Hamid R. Sharif
Date of this Version
Taylor, J. M. (2020). A Novel Path Loss Forecast Model to Support Digital Twins for High Frequency Communications Networks (Doctoral dissertation). University of Nebraska - Lincoln.
The need for long-distance High Frequency (HF) communications in the 3-30 MHz frequency range seemed to diminish at the end of the 20th century with the advent of space-based communications and the spread of fiber optic-connected digital networks. Renewed interest in HF has emerged as an enabler for operations in austere locations and for its ability to serve as a redundant link when space-based and terrestrial communication channels fail. Communications system designers can create a “digital twin” system to explore the operational advantages and constraints of the new capability. Existing wireless channel models can adequately simulate communication channel conditions with enough fidelity to support digital twin simulations, but only when the transmitter and receiver have clear line of sight or a relatively simple multi-path reflection between them. With over-the-horizon communications, the received signal depends on refractions of the transmitted signal through ionospheric layers. The time-varying nature of the free electron density of the ionosphere affects the resulting path loss between the transmitter and receiver and is difficult to model over several days. This dissertation examined previous efforts to characterize the ionosphere and to develop HF propagation models, including the Voice of America Coverage Analysis Prediction (VOACAP) tool, to support path loss forecasts. Analysis of data from the Weak Signal Propagation Reporter Network (WSPRnet), showed an average Root Mean Squared Error (RMSE) of 12.9 dB between VOACAP predictions and actual propagation reports on the WSPRnet system. To address the significant error in VOACAP forecasts, alternative predictive models were developed, including the Forecasting Ionosphere-Induced Path Loss (FIIPL) model and evaluated against one month of WSPRnet data collected at eight geographically distributed sites. The FIIPL model leveraged a machine learning algorithm, Long Short Term Memory, to generate predictions that reduced the SNR errors to an average of 4.0 dB RMSE. These results could support more accurate 24-hour predictions and provides an accurate model of the channel conditions for digital twin simulations.
Advisor: Hamid R. Sharif-Kashani