ML Architecture Selection for Network Slice QoS
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless communication networks face challenges in meeting diverse quality-of-service (QOS) levels requested by various applications, such as voice calling, web access, and video streaming, due to varying user equipment capabilities.
Innovation Solution
A machine-learning (ML) architecture is determined for network slicing by communication between user equipment (UE) and a network-slice manager, allowing the UE to select an ML architecture that meets the requested QOS level, and the network-slice manager to accept or reject the request based on available end-to-end ML architectures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a single wireless communication network is used to meet diverse quality-of-service levels, then network simplicity is maintained, but the ability to satisfy diverse QOS requirements deteriorates
Solution Approach 1:
The network is segmented into multiple network slices, each optimized for specific QOS requirements (e.g., voice calling, web access, video streaming). Each slice operates independently with dedicated resources, allowing the system to meet diverse QOS levels without requiring complete network redesign. The segmentation enables targeted optimization for different service types while maintaining overall network manageability.
Solution Approach 2:
The network slice manager dynamically selects and configures machine-learning architectures based on real-time QOS requirements and available resources. This dynamic adaptation allows the network to flexibly respond to changing service demands, selecting appropriate ML models and architectures without manual reconfiguration, thereby maintaining versatility while controlling complexity through automated management.
2Reliability
If machine-learning architectures are selected to meet specific QOS levels, then QOS satisfaction is improved, but the complexity of architecture selection and management increases
Solution Approach 1:
The system implements self-service through automated ML architecture selection and configuration. The network slice manager autonomously evaluates available ML architectures, selects the most appropriate one based on QOS requirements, and configures it without human intervention. This self-service approach ensures reliable QOS satisfaction while reducing management complexity by eliminating manual configuration processes.
Solution Approach 2:
The network slice manager continuously monitors QOS performance and uses feedback to dynamically adjust ML architecture selections. When QOS requirements change or performance degradation is detected, the system re-evaluates and reconfigures ML architectures accordingly. This feedback mechanism ensures sustained QOS satisfaction while simplifying management through automated closed-loop control.
Data Source
AI summary
This document describes techniques and devices for determining a machine-learning architecture. A user equipment (UE) transmits, to a network-slice manager of a wireless network, a first machine-learning architecture request message to request permission to use a first machine-learning architecture. The UE receives a first machine-learning architecture response message that grants permission to use the first machine-learning architecture based on a first network slice, the first machine-learning architecture forming a portion of at least one first end-to-end machine-learning architecture associated with the first network slice, the at least one first end-to-end machine-learning architecture being a distributed machine-learning architecture that is configured to process wireless communication signals and is formed by the first machine-learning architecture implemented by the user equipment, a machine-learning architecture implemented by a base station, and a machine-learning architecture implemented by an entity of a core network. The UE wirelessly communicates data using the first machine-learning architecture.