Doppler Shift Pre-compensation via LSTM Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wireless communication networks face challenges in efficiently handling high Doppler shifts experienced by high-speed trains due to their mobility, leading to increased signaling overhead and reduced communication reliability.
Innovation Solution
A method using a long short-term memory neural network for predicting Doppler shift pre-compensation, where network nodes initiate a training phase to determine a predictive model based on Doppler shift data from wireless devices, allowing for reduced signaling and improved Doppler shift compensation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional Doppler shift compensation methods are used in high-speed train scenarios, then communication reliability is maintained, but signaling overhead increases significantly
Solution Approach 1:
The patent applies preliminary action by pre-compensating for Doppler shift effects before they occur in the downlink transmission. The network node calculates and applies frequency offset compensation based on predicted Doppler values obtained from uplink signals, thereby preparing the communication system in advance to handle high-speed train scenarios without requiring continuous heavy signaling during the actual data transmission.
Solution Approach 2:
The patent implements feedback by using uplink signals from the wireless device to estimate Doppler shift values, which are then fed back to the network node to adjust downlink transmission parameters. This feedback mechanism allows the system to adapt to the actual Doppler conditions experienced by the high-speed train while maintaining efficient signaling overhead through predictive modeling.
2Reliability
If frequent Doppler shift measurements are performed to maintain communication quality at high speeds, then communication reliability is improved, but signaling overhead increases
Solution Approach 1:
The system performs preliminary Doppler estimation using uplink signals before the actual downlink communication begins. By predicting future Doppler values based on current measurements and train velocity information, the system prepares compensation parameters in advance, eliminating the need for continuous frequent measurements and reducing signaling overhead while maintaining communication quality.
Solution Approach 2:
The patent applies dynamics by adapting the Doppler compensation strategy based on the train's velocity and position. The system dynamically adjusts the compensation parameters according to the actual motion state, allowing for optimal performance at high speeds without requiring continuous heavy signaling. The predictive model adapts to changing conditions smoothly, reducing the need for frequent re-measurements.
3Measurement precision
If machine learning-based predictive models are trained continuously, then Doppler shift prediction accuracy is improved, but processing time and computational resources increase
Solution Approach 1:
The patent applies preliminary action by performing machine learning model training in advance during periods when computational resources are available, rather than continuously during operation. The model is pre-trained on historical Doppler data and train velocity information, then deployed for real-time prediction. This allows high prediction accuracy to be achieved without requiring continuous training during active communication, thereby reducing processing time and computational resource consumption during critical operations.
Data Source
AI summary
A method, performed by a first network node. The method is for handling Doppler shift pre-compensation. The first network node sends a first indication towards a first wireless device. The first indication indicates a start of a training phase. The first network node obtains, based on the sent first indication, a set of information from the first wireless device. The set of information indicates: i) a Doppler shift experienced by the first wireless device while moving along a pre-defined trajectory to which a static set of radio network nodes provide coverage, and ii) a set of features characterizing how the first wireless device experienced the Doppler shift. The first network node also initiates determining, using machine-learning, and based on the received set of information, a predictive model of Doppler shift pre-compensation. The training phase is of the predictive model.