Integration of Subspace Machine Learning System and Method for Enhanced Terrestrial Navigation

Abstract

This invention introduces an innovative signal processing and forecasting approach using subspace-based methods, particularly tailored for ionospheric Total Electron Content (TEC) predictions. At the core of this methodology is the integration of Singular Spectrum Analysis (SSA), Linear Recurrent Formula (LRF), and Artificial Neural Networks (ANN), forming a robust SSA-LRF-ANN framework. This hybrid model leverages the strengths of each component: SSA for decomposing and reconstructing the TEC time series data by distinguishing between signal and noise subspaces, LRF for parameter estimation and forecasting, and ANN for refining the residuals and capturing complex nonlinear patterns. The application of this framework is demonstrated using GPS-derived TEC data from the Bangalore grid (13.02° North and 77.57° East) during sunspot cycle 25 (2020). This novel SSA-LRF-ANN approach substantially improves TEC forecasting, providing more accurate and reliable predictions critical for enhancing GPS accuracy and other Earth Observation applications. The presented numerical results validate the practical implementation and effectiveness of the proposed methodology, making it a promising tool for advanced time series forecasting in various domains.

Specification

Description:The present disclosure proposes an integrated cloud-based terrestrial navigation system and method thereof. The following presents a simplified summary in order to provide a basic understanding of some aspects of the claimed subject matter. This summary is not an extensive overview. It is not intended to identify key/critical elements or to delineate the scope of the claimed subject matter. Its sole purpose is to present some concepts in a simplified form as a prelude to the more detailed description that is presented later.
In order to overcome the above deficiencies of the prior art, the present disclosure is to solve the technical problem of providing an integrated cloud-based terrestrial navigation system that aids in enhancing the reliability of navigation services where the receiver line of sight is interrupted.

SSA decomposes complex TEC time series data into its fundamental components. This technique captures underlying patterns and variability within the data, outperforming decomposition methods like Principal Component Analysis (PCA) and Singular Value Decomposition (SVD).
Combining SSA with LRF and ANN techniques enhances the model's ability to reconstruct and forecast TEC values accurately. This integrated approach shows superior performance, particularly effective in capturing the intricate dynamics of TEC influenced by seasonal variations and solar activity.
The computational efficiency of the SSA-based models, including SSA-LRF-ANN, is a significant advantage and makes these models suitable for real-time forecasting applications where timely data processing is critical. The low computational burden and high forecasting accuracy position these models as viable solutions for operational environments requiring rapid and reliable TEC predictions.
Furthermore, these methods can be adapted to various domains, including Earth observation, satellite communication, and other fields that require precise and robust forecasting techniques.

Further, objects and advantages of the present invention will be apparent from a study of the following portion of the specification, the claims, and the attached drawings.


, Claims:1. The method of claim 1, wherein the reconstructed components singular spectrum analysis (SSA) classifies the original time series using the reconstructed components (RCs) time series.
2. The method of claim 1, Integration of SSA, LRF, and ANN: The combination of SSA for decomposition and reconstruction, LRF for parameter estimation, and ANN for residual modeling provides a novel and efficient framework for forecasting ionospheric TEC.
3. The method of claim 1, superior Forecasting Accuracy: The proposed method yields more accurate TEC forecasts than traditional models and other SSA-based approaches, as demonstrated by lower MAE and RMSE.
4. The method of claim 1, Real-Time Forecasting Capability: The computational efficiency of the SSA-LRF-ANN framework enables its application in real-time forecasting scenarios, making it suitable for dynamic and time-sensitive environments.
5. The method of claim 1, Versatility and Scalability: The methodology's design allows its application to a wide range of time series forecasting problems, from Earth Observation to financial market analysis, without significant computational constraints.

Documents