Edge RAG Resource Adaptation for Secure Low-Latency Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing complexity of machine learning models and data management overhead have led to inefficient and insecure data processing, with cloud computing exposing data to risks and edge computing introducing inefficiencies due to unsolved supply oversupply issues.
Innovation Solution
A computer-implemented method that dynamically adapts the amount and type of resources sent to edge nodes based on real-time performance, using AI models to evaluate and output relevant subsets of RAG data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Power
If cloud computing is implemented to improve computational throughput, then processing capability is improved, but data security deteriorates due to centralized data exposure to attacks
Solution Approach 1:
The patent implements federated learning where different edge nodes train local model instances with locally stored data, creating localized processing capabilities. Each edge node maintains data security locally while contributing to global model improvement, resolving the contradiction between centralized computational power and data security.
2Object-affected harmful factors
If edge computing is implemented to improve data security, then data security is improved, but system efficiency deteriorates due to resource oversupply
Solution Approach 1:
The patent implements dynamic resource allocation where the central server adjusts the amount and type of data sent to each edge node based on real-time performance feedback. This dynamic adaptation allows edge nodes to operate efficiently with optimized resource levels, preventing both oversupply and undersupply of resources.
Solution Approach 2:
The patent establishes a feedback loop where edge nodes send performance information back to the central server, which then adjusts resource allocation accordingly. This feedback mechanism enables continuous optimization of resource distribution, improving system efficiency while maintaining data security at edge locations.
3Adaptability or versatility
If machine learning model complexity is increased to improve AI capability, then AI performance is improved, but processing overhead increases
Solution Approach 1:
The patent divides the global machine learning model into multiple local model instances distributed across different edge nodes. Each local model processes data locally, segmenting the computational workload and reducing the processing overhead at any single location while maintaining overall AI capability through coordinated training.
Data Source
AI summary
A computer-implemented method, according to one approach, includes: receiving information from an edge node, where the information outlines specific retrieval-augmented generation (RAG) data applied at the edge node, as well as a condition of the edge node, in real-time. A knowledge database which maps embeddings of RAG data to various edge node conditions is further updated with the received information. Moreover, one or more trained artificial intelligence based models are used to dynamically evaluate the received information and the knowledge database. The artificial intelligence based models are also used to output a relevant subset of RAG data. The relevant subset of RAG data is further sent to the edge node.