Cellular Network User Clustering to Reduce Data Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current cellular networks lack a powerful and compact representation of user entities, failing to exploit strong similarities in activity patterns, leading to inefficient decision-making due to the need for large data collection and processing.
Innovation Solution
A method for grouping user entities into clusters based on activity parameters, using unsupervised learning to generate user profiles and group UEs with similar characteristics, allowing for reduced data storage and processing while enabling advanced network control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If large amounts of user data are collected and stored to enable accurate user representation and decision-making, then measurement precision and reliability improve, but device complexity and loss of substance increase due to expensive hardware requirements
Solution Approach 1:
The patent merges multiple user entities with similar activity patterns into clusters, combining their characteristics into cluster-specific user profiles. This allows the system to represent many users through a limited number of cluster profiles, reducing the need for individual user data storage and processing while maintaining accurate user representation.
Solution Approach 2:
Instead of representing each user individually with full data sets, the patent inverts the approach by creating cluster profiles that represent groups of users. The system stores cluster characteristics and activity patterns rather than individual user data, achieving efficient representation with reduced hardware requirements.
2Measurement precision
If individual user profiles are maintained for all user entities, then measurement precision improves, but loss of information increases due to inability to exploit user similarities
Solution Approach 1:
The patent merges user entities into clusters based on similar activity patterns, combining individual user characteristics into cluster-level representations. This preserves important user similarities while reducing data redundancy, allowing the system to exploit user pattern similarities for more efficient information utilization.
3Loss of substance
If cluster-based grouping is implemented to reduce data storage, then loss of substance decreases, but measurement precision may worsen due to aggregation
Solution Approach 1:
The patent applies local quality by maintaining cluster-specific user profiles that capture the distinctive characteristics of each cluster. Each cluster profile contains activity parameters and patterns specific to that group, ensuring that aggregated representations still preserve the essential qualities that differentiate various user types.
Solution Approach 2:
The clustering approach is dynamic, allowing users to be grouped based on their activity patterns rather than static attributes. The system can adapt cluster compositions as user behaviors change, maintaining measurement precision while benefiting from reduced data storage through clustering.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for grouping a plurality of user entities connected to a cellular network into different clusters, the method comprising: - collecting, for each of the plurality of user entities, activity information comprising different activity parameters, each activity parameter describing an activity of the corresponding user entities in the cellular network, - generating, for each of the plurality of user entities, a user profile comprising the activity parameters for the corresponding user entity, - grouping the plurality of user entities into the different clusters based on a similarity of the activity parameters of the user profiles.