AI Planning System for Self-Organizing MEC Applications
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The 5G network lacks self-organizing functionality for Multi-Access Edge Computing (MEC) systems, leading to increased manual development and maintenance costs due to the need for customized applications for user groups, unlike conventional public cloud applications.
Innovation Solution
An artificial intelligence planning method and device that uses a problem file and domain file to construct planning trees based on deep learning models, generating scheduling plans to guide specific applications from initial to target states, enabling self-configuration, self-optimization, and self-healing capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If manual development and maintenance of customized applications is performed for each user group, then application functionality and user-specific requirements are satisfied, but development time and costs increase significantly
Solution Approach 1:
The patent implements self-service through automated planning systems that enable MEC applications to automatically adapt to different user groups and network conditions. The system uses planning trees and deep learning models to autonomously generate customized application configurations without manual intervention, allowing the application to serve itself across multiple user scenarios.
Solution Approach 2:
The patent creates a universal planning framework that can handle multiple user groups and application types through a single system. The planning tree structure and domain files enable one system to generate customized solutions for diverse scenarios including video streaming, online gaming, and telemedicine, eliminating the need for separate manual development for each user group.
2Adaptability or versatility
If manual development and maintenance of customized applications is performed for each user group, then application functionality and user-specific requirements are satisfied, but maintenance costs increase significantly
Solution Approach 1:
The system enables automated maintenance through the planning framework that can automatically detect changes in network conditions or user requirements and regenerate appropriate configurations. The self-service mechanism reduces maintenance burden by automatically handling updates, optimizations, and adaptations without requiring manual developer intervention for each change.
Solution Approach 2:
The patent implements dynamic adaptation through planning trees that can be automatically updated and regenerated based on changing conditions. The system maintains customized applications dynamically by continuously evaluating network states and user requirements, automatically adjusting configurations to maintain optimal performance without manual maintenance.
3Ease of operation
If conventional public cloud application models are applied to MEC systems, then deployment simplicity is maintained, but user-specific customization and local optimization capabilities are lost
Solution Approach 1:
The patent segments the application deployment into multiple planning layers (application layer, service layer, infrastructure layer) that can be independently configured and optimized. This segmentation allows automated customization at each layer while maintaining overall deployment simplicity through the structured planning framework that guides configuration across all layers systematically.
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
An artificial intelligence planning method and an artificial intelligence planning device (100) are provided. When a considered specific application (173) has a plurality of planning layers, a corresponding planning node set (310, 330, 350) and a planning tree (500) are built for each of the planning layers according to a domain file (Fl), and all planning nodes (311-31M, 331-33N, 351-35O) in the planning node set (310, 330, 350) of each of the planning layers satisfy a plurality of corresponding preset conditions. In addition, a corresponding planning path (PI, P2) can be determined in each of the planning trees (500), and a plurality of action commands (al-a3, a11-a13, A1-AZ) can be accordingly issued to the specific application (173) to achieve a corresponding target state (eZ). In this way, the specific application (173) having multiple planning layers can be equipped with self-configuration, self-optimizing, and self-healing functions and is then implemented as a self-organizing application.