Background

Thanks to the ubiquitous nature of Wi-Fi access points and devices, Wi-Fi sensing enables transformative applications in remote health care, home/office security, and surveillance. Existing work has explored the usage of machine learning on channel state information (CSI) computed from Wi-Fi packets to classify events of interest. However, most of these algorithms require a significant amount of data collection, as well as extensive computational power for additional CSI feature extraction. Moreover, the majority of these models suffer from poor accuracy when tested in a new or untrained environment.

Technology Overview

To address these issues, Northeastern University researchers developed Reliable Wi-Fi Sensing (ReWiS) which is a novel framework for robust and environment-independent CSI-based Wi-Fi sensing. The core design principles behind the ReWiS are:

(i) leveraging multi-antenna, multi-receiver diversity, as well as fine-grained frequency resolution to improve the overall robustness of the algorithms

(ii) utilizing a customized version of Few Shot Learning (FSL) to remove the need for application-specific feature extraction and help generalize to new environments by leveraging a limited number of new samples

The researchers also proposed a technique based on SVD (singular value decomposition) to make the FSL input constant irrespective of the number of receiver antennas and window size. To this end, the researchers were able to reduce the input size by about 80% of the original size in their dataset. 

Benefits

• A customized few-shot learning model to learn from the macro- and micro- diversity provided through using multiple antennas and receivers 
• High-frequency resolution (data collection over 242 subcarriers) data compression using SVD
• ReWiS improves the performance by about 40% with respect to existing single-antenna low-resolution approaches 
• Compared to a CNN-based approach, ReWiS shows a 35% more accuracy and less than 10% drop in accuracy when tested in unseen environments (the CNN performance drops by more than 45%)

Applications

 Can be used in all CSI sensing applications, such as: 

• Human activity recognition
• Wi‑fi surveillance
• Remote healthcare
• Safety applications 

Opportunity

• License
• Research collaboration
• Partnering