ML Model Predicting Channel Interference for Cellular Network Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile network operators face challenges in optimizing cellular network performance due to high processing resource usage and delays when directly applying predefined models for channel interference prediction during on-the-fly reconfiguration, which affects handover events and channel interference metrics.
Innovation Solution
A machine learning model is trained using a training set generated from simulated radio propagation data to predict channel interference indicators, reducing processing burden and enabling efficient network optimization without the need for real-time application of complex propagation models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If predefined models for channel interference prediction are directly applied during on-the-fly reconfiguration, then prediction accuracy is maintained, but processing resource usage increases and delays occur
Solution Approach 1:
The patent pre-calculates and stores channel interference indicators for multiple possible network configurations before actual reconfiguration events occur. When reconfiguration is needed, the pre-computed indicators are directly retrieved and applied, eliminating the need for real-time complex calculations. This preliminary action resolves the contradiction by maintaining prediction accuracy through pre-computed precise models while achieving fast reconfiguration through simple lookup operations.
Solution Approach 2:
The system prepares multiple pre-computed channel interference indicators for different potential reconfiguration scenarios in advance, creating a buffer of ready-to-use prediction results. This cushioning approach ensures that when reconfiguration events occur, accurate predictions are immediately available without requiring real-time computation, thus maintaining both accuracy and speed.
2Measurement precision
If complex propagation models are applied in real-time, then accurate channel interference indicators are obtained, but processing delays increase
Solution Approach 1:
The patent performs complex propagation model calculations in advance during off-line or idle periods, storing the results as pre-computed channel interference indicators. During actual network operation and reconfiguration events, these pre-computed indicators are directly applied without re-running the complex models, thereby maintaining high accuracy while eliminating processing delays.
Solution Approach 2:
Instead of executing the complex propagation model in real-time, the system creates and uses copies of pre-computed channel interference indicators for different network configurations. These copies are generated beforehand through detailed modeling and then reused during actual operations, preserving the accuracy of the original complex model calculations while avoiding their computational overhead during time-critical operations.
3Reliability
If traditional network optimization methods are used, then comprehensive analysis is performed, but processing resource usage increases
Solution Approach 1:
The patent pre-computes channel interference indicators for multiple network configurations during periods when processing resources are more available, storing these results for later use. This shifts the computational burden to off-peak times, reducing processing resource usage during critical network operations while maintaining comprehensive analysis capability through the pre-computed indicators.
Solution Approach 2:
The system creates pre-computed copies of channel interference analysis results for various network configurations. These copies encapsulate the outcomes of comprehensive traditional analysis methods but can be applied with minimal processing resources. By using these pre-generated copies instead of re-running comprehensive analyses, the system maintains reliable network optimization while dramatically reducing processing resource consumption during operational decision-making.
Data Source
AI summary
Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a network node may obtain a machine learning (ML) model trained to provide one or more predicted channel interference (CI) indicators informative of channel interference between cells of a cellular network. The network node may calculate, using the ML model, the one or more predicted CI indicators using data characterizing a given cell and one or more neighbor cells of the given cell. The network node may provide the one or more predicted CI indicators. Numerous other aspects are described.