Neural Network Beam Failure Detection in 5G Terminals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for beam failure detection and recovery in New Radio (NR) systems are inadequate due to reliance on Layer-1 measurements, which are prone to noise and interference, leading to inaccurate determination of new candidate beam reference signals.
Innovation Solution
Implementing machine learning mechanisms, specifically neural networks, in terminal devices to detect beam failures and determine new candidate beam reference signals, considering current beam failure status to avoid selecting failed beams as new candidates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional Layer-1 measurement methods are used for beam failure detection, then the detection process is simple and fast, but the accuracy is poor due to noise and interference
Solution Approach 1:
The patent replaces the conventional mechanical measurement system (Layer-1 physical measurements) with an artificial intelligence-based system. Specifically, it uses a neural network model that processes multiple parameters including Layer-1 measurements (RSRP, SINR), Layer-2 measurements (BLER), and historical beam failure information to detect beam failures. This substitution eliminates the limitations of simple physical measurements by incorporating multiple data sources and intelligent processing, thereby significantly improving detection accuracy while managing system complexity through automated AI-based decision making.
2Measurement precision
If Layer-1 RSRP measurement is used to determine new candidate beam RS, then the process is straightforward, but the result is unsatisfactory due to estimation noise and interference
Solution Approach 1:
The patent implements a multi-functional beam selection system that considers multiple types of measurements and historical data simultaneously. The neural network processes Layer-1 measurements (RSRP, SINR), Layer-2 measurements (BLER), and historical beam failure information to comprehensively evaluate candidate beams. This universal approach allows the system to make informed decisions about beam selection by integrating diverse data sources, thereby improving the accuracy of new candidate beam RS determination while avoiding the limitations of single-measurement approaches.
Solution Approach 2:
The patent incorporates feedback mechanisms by using historical beam failure information as input to the neural network model. The system learns from previous beam failure experiences and adjusts its decisions accordingly. This feedback loop enables the system to avoid selecting beams that have previously failed, thereby improving the accuracy of candidate beam selection. The feedback mechanism processes historical data and integrates it with current measurements to make more informed beam selection decisions.
3Adaptability or versatility
If simple threshold-based beam failure detection is used, then the implementation is easy, but it fails to consider complicated cellular communication environment factors
Solution Approach 1:
The patent changes the detection parameters from simple threshold-based criteria to a multi-parameter neural network model. The system processes multiple parameters including Layer-1 measurements (RSRP, SINR), Layer-2 measurements (BLER), and historical beam failure information. By transforming the detection approach from single-threshold to multi-parameter analysis, the system becomes adaptable to complex communication environments while managing algorithm complexity through automated neural network processing.
Data Source
AI summary
Methods and systems for enabling a terminal device to perform a link recovery process are provided. In some embodiments, the method includes (1) receiving, by the terminal device, a set of Channel State Information Reference Signal (CSI-RS) resources for a beam failure detection; (2) receiving, by the terminal device, configuration information of a first neural network for the beam failure detection; (3) performing, by the terminal device, a measurement on the set of CSI-RS resources; and (4) generating, by the terminal device, a beam failure detection result by applying the first neural network on a result of the measurement on the set of CSI-RS resources.