Skip to main content
  • Yongsen Ma, William & Mary
In recent years, WiFi has a very rapid growth due to its high throughput, high efficiency, and low costs. Multiple-Input Multiple-Output (MIMO) and Orthogonal Frequency-Division Multiplexing (OFDM) are two key technologies for providing high throughput and efficiency for WiFi systems. MIMO-OFDM provides Channel State Information (CSI) which represents the amplitude attenuation and phase shift of each transmit-receiver antenna pair of each carrier frequency. CSI helps WiFi achieve high throughput to meet the growing demands of wireless data traffic. CSI captures how wireless signals travel through the surrounding environment, so it can also be used for wireless sensing purposes. This dissertation presents how to improveĀ  WiFi sensing and networking with CSI. More specifically, this dissertation proposes deep learning models to improve the performance and capability of WiFi sensing and presents network protocols to reduce CSI feedback overhead for high efficiency WiFi networking.

For WiFi sensing, there are many wireless sensing applications using CSI as the input in recent years. To get a better understanding of existing WiFi sensing technologies and future WiFi sensing trends, this dissertation presents a survey of signal processing techniques, algorithms, applications, performance results, challenges, and future trends of CSI-based WiFi sensing. CSI is widely used for gesture recognition and sign language recognition. Existing methods for WiFi-based sign language recognition have low accuracy and high costs when there are more than 200 sign gestures. The dissertation presents SignFi for sign language recognition using CSI and Convolutional Neural Networks (CNNs). SignFi provides high accuracy and low costs for run-time testing for 276 sign gestures in the lab and home environments.

For WiFi networking, although CSI provides high throughput for WiFi networks, it also introduces high overhead. WiFi transmitters need CSI feedback for transmit beamforming and rate adaptation. The size of CSI packets is very large and it grows very fast with respect to the number of antennas and channel width. CSI feedback introduces high overhead which reduces the performance and efficiency of WiFi systems, especially mobile and hand-held WiFi devices. This dissertation presents RoFi to reduce CSI feedback overhead based on the mobility status of WiFi receivers. CSI feedback compression reduces overhead, but WiFi receivers still need to send CSI feedback to the WiFi transmitter. The dissertation presents EliMO for eliminating CSI feedback without sacrificing beamforming gains.
  • Channel State Information,
  • Networking,
  • Sensing,
  • WiFi
Publication Date
Spring January 15, 2020
Citation Information
Yongsen Ma. "dissertation_Oct20_2019.pdf" (2020)
Available at:
Creative Commons license
Creative Commons License
This work is licensed under a Creative Commons CC_BY-NC-ND International License.