Title:  THz Wireless Channel Characterization and Modeling for Chip-to-Chip Communication in Computing Systems

Committee:

Dr. Alenka Zajic, ECE, Chair, Advisor

Dr. Andrew Peterson, ECE

Dr. Gregory Durgin, ECE

Dr. Gordon Stuber, ECE

Dr. Milos Prvulovic, CoC

Abstract: This research focuses on the characterization and modeling of the THz wireless channel for chip-to-chip communication in computing systems. To understand the signal propagation mechanisms in a metal enclosure, this theis presents the channel characterizations inside a desktop sized metal cavity with the consideration of several potential scenarios. Based on the measurement findings, a path loss model for THz chip-to-chip communication is proposed. According to the cavity environment and the statistical properties of the channel inside the metal cavity, a geometry based statistical channel model is constructed. Afterwards, a more practical motherboard desktop environment is investigated by putting the densely populated motherboard in the metal cavity. Both channel characterization and modeling are presented in the thesis for this practical environment. Besides that, deep learning methods are applied on the property prediction of THz wireless channel in the motherboard desktop environment. A ResNet based model is propsed and analyzed for the prediction of the scenario the channel is under and the attribute of the predicted scenario. The objective of this research is to provide other researchers with guidelines on how to characterize and model the wireless channel in computing systems, and provide the channel information for future THz chip-to-chip wireless system design.