Part 7: Neural Scene Representations
Chapter 24: NeRF for RF
Research~160 min
Learning Objectives
- Derive the optical volume rendering equation and explain positional encoding, hierarchical sampling, and Instant-NGP acceleration
- Identify the five key differences between optical and RF rendering and formulate the complex-valued RF rendering integral
- Analyze NeRF-Squared architecture, training objective, and limitations for wireless propagation modelling
- Survey RF-NeRF variants (WiNeRT, R-NeRF, VoxelRF, NeRF-APT, DART, ISAR-NeRF) and their design trade-offs
- Connect neural volumetric rendering to the Born-approximation forward model from Chapters 5-6
Sections
💬 Discussion
Loading discussions...