Intelligent scheduling method and system based on distributed computing resources

By calculating the load phase divergence coefficient and hysteresis comparison interval, the resource allocation strategy is dynamically adjusted, which solves the problems of invalid computation and mode jitter in the chaotic state of edge computing node load, and achieves high efficiency in resource utilization and stability in scheduling.

CN122363941APending Publication Date: 2026-07-10QUANZHOU INST OF INFORMATION ENG
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-06-08
Publication Date
2026-07-10

AI Technical Summary

Technical Problem

When the load on edge computing nodes is in a highly chaotic state, existing scheduling methods cannot effectively avoid the ineffective consumption of computing resources and frequent pattern jitter of deep learning prediction models, leading to scheduling continuity and stability issues.

Method used

By calculating the load phase divergence coefficient and introducing a hysteresis comparison interval, combined with lightweight threshold scheduling, the resource allocation strategy is dynamically adjusted to avoid the ineffective operation of the deep prediction model. When the load is chaotic, it switches to lightweight scheduling to maintain the continuity and stability of scheduling.

Benefits of technology

It effectively reduces the burden of invalid computation, improves the resource utilization efficiency of edge nodes, reduces task erroneous eviction and invalid resource reservation, and enhances the continuity and stability of scheduling strategies.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122363941A_ABST
    Figure CN122363941A_ABST
Patent Text Reader

Abstract

This invention relates to the field of distributed computing resource scheduling technology, specifically disclosing a method and system for intelligent scheduling of distributed computing resources. The method involves collecting historical load data from edge nodes to form a time-series queue, reconstructing it into a high-dimensional set of phase points, and calculating the load phase divergence coefficient as a chaos discrimination value. This value is compared with the rising and falling switching thresholds in the hysteresis comparison interval, and updated to either a chaotic mode or a predictable mode flag based on the current mode flag. In the chaotic mode, resource reservation or task eviction is performed based on the comparison of real-time load with high and low water level thresholds. In the predictable mode, load prediction is derived through deep lightweight convolutional mapping, and scheduling is then performed according to the same threshold. This invention can adaptively switch scheduling strategies according to the degree of load chaos, avoiding ineffective prediction overhead and suppressing mode jitter, making it suitable for highly dynamic distributed environments such as edge computing.
Need to check novelty before this filing date? Find Prior Art