ML-Based Secondary Cell Selection for Wireless Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wireless communication systems face challenges in optimizing secondary cell selection for user equipment (UE) in carrier aggregation and dual/multi-connectivity scenarios, as they lack direct information on the achievable channel quality of secondary cells, leading to suboptimal selection and increased overhead in inter-frequency measurements.
Innovation Solution
A method using a machine learning algorithm, specifically a deep neural network, to predict the achievable channel quality for user equipment on secondary cells based on measurements from the primary cell, allowing for informed secondary cell selection without the need for extensive inter-frequency measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional inter-frequency measurements are performed to determine secondary cell quality, then cell selection accuracy is improved, but measurement overhead and battery consumption increase
Solution Approach 1:
The patent introduces a machine learning model as an intermediary that predicts secondary cell channel quality based on primary cell measurements. This mediator translates readily available primary cell measurement data into accurate secondary cell quality estimates without requiring direct secondary cell measurements, thus reducing measurement overhead while maintaining selection accuracy
Solution Approach 2:
The system performs preliminary machine learning training offline to establish the relationship between primary and secondary cell channel qualities. This preliminary action prepares the prediction model in advance, enabling accurate secondary cell quality estimation during runtime without requiring extensive real-time inter-frequency measurements
2Loss of information
If extensive inter-frequency measurements are performed for secondary cell selection, then channel quality information is improved, but battery consumption increases
Solution Approach 1:
The patent creates a predictive copy of secondary cell channel quality information by training a machine learning model on available data. This copied information, generated through ML prediction rather than direct measurement, provides sufficient channel quality data for cell selection while avoiding the energy-intensive process of actual inter-frequency measurements
3Quantity of substance
If machine learning prediction is used for secondary cell selection, then measurement overhead is reduced, but algorithm complexity increases
Solution Approach 1:
The patent transforms the cell selection problem by changing the parameters used for decision-making. Instead of using raw measurement quantities, the system uses machine learning model predictions as the decision parameter. This parameter change reduces the need for extensive measurements while the model complexity is managed through efficient architecture design and offline training
Data Source
AI summary
UE-related measurements taken on a Pcell in a wireless communication system are formed into a set of data. The Pcell overlaps with Scell(s). The UE-related measurements on the Pcell are for a specific UE in the Pcell. Using a ML algorithm applied to the set of data, achievable channel quality is predicted for the specific UE for each of the Scell(s). The predicted achievable channel qualities are output for the specific UE to be used for Scell selection. At a RAN node, the set of data is sent toward an Scell prediction module for the module to determine information suitable to enable Scell selection for the specific UE. The RAN node receives information from the module allowing the RAN node to inform the selected UE of Scell(s) to be used for Scell selection for the specific UE. A node may train the ML algorithm using UE-related measurements on the Pcell.