ML-Based Auto-Scaling for MEC Resource Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-access edge computing (MEC) networks often face insufficient resources due to the number of end devices and applications, leading to degraded performance metrics like latency, error rate, and packet loss, with existing auto-scaling techniques being static and inefficient in resource management.
Innovation Solution
A machine learning-based resource management service that dynamically adjusts vertical and horizontal scaling rules using network and end device information to optimize resource allocation, employing supervised, unsupervised, and reinforcement learning to continuously evaluate and adjust auto-scaling configurations without human intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If static auto-scaling rules are used, then resource allocation is simplified, but resource utilization efficiency deteriorates
Solution Approach 1:
The patent implements dynamic auto-scaling rules that automatically adjust resource allocation based on real-time network conditions, traffic patterns, and performance metrics. The system continuously monitors multiple parameters and dynamically modifies scaling decisions without manual intervention, transforming static rules into adaptive, condition-based policies that optimize resource utilization while maintaining manageable complexity through automated decision-making algorithms.
2Reliability
If more resources are allocated to accommodate all end devices, then service reliability improves, but resource waste increases
Solution Approach 1:
The system dynamically adjusts resource allocation parameters based on changing network conditions, traffic demands, and service requirements. By monitoring multiple parameters simultaneously (traffic volume, latency, error rates, device counts) and adjusting scaling decisions in response to parameter changes, the system ensures sufficient resources are allocated only when needed, maintaining service reliability while avoiding persistent over-provisioning and resource waste during low-demand periods.
Solution Approach 2:
The patent implements a feedback mechanism that continuously monitors service performance metrics and resource utilization levels, then uses this information to adjust auto-scaling decisions. The system evaluates whether current resource allocation is sufficient or excessive based on observed performance, and automatically modifies scaling policies to maintain optimal service reliability while minimizing resource waste through continuous adaptive adjustment.
3Measurement precision
If manual configuration of auto-scaling rules is used, then control precision is high, but response time to changing conditions deteriorates
Solution Approach 1:
The system enables self-service automated configuration of auto-scaling rules through machine learning algorithms that autonomously analyze network conditions, traffic patterns, and performance metrics to generate and adjust scaling policies without manual intervention. The ML models learn from historical data and real-time observations to make precise scaling decisions automatically, achieving both high control precision through sophisticated analysis and rapid response times by eliminating human configuration delays.
Solution Approach 2:
The patent implements preliminary action by pre-training machine learning models with historical network data and traffic patterns before deployment. This allows the system to have pre-configured knowledge of scaling behaviors for various conditions, enabling rapid response to changing network states without requiring manual rule configuration at the time of change. The preliminary ML model training establishes baseline scaling policies that can be quickly adjusted based on real-time conditions.
Data Source
AI summary
A method, a device, and a non-transitory storage medium are described in which a machine learning-based resource management service is provided. A network device obtains network and end device information, and uses machine learning to determine whether to adjust an auto-scaling rule pertaining to the provisioning of an application service. The network device generates a modified auto-scaling rule based on the analysis.