Deep Learning S-Curve Delay Regression for Multicarrier Signal Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current positioning methods, especially GNSS and multi-source sensor positioning, face challenges in urban and indoor areas due to signal attenuation, multipath phenomena, and high environmental noise, leading to inaccurate navigation and positioning.
Innovation Solution
A multicarrier signal tracking method based on deep learning that utilizes S-curve features to regress delay estimation, leveraging pre-trained networks to handle multipath signals and varying signal-to-noise ratios, and integrates with existing large-scale networks like LTE and 5G for cost-effective and precise indoor positioning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional GNSS positioning is used, then positioning can be achieved in open areas, but positioning accuracy deteriorates in urban and indoor areas due to signal attenuation and multipath phenomena
Solution Approach 1:
The patent introduces an intermediary mechanism (deep learning-based delay estimation model) between the received signal and the positioning calculation. This intermediary processes the corrupted signals by learning the mapping between S-curve features and true delays, effectively mediating the transition from noisy multipath-contaminated signals to accurate delay estimates for positioning
Solution Approach 2:
The patent transforms the positioning problem by changing the parameters used for estimation. Instead of directly estimating delay from raw signals, it extracts S-curve features (amplitude, width, shape parameters) and uses these transformed parameters as inputs to the deep learning model, which then outputs corrected delay estimates that account for multipath effects
2Device complexity
If conventional delay estimation methods are used, then the system structure remains simple, but positioning accuracy deteriorates in multipath environments
Solution Approach 1:
The patent replaces traditional mechanical/signal-processing-based delay estimation methods (such as peak detection, correlation methods) with a data-driven deep learning approach. The neural network learns complex nonlinear mappings from S-curve features to delay values, substituting conventional signal processing algorithms with an intelligent model that adapts to multipath conditions
3Measurement precision
If fingerprint positioning is used, then indoor positioning can be achieved, but significant manpower, material resources, and cost are required for data collection and maintenance
Solution Approach 1:
The patent enables the positioning system to serve itself by using existing broadcast signals (navigation signals already transmitted by satellites) as the positioning source. The deep learning model processes these freely available signals directly, eliminating the need for external infrastructure like fingerprint databases or additional indoor transmitters, making the system self-sufficient
4Ease of manufacture
If multi-source sensor positioning is used, then positioning can be achieved without additional infrastructure, but positioning errors accumulate continuously and costs increase
Solution Approach 1:
The patent introduces periodic correction by using satellite-based signals that provide absolute position references at regular intervals. The deep learning delay estimation periodically corrects the positioning solution using these external references, preventing the continuous error accumulation that occurs in inertial or sensor-only systems
Data Source
AI summary
The present invention discloses a method and system for multicarrier signal tracking based on deep learning and high precision positioning. Using the data characteristics of S-curve, and using S-curve which contains multipath signals as feature data for training deep learning networks under different signal-to-noise ratios. The delay regression results of receiving signal can be directly obtained by the S-curve of real-time receiving signal and the pre-trained network. The motivation of this method is to fully utilize the advantages of deep learning networks in accurately regressing complex problems with a large amount of data, fundamentally solving the impact of multipath signals on the delay estimation of the main path signal in traditional software defined receivers, extracting the corresponding relationship between the delay of main path and S-curve under the influence of different signal-to-noise ratios and different multipath signals.