Unified Vector Repository for Subscriber-Aware Telco LLM Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large language models in telecommunications networks primarily focus on general use cases without considering subscriber-related data or behavior, limiting their effectiveness in enhancing network service levels and efficiency.
Innovation Solution
Implement a unified vector repository entity that stores subscriber-specific embeddings, allowing for intelligent use of large language models by embedding subscriber data into queries, thereby enabling behavior-specific decisions and results, and integrating this entity within the 5G core network's unified data management layer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If large language models are used for general use cases without subscriber-related data, then the implementation is simpler and more cost-effective, but the network service level improvement and network efficiency enhancement are limited
Solution Approach 1:
The patent implements a nested architecture where the unified vector repository is integrated within the 5G core network's unified data management layer. The vector repository contains subscriber embeddings that are nested within the existing network architecture, allowing LLMs to access subscriber-specific data without requiring complete system redesign. This nested structure enables progressive enhancement from general to personalized applications.
Solution Approach 2:
The patent performs preliminary action by pre-processing and embedding subscriber-related data into the unified vector repository before the LLM queries are executed. Subscriber profiles, behavior patterns, and network data are transformed into vector embeddings in advance, stored in the vector repository, and then retrieved during query processing. This pre-processing enables faster and more accurate personalized responses.
2Productivity
If subscriber-specific data is embedded into LLM queries, then behavior-specific decisions and network efficiency improve, but the system complexity and data management requirements increase
Solution Approach 1:
The unified vector repository serves multiple functions: it stores subscriber profiles, behavior patterns, network data, and provides vector embeddings for various LLM applications. This multi-functional repository reduces the need for separate data storage systems for different purposes, managing complexity through consolidation rather than proliferation of separate systems.
Solution Approach 2:
The unified vector repository acts as an intermediary between the existing 5G core network data management system and the LLM application layer. It transforms diverse subscriber data into standardized vector embeddings that the LLM can process, while maintaining compatibility with existing network data structures. This intermediary layer shields the complexity of data transformation from both the network core and the LLM application.
3Adaptability or versatility
If a unified vector repository is integrated into the 5G core network, then subscriber-specific embeddings can be accessed for personalized services, but the integration complexity with existing network functions increases
Solution Approach 1:
The patent segments the data management functionality by creating a distinct unified vector repository component within the 5G core network architecture. This segmented approach allows the vector repository to be developed, maintained, and scaled independently from other network functions, reducing integration complexity. The repository can be implemented as a separate microservice or module that interfaces with existing network functions through standardized APIs.
Data Source
Figure 1~2
Figure 3~5
AI summary
The invention relates to a method for large language model usage, especially for realizing network service level improvement and/or for enhancing network efficiency within or related to a mobile communication network, wherein the mobile communication network comprises a radio access network and a core network, wherein the mobile communication network and/or the core network comprises a plurality of network functions or services, wherein the plurality of network functions or services at least comprises a telco application agent and a unified vector repository entity or functionality, wherein, in order to realize large language model usage, especially associated with network service level improvement and/or an increase of network efficiency within or related to the mobile communication network, the telco application agent exchanges messages with the unified vector repository entity or functionality, wherein the method comprises the following steps: -- in a first step, the unified vector repository entity or functionality receives a unified vector repository request message, -- in a second step, the unified vector repository entity or functionality transmits a unified vector repository response message.