Network Slice Feasibility via ML Resource Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wireless network technologies face challenges in efficiently managing network slices due to inadequate resource allocation prediction, leading to potential over-dimensioning or under-dimensioning of physical resource blocks (PRBs), which can result in spectrum loss or SLA violations.
Innovation Solution
A device associated with service management in a wireless network is configured to receive requests for network slices, predict PRB allocations based on Service Level Agreement (SLA) parameters and observed network conditions, and output recommendations for approval or rejection, using machine learning models trained with network snapshots and performance indicators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional resource allocation methods are used for network slices, then network management is simpler, but resource allocation accuracy deteriorates leading to over-dimensioning or under-dimensioning of PRBs
Solution Approach 1:
The service management device performs preliminary prediction of PRB allocations using machine learning models before actual network slice deployment. By predicting resource requirements in advance based on SLA parameters and network conditions, the system avoids over-dimensioning or under-dimensioning of physical resource blocks, thereby improving resource allocation accuracy while maintaining manageable complexity through automated preprocessing.
Solution Approach 2:
The patent replaces traditional mechanical/resource-based allocation methods with machine learning-based prediction systems. The service management device uses trained ML models to automatically predict PRB allocations, substituting manual or rule-based resource management with intelligent algorithms that adapt to changing network conditions, thus improving prediction accuracy without linearly increasing system complexity.
2Loss of energy
If PRB allocation is optimized for better spectral efficiency, then spectrum loss is reduced, but the complexity of resource management increases
Solution Approach 1:
The service management device implements feedback mechanisms where machine learning models continuously learn from actual network performance and PRB utilization patterns. The system compares predicted allocations with actual outcomes, uses this feedback to refine predictions, and adjusts future resource allocations accordingly. This closed-loop approach reduces spectrum loss through progressively improved accuracy while the automated feedback processing keeps management complexity manageable.
Solution Approach 2:
The machine learning models perform self-optimization by automatically adjusting predictions based on learned patterns from historical data and current network conditions. The service management device enables itself to improve resource allocation accuracy over time without requiring manual intervention for each optimization decision, thereby reducing spectrum loss while preventing management complexity from escalating.
3Reliability
If network slice requests are processed with detailed SLA parameters, then service quality is improved, but processing time and complexity increase
Solution Approach 1:
The service management device performs preliminary filtering and pre-assessment of network slice requests using machine learning models before detailed processing. By predicting resource requirements and feasibility in advance based on SLA parameters and current network conditions, the system identifies promising candidates early, enabling faster decision-making while maintaining high SLA compliance through thorough evaluation of pre-selected requests.
Solution Approach 2:
The system applies partial processing to routine slice requests by using pre-trained machine learning models to make quick predictions without full detailed analysis. For non-critical or standard SLA parameters, the system performs sufficient but not excessive evaluation, achieving acceptable SLA compliance with reduced processing time. Full detailed processing is reserved only for complex or high-stakes slice requests.
Data Source
AI summary
This disclosure provides systems, methods and apparatus, including computer programs encoded on computer storage media, for network slice feasibility assessment for slice orchestration in a wireless network. Some aspects relate to providing on-demand approval or rejection of a requested network slice at a device associated with service management. The device may select respective resource allocations of the requested network slice for each cell of a set of cells in a wireless network, add the respective predicted resource allocations to respective current resource utilizations at each of the cells, and output a recommendation associated with the requested network slice in accordance with the summations. The device may select the respective resource allocations of the requested network slice in accordance with a service level agreement (SLA) of the requested network slice and, in some implementations, observed network conditions at each of the cells in the wireless network.