| Title | : | Mapping Pervasive Environments using Radio Tomography and Neural Radiance Fields |
| Speaker | : | Amartya Basu (IITM) |
| Details | : | Mon, 25 Nov, 2024 11:00 AM @ SSB 334 Turing Hall |
| Abstract: | : | The proliferation of radio signals around us has transformed the way we perceive the physical world. Traditionally used for long-distance communication, Radio Frequency (RF) has recently emerged as a powerful sensing modality capable of addressing real-world challenges. Its unique ability to penetrate structures, interact with materials, and propagate through diverse environments enables it to serve applications beyond communication. In this thesis, we harness RF signals for two key purposes: sensing building structures within indoor environments and capturing the physical dynamics of outdoor terrains to facilitate the deployment of reliable communication infrastructure. We present UbiqMap for sensing and building indoor maps, and SpecNeRF for Radio Environment Map (REM) estimation. The demand for real-time, accurate mapping is ubiquitous, particularly in complex indoor settings. While Simultaneous Localization and Mapping (SLAM)- based methods are widely used, Radio Tomographic Imaging (RTI) provides distinct advantages, such as mapping inaccessible or enclosed spaces, shorter scanning trajectories, and identifying material properties of structures. However, existing RTI systems often rely on pre-deployed, precisely calibrated infrastructure and significant computational resources, limiting their applicability in ubiquitous settings. To address these challenges, UbiqMap is designed as a lightweight, end-to-end RTI-based system for real-time indoor mapping that minimizes reliance on pre-deployed infrastructure. In addition, Neural Radiance Fields (NeRF) have emerged as a powerful technique for synthesizing novel views of complex 3D scenes from sparse image sets. Recent advances highlight their potential in wireless networks. This talk explores the application of NeRF to spectrum sensing, proposing SpecNeRF, which leverages RF-based NeRF to improve the accuracy and efficiency of spectrum sensing in wireless communication networks. Extensive experiments demonstrate that SpecNeRF offers significant improvements in scalability, robustness to environmental changes, and adaptability to varying signal conditions. SpecNeRF not only addresses current spectrum sensing challenges but also lays the foundation for innovative applications in future wireless networks, including cognitive radio and 6G technologies |
