Bringing Motion-Awareness To WLANs Using PHY Layer Information
In recent years, with the proliferation of smartphones, tablets, and laptops, mobile devices are becoming the preferred medium of Internet access in Wireless LANs (WLANs). Such a trend has two major impacts: on one hand, it calls for new, mobility-aware WLAN protocols that are optimized for client mobility in order to sustain high performance; on the other hand, the small form factor of mobile devices calls for better ways of interacting with device interfaces beyond the traditional touchscreen, keyboard, and mouse. Central to dealing with both challenges is the ability to track human motion. Theoretically, human motion can be detected by studying its impact on wireless signal propagation. A major obstacle towards this direction until recently has been the limited amount of PHY layer information exposed by WiFi chipsets. In this thesis, we demonstrate that it is possible to leverage WiFi signals from commodity mobile devices for fine-grained human motion detection by exploiting new types of PHY layer information exposed by today's WiFi chipsets. We propose two motion detection solutions for two different application scenarios and explore their benefits on protocol performance and human-device interaction, respectively. Our solutions require no external hardware and can be enabled on today's mobile devices only via software modifications. In the first part of the thesis, we present a solution for detecting client mobility on the access point (AP) side. We leverage channel state information (CSI) and time-of-flight (ToF) to classify WiFi clients into four categories according to their mobility types. In addition, we demonstrate how fine-grained mobility determination can be exploited to improve performance of a number of WLAN protocols. Our testbed experiments show that our mobility classification algorithm achieves more than 92% accuracy in a variety of scenarios, and the combined throughput gain of all four mobility-aware protocols over their mobility-oblivious counterparts can be up to 98%. In the second part of the thesis, we are developing a system that allows a mobile device to perform fine-grained tracking of the user's hand motion. Our idea is to leverage signal angle-of-arrivals (AoA) from neighboring WiFi devices in the 3-D space. In contrast to prior solutions that can only recognize a small predefined set of gestures, we are targeting a solution that requires no a priori training and can track an arbitrary set of gestures enabling a general-purpose hands-free interface for interacting with the mobile device. Our solution does not require the use of any wearable or any dedicated hardware setup. We demonstrate through our prototype that commodity wireless cards can track the user's hand with less than 5 cm error on average. We further implement an in-air handwriting application that achieves an average word recognition accuracy of 91%. We believe our solution can open up a whole new class of applications in human-computer interaction.