We build radio frequency (RF) digital twins by reconstructing urban 3D geometry and RF signal propagation environments from mobile sensing data, leveraging neural networks as a powerful tool for accurate and efficient reconstruction.

We study low-power wireless sensor networks by estimating interference graphs and link connectivity in multi-hop BLE systems. Enabled by our hardware design, nodes can probe the channel during ongoing communication without reserving additional dedicated time-frequency resources, thereby supporting efficient and reliable network operations.