Wireless Distributed Learning Terminal Grouping for Abnormal Device Filtering

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In wireless distributed learning systems, abnormal operations from some wireless communication devices can degrade the overall learning performance by affecting the learning data, necessitating a method to filter out the influence of such devices.

Innovation Solution

A base station identifies and filters abnormal wireless communication devices by creating multiple terminal groups, allocating different resources to each group, and identifying a final group with high performance learning data, thereby minimizing the impact of abnormal devices on the system update.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If wireless distributed learning is performed using multiple terminals, then learning performance can be improved through aggregated data, but abnormal terminals can degrade overall system performance by contaminating learning data

Engineering Contradiction:

Improvelearning performanceVSAvoidabnormal terminal influence

Core Design Contradiction:

ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The base station divides terminals into multiple groups (first terminal group and second terminal group) and performs distributed learning separately for each group. This segmentation isolates abnormal terminals within one group from affecting the other group's learning results, thereby maintaining overall system reliability while still utilizing multiple terminals for learning.

Inventive Principle:

Principle #1Segmentation

2Measurement precision

If all terminals are used for distributed learning, then data diversity and learning accuracy are improved, but abnormal terminals can corrupt the learning results

Engineering Contradiction:

Improvelearning accuracyVSAvoidlearning result integrity

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The base station identifies and extracts abnormal terminals from the terminal population by analyzing learning data characteristics. These abnormal terminals are then separated into a distinct group, preventing their contaminated data from corrupting the overall learning results while still allowing normal terminals to contribute to accurate learning.

Inventive Principle:

Principle #2Taking out (Extraction)

3Reliability

If multiple terminal groups are created to filter abnormal terminals, then system performance is protected, but device complexity increases

Engineering Contradiction:

Improvedistributed learning performanceVSAvoidbase station processing complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The base station applies different processing qualities to different terminal groups: simple aggregation for normal terminals and filtered processing for groups containing abnormal terminals. This local differentiation allows the system to maintain high reliability for normal learning while using simplified processing, reducing overall complexity compared to uniform complex processing of all terminals.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS11659437B2Wireless distributed learning system including abnormal terminal and method of operation thereof

Publication Date: 2023.05.23 SEOUL NATIONAL UNIVERSITY R&DB FOUNDATION

AI summary

An electronic device is provided. The electronic device includes a communication circuit and a processor. The processor may be configured to obtain information on the number of predicted abnormal terminals, allocate different resources respectively to a plurality of terminal groups, wherein the number of the plurality of terminal groups is greater than the number of predicted abnormal terminals, obtain learning data of each of the plurality of terminal groups, and identify a final terminal group among the plurality of terminal groups, based on the learning data.