Dynamic AI Model Split Processing for Edge Computing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current mobile edge computing methods for AI model processing face challenges in optimizing learning speed and reliability due to static split processing of models across cloud servers, edge servers, and user equipment, which do not dynamically adapt to computing node conditions or requirements.
Innovation Solution
A control apparatus and method that dynamically control split processing of AI models by obtaining information on computing nodes, including computing latency and accuracy requirements, to determine optimal split points and data transmission between nodes, thereby enhancing learning speed and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If static split processing of models is used across cloud servers, edge servers, and user equipment, then device complexity is reduced and ease of operation is improved, but learning speed and reliability deteriorate due to inability to adapt to computing node conditions
Solution Approach 1:
The patent implements dynamic split processing by allowing the split point of the AI model to be adjusted based on real-time conditions of computing nodes. The control apparatus determines optimal split points by evaluating computing capabilities, latency requirements, and accuracy requirements of different nodes, transforming the static split configuration into a dynamic adaptive system that responds to changing conditions.
Solution Approach 2:
The control apparatus receives feedback information from computing nodes including computing capability data, latency measurements, and accuracy metrics. This feedback loop enables the system to continuously evaluate performance and adjust the split processing configuration accordingly, improving learning reliability through data-driven optimization rather than fixed predetermined settings.
2Productivity
If dynamic split processing control is implemented based on computing node information, then learning speed and reliability improve, but device complexity and information processing requirements increase
Solution Approach 1:
The control apparatus segments the AI model into multiple parts that can be distributed across different computing nodes. By dividing the model processing into discrete segments that can be independently configured and assigned to specific nodes based on their capabilities, the system achieves faster parallel processing while managing complexity through modular organization of model components.
Solution Approach 2:
The patent assigns different parts of the AI model to different computing nodes based on their specific local qualities or characteristics. Each node processes the segments for which it is best suited, with the control apparatus matching model segments to nodes according to their computing capabilities, latency performance, and accuracy requirements, thereby optimizing overall learning speed.
3Adaptability or versatility
If model split points are fixed, then ease of operation is maintained, but adaptability to different computing node conditions deteriorates
Solution Approach 1:
The control apparatus automatically determines optimal split points and configurations by evaluating computing node conditions without requiring manual intervention. The system self-adjusts the model segmentation and node assignment based on received feedback about computing capabilities, latency, and accuracy requirements, providing adaptability while maintaining ease of operation through automation.
Data Source
AI summary
An apparatus and method for split processing of a model are provided. The apparatus for the split processing of the model includes a memory including instructions and a processor electrically connected to the memory and configured to execute the instructions. When the instructions are executed by the processor, the processor may be configured to perform a plurality of operations. The plurality of operations may include obtaining information on a plurality of computing nodes that uses at least one layer among a plurality of layers of a model for an artificial intelligence (AI)-based service, obtaining a requirement for the AI-based service, and controlling split processing of the model based on the information and the requirement.