Signal Harmonization via Adaptive Predistortion and Estimation (SHAPE)

This study addresses the need for accurately replicating threat-representative radio-frequency (RF) waveforms emitted from low-cost, open-air RF emitter systems. Commercially available power amplifiers (PAs) and other RF components introduce distortions, creating a discrepancy between theoretical waveforms and their over-the-air emissions. To overcome this challenge, an adaptive filtering approach that leverages machine learning (ML) and an integrated feedback loop is proposed. This iterative technique dynamically modifies a software-based pre-distortion filter, aligning transmitted RF outputs precisely with the targeted digital twin waveform model, which results in a predictable response for the entire system. Toyon will collaborate with the Government to define system requirements, ensuring alignment with operational objectives and constraints. Following requirements definition, we will evaluate multiple candidate architectures including ML-based and Support Vector Regression (SVR)-based systems. We will then identify the optimal solution capable of real-time waveform correction and validation. The most promising architecture and adaptive algorithms will be recommended for detailed implementation in the subsequent development phase. The final report will describe the selected system architecture, specifying all critical hardware and software components. Additionally, a preliminary hardware and software design roadmap will be provided, setting the stage for a smooth transition into prototype development and demonstration in the follow-on effort.