benefit from both resource efficiency and interference robustness

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system dynamically adjusts transmission parameters such as modulation schemes and coding rates based on channel conditions and interference levels. By changing these parameters adaptively, the system can maintain resource efficiency while becoming more resilient to interference when conditions warrant it

Inventive Principle:

Principle #35Parameter changes

3Device complexity

If model updates are transmitted using shared channels, then network resources are saved, but communication latency increases due to interference

Engineering Contradiction:

Improvenetwork resourcesVSAvoidcommunication latency

Core Design Contradiction:

Device complexityVSLoss of time

Solution Approach 1:

The patent replaces traditional mechanical time-division multiplexing approaches with signal processing-based interference mitigation techniques. Instead of allocating different time slots to different agents (which increases latency), the system uses digital modulation and coding to handle simultaneous transmissions, reducing communication latency while maintaining resource efficiency

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS12627573B2Server and agent for reporting of computational results during an iterative learning process

Publication Date: 2026.05.12 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)

AI summary

There is provided mechanisms for performing an iterative learning process with agent entities. A method is performed by a server entity. The method includes selecting a linear mapping to be used by the agent entities when reporting computational results of a computational task to the server entity. The linear mapping has an inverse. The linear mapping is selected as a function of an estimated interference level for a radio propagation channel over which the agent entities are to report the computational results of the computational task to the server entity. The method includes configuring the agent entities with the computational task. The method includes performing the iterative learning process with the agent entities until a termination criterion is met. The server entity as part of performing the iterative learning process applies the inverse of the linear mapping to a sum of the computational results.