AI-Based UPF Selection for Adaptive Network Load Balancing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network function (NF) selection methods in communication systems struggle with inefficiencies in load balancing, particularly in complex network environments, where manual setting of selection ratios is difficult and fails to adapt to dynamic changes in network traffic patterns and NF additions or deletions.
Innovation Solution
An apparatus and method utilizing an artificial intelligence model to collect and analyze load information from NFs, enabling adaptive NF selection based on real-time data to optimize load balancing and dynamically adjust to network changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If manual setting of selection ratios is used for NF selection, then the system is simple to operate, but it fails to adapt to dynamic changes in network traffic patterns and NF additions or deletions
Solution Approach 1:
The system enables NF selection to occur automatically based on real-time load information and AI predictions, without requiring manual configuration of selection ratios. The NF selection apparatus autonomously monitors load conditions, generates predictions, and makes selection decisions, allowing the system to self-adapt to dynamic network changes including traffic pattern variations and NF additions or deletions
Solution Approach 2:
The system transitions from static manual selection ratios to dynamic AI-driven selection that continuously adapts to changing network conditions. Load information is collected in real-time, predictions are generated dynamically, and selection ratios are automatically adjusted based on current network state, enabling the system to respond flexibly to dynamic changes while maintaining operational simplicity
2Productivity
If AI model is used for predictive NF selection, then load balancing is optimized, but computational resources and processing time increase
Solution Approach 1:
The system performs predictive analysis by generating future load predictions before actual load conditions occur. The AI model forecasts upcoming traffic patterns and NF load states, allowing the system to proactively make NF selections that optimize future load distribution rather than merely reacting to current conditions, thereby improving load balancing efficiency
Solution Approach 2:
The system collects and processes load information from multiple NFs and time intervals, using AI models to generate predictions. By utilizing partial information (selected NFs, specific time intervals) and AI-driven processing, the system achieves optimized load balancing while managing computational resource consumption through targeted rather than exhaustive analysis
3Measurement precision
If real-time load information from multiple NFs is collected and analyzed, then NF selection accuracy is improved, but data collection and processing complexity increases
Solution Approach 1:
The NF selection apparatus acts as an intermediary that centralizes the collection, analysis, and processing of load information from multiple NFs. By consolidating these functions in a dedicated apparatus, the system simplifies the complexity of managing distributed data collection while maintaining high selection accuracy through comprehensive analysis of load conditions across the network
Data Source
AI summary
An electronic device for a session management function (SMF) may comprise memory storing instructions and at least one processor. The instructions may cause the electronic device to: obtain a first load value for a second time interval before selecting a serving user plane function (UPF), estimated based on an artificial intelligence model (AI model) using first load information of each of UPFs measured within a first time interval before the second time interval, obtain a second load value for the second time interval, calculated by using second load information of each of the UPFs measured within the second time interval, determine a difference between the first load value and the second load value, and determine, using the difference, whether to use a predicted load value of each of the UPFs obtained based on the AI model to select the serving UPF from among the UPFs.