Enhancing wireless signal perception through combined processing and learning methods

In recent years, wireless signals have transcended their traditional role in communication systems, emerging as a powerful medium for sensing and perception. These electromagnetic waves have demonstrated remarkable versatility, enabling applications ranging from coarse-grained autonomous vehicle perception to finegrained physiological monitoring of human vital signs. While the underlying sensing mechanisms vary across applications, they fundamentally leverage the same physical properties of radio frequency propagation, creating opportunities for unified theoretical frameworks and methodological approaches. This dissertation introduces novel hybrid methodologies that bridge classical signal processing techniques with modern deep learning approaches. We build upon established wireless sensing foundations while introducing innovative processing pipelines that combine domain-specific signal transformations with adaptive neural architectures. Our framework extends beyond application-specific solutions, presenting a generalizable approach to wireless signal processing that maintains theoretical rigor while embracing the flexibility of data-driven methods. Through extensive experimental validation, we demonstrate the efficacy of our proposed methods across multiple perception tasks. Specifically, we address critical challenges in autonomous vehicle sensing and indoor object localization and tracking. Our approach leverages domain knowledge of wireless propagation characteristics to inform the design of specialized learning architectures, resulting in significant performance improvements over traditional methods. The methodologies developed in this work not only advance the state-of-the-art in wireless sensing but also establish new paradigms for integrating physical understanding with learning-based approaches in signal processing systems.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Electrical & Computer Engr

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2024

Author dc:creator

• Madani, Sohrab

Contributors dc:contributor

• Al-Hassanieh, Haitham
• Patel, Sanjay J
• Gupta, Saurabh
• Mitra, Sayan

• Wireless Sensing
• Deep Learning
• Signal Processing

Statement dc:rights

• Copyright 2024 Sohrab Madani

Language dc:language

eng, en

Handle dc:identifier

https://hdl.handle.net/2142/127337

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F127337