Universal Wi-Fi signal encoder improves human sensing across devices
UniCSI Towards a Universal Wi-Fi CSI Encoder for Ubiquitous Human Sensing
Machine LearningNetworking and Internet Architecture
Summary
Wi-Fi signals can be used to sense people's movements and activities, but different devices capture the Wi-Fi data in many incompatible ways. The authors created UniCSI, a system that understands Wi-Fi data from many different devices without losing important details. It uses physics-based encoding to handle the varied Wi-Fi frequencies and organizes the information into a fixed format that works for lots of sensors. This approach helps computer models learn better and perform more accurately when detecting humans from Wi-Fi signals across many setups.
What this means in practice
- •For smart home developers: Build Wi-Fi-based systems that detect human presence or activity reliably across different devices and Wi-Fi configurations.$Commercial implications: Enables consumer products that use Wi-Fi signals for human sensing without device-specific retraining, allowing broad market deployment.
- •For security system integrators: Implement consistent human detection using Wi-Fi signals from heterogeneous hardware to monitor secured areas without dedicated sensors.
Authors
Daniel Eckhoff, Hua Kang, Zhitang Chen, Jie Chuai
Abstract
Wi-Fi sensing promises to turn the everyday wireless signals that already surround us into ubiquitous sensors for human sensing. However, a fundamental obstacle is that CSI is acquired under diverse device-specific configurations, including different subcarrier counts, bandwidths, and carrier bands. Consequently, the resulting CSI tensors vary in both spectral resolution and tensor shape, making heterogeneous modeling challenging. Standard architectures struggle with such heterogeneity, forcing lossy pre-processing which compromises the underlying signal. To bridge this gap, we present UniCSI, a unified foundation architecture that directly operates on heterogeneous CSI while preserving the integrity of the native waveform. UniCSI hinges on two core innovations: (1) a physics-informed RF tokenizer that encodes each frequency channel based on its fractional position within the physical spectrum rather than rigid array indices. It preserves intrinsic spectral coherence and enables seamless, frequency resolution-agnostic processing across arbitrary sensing configurations. (2) a spectral aggregator that distills variable-length channel sequences into a fixed-size spectral signature, effectively decoupling the feature dimensionality from the physical subcarrier spacing. Extensive evaluations on a large-scale corpus of 25 heterogeneous public datasets, spanning 14 to 2048 subcarriers, 20 to 160 MHz bandwidth, and the 2.4 and 5 GHz bands, demonstrate that native heterogeneous ingestion substantially improves cross-domain transfer under both supervised and self-supervised training schemes, particularly in regimes where fixed-grid architectures fail to generalize.