Design and development of a decompression-based receiver for ice sounding radar and investigative signal recovery

Ice-penetrating radar systems are critical tools in glaciology and climate research, supporting scientific missions such as that of the Center for Oldest Ice Exploration (COLDEX). A primary challenge for these radars is achieving sufficient dynamic range to capture both strong, shallow reflections from the ice surface without saturating the radar's analog to digital converter (ADC), and extremely weak signals from the deep bedrock. This thesis presents a non-conventional analog receiver architecture and signal processing methodology designed to enhance the dynamic range of a radar system by utilizing characterized signal compression. The core of this approach relies on the non-linear properties of a set of RF power limiters to compress high-power received signals.

A complete receiver module was designed, simulated, implemented on a 4-layer printed circuit board for operation in the 600-900 MHz band, with the design being adaptable to other frequency ranges (e.g. 140-215 MHz). Multiple modules based on this design were manufactured for three different multichannel radar systems. Characterization of the manufactured receiver blocks demonstrates reproducible performance, confirming the well-defined non-linear input and output power relationship, which is essential for this technique.

To recover the original signal from the compressed data, this work approaches the inversion problem using a machine learning technique. A 3-layer neural network was trained on a test data set generated from an exponentially-varying, single-tone waveform, mapping the compressed receiver output back to the original input envelope. The trained model was then validated using a distinct, triangular-amplitude-modulated test signal. The results show that the neural network can accurately predict and reconstruct the original, uncompressed waveform envelope from the compressed receiver output for discrete frequencies within the band of operation. This work serves as a successful proof-of-concept for a decompression-based analog receiver, offering an alternate and effective pathway to enhancing the dynamic range of ice-sounding radar systems.