Project

Can Next-Generation WiFi Routers Reconstruct Indoor Layouts?

Signal Kinetics

Groups

Media Lab Research Theme: Future Worlds

Indoor layout reconstruction has broad applications in various scenarios, such as smart homes and virtual reality. However, optical sensors like cameras and LiDAR can raise privacy concerns and often struggle in challenging conditions, such as darkness or smoke. This raises the question: can we achieve accurate indoor layout reconstruction using a single, static millimeter-Wave (mmWave) wireless device? mmDiff addresses this challenge by proposing an efficient method that leverages a single static mmWave sensor—the same type of signal used in 5G and next-generation WiFi—for accurate and privacy-preserving indoor layout reconstruction.

How does it work?

mmDiff leverages human mobility to reconstruct indoor layouts using a single static mmWave sensor. As a person moves through the environment, their body causes mmWave signals to reflect off surrounding surfaces and objects before reaching the receiver—an effect known as multi-path propagation. These reflections provide rich spatial information that mmDiff uses to infer the layout of the environment.

This design introduces three main innovations:

1) mmDiff enhances the visibility of signals caused by multi-path (ghost signals). Because ghost signals result from indirect reflections, their transmitted and received angles differ, making them distinguishable through angular analysis.

2) mmDiff uses the geometric relationships between signal reflections (also called multipath ghosts), real objects, and surfaces to estimate the layout of a room. By analyzing how these reflections behave, mmDiff can piece together the structure of the environment, using just the data from a single mmWave sensor.

3) mmDiff proposes a generative model that can fill in missing details in ghost-based layout reconstruction. Additionally, the model is capable of distinguishing furniture from the surrounding environment.

How well does mmDiff work?

mmDiffis is very accurate:

· mmDiff achieves a Chamfer distance of approximately 16 cm in indoor layout reconstruction.

· mmDiff achieves an IoU-based accuracy of over 58% in furniture detection.

This paper introduces a novel method for accurate indoor layout reconstruction using a single static mmWave sensor. It highlights the potential of privacy-preserving mmWave sensing for applications in smart homes, security, and virtual/augmented reality.