Recent Research Themes

Understanding how radio waves travel through real environments is the foundation of every wireless system — and it has been the core expertise of our group for over two decades. We measure, model, and emulate the radio propagation channel: from double-directional channel sounding in the field, to physics-based models that link the geometry of an environment to the behavior of the channel, to real-time emulation platforms that let new wireless systems be tested without ever going on air. While channel models are broadly classified as deterministic or stochastic, our interest lies chiefly in site-specific modeling, where each interaction between a radio wave and the environment is described deterministically by physics or its well-founded approximations — so that the model explains not only how the channel behaves, but why.

Every radio signal that carries data also carries an imprint of the world it traveled through. We investigate integrated sensing and communication (ISAC) at the physical layer — turning the wireless channel itself into a sensor. Using signals that already surround us, from Wi-Fi to cellular waveforms across frequency bands, we characterize how multipath propagation responds to sensing targets, most notably the movement of people and vehicles. Because wireless coverage is everywhere, this approach brings passive sensing everywhere too — without cameras, and without requiring people to carry any device.

Between the antenna and the information lies signal processing. We develop the estimation and detection algorithms that make wireless links work in difficult conditions — channels that change rapidly with motion, signals buried in interference, and waveforms designed to serve communication and sensing at once. Our work spans both simulation and reality: we apply these techniques to site-specific channels reproduced by deterministic modeling such as ray tracing, and to signals and channels measured in the real world. This deterministic, physics-grounded view of the channel is a deliberate choice — when the interaction between waves, environment, and targets is explicitly modeled rather than treated statistically, signal processing can exploit that structure, and its performance can be traced back to physical causes. Stochastic, site-general channel models still have their place in our work: we use them to evaluate algorithms under standardized conditions and to verify that what works in a specific site generalizes beyond it.