Interwoven Waves: Enhancing the Scalability and Robustness of Wi-Fi Channel State Information for Human Activity Recognition

This PhD dissertation investigates the future of unobtrusive radio wave-based sensing, specifically focusing on Wi-Fi sensing in realistic healthcare scenarios. Wi-Fi sensing leverages the analysis of multi-path reflections of radio waves to monitor human activities and physiological states, providing a scalable solution without intruding on daily life.

Wi-Fi-based sensing, particularly through channel state information, fits well in healthcare due to its ubiquitous presence and unobtrusiveness. As our society ages and populations grow, continuous health monitoring becomes increasingly critical. Existing solutions like wearable devices, audiovisual technologies, and expensive infrastructure modifications each have limitations, such as forgetting to wear devices, privacy invasions, and high costs. Channel state information-based sensing offers a promising alternative, enabling remote monitoring without the need for additional infrastructure changes.

Nevertheless, implementing channel state information-based sensing in already congested Wi-Fi bands could present challenges in the future. Current solutions often exacerbate congestion by adding random noise, which can degrade network performance. These solutions also tend to address niche problems in idealistic settings, making it difficult to justify their use in everyday environments due to potential impacts on network latency and overall user experience.

To realise the potential of Wi-Fi sensing, future solutions must integrate seamlessly with wireless communication networks, ensuring that sensing and communication processes coexist and collaborate effectively. This dissertation categorises the relationship between sensing and communication into three models: parasitic, opportunistic, and mutualistic. In the parasitic model, sensing operates independently of the wireless infrastructure, potentially adding noise and congestion. The opportunistic model leverages existing traffic flows, avoiding adverse effects on communication. The mutualistic model aims for a balance, enhancing both sensing and communication without compromising either function.

The primary research objective is to enhance the robustness and scalability of channel state information-based sensing for human activity recognition, facilitating seamless integration into home environments with minimal impact on existing infrastructure. Overall, this dissertation provides an exploration of the challenges and solutions for unobtrusive Wi-Fi sensing in healthcare, paving the way for future advancements in the field.