Digital Twin-Enabled Waveform Optimization for VHF Radio Communication Systems.

Area of Science:

• Electrical Engineering
• Computer Science
• Communications Engineering

Background:

• Modern Very High Frequency (VHF) radio systems struggle with spectrum congestion and dynamic interference.
• Current adaptive methods lack experience accumulation across decision cycles, limiting effectiveness.

Purpose of the Study:

• To introduce a digital twin-enabled online learning framework (DT-MAB) for adaptive waveform selection in tactical VHF communication.
• To improve system resilience and performance in complex electromagnetic environments.

Main Methods:

• Utilized a contextual multi-armed bandit algorithm (Lin-UCB) for continuous learning.
• Integrated a digital twin as a virtual sandbox for safe configuration screening.
• Developed a hierarchical performance model using Analytic Hierarchy Process (AHP) for reward calculation.
• Expanded the configuration space to 63 candidates.

Main Results:

• DT-MAB significantly reduced mean cumulative regret by 29% compared to a digital twin-less approach.
• DT-MAB outperformed Particle Swarm Optimization (PSO)-based methods by 16.5% on average.
• Ablation studies showed a 49% increase in performance drop events without the digital twin's virtual exploration.

Conclusions:

• The digital twin is crucial for the online learning architecture, preventing link disruptions during exploration.
• DT-MAB offers a superior adaptive waveform selection strategy for tactical VHF communications.
• The framework demonstrates enhanced learning efficiency and system stability.