Talk: Enabling High-Fidelity Acoustic Sensing on Smartphones

Prof. Xingyu Zhang (
Department of Electrical and Computer Engineering, University of Wisconsin-Madison

2pm, August 21, 2014.

Xinyu Zhang is an Assistant Professor in the Department of Electrical and Computer Engineering at the University of Wisconsin-Madison. He received the B.E. degree in 2005 from Harbin Institute of Technology, the M.S. degree in 2007 from the University of Toronto, and the Ph.D. degree in 2012 from the University of Michigan. He worked as a research intern at Microsoft Research Asia from May to Aug. 2010, and at NEC Labs American from May to Dec. 2011. His research interest lies in designing cross-layer protocols that improve wireless network performance, as well as mobile applications that enable fine-grained context sensing. His work spans the areas of wireless networking, communications engineering and mobile computing, involving both mathematical analysis and systems implementation. He received ACM MobiCom Best Paper Award in 2011, and NSF CAREER award in 2014.

High-end acoustic sensing applications, such as directive audio capturing, speaker tracking, and localization, are typically realized using dedicated microphone arrays. Modern smartphones fail to support such demanding applications because they use dumb microphones as basic recording apparatus. In this talk, I will introduce our recent work that aims to overcome this limitation by augmenting novel acoustic signal processing frameworks on smartphones. First, I will present a system called UbiK that achieves centimeter-scale acoustic localization, and can accurately identify keystrokes on solid surfaces. UbiK extracts and optimizes the location-dependent multipath fading features from audio signals, and leverages the dual-microphone interface on smartphones to improve signal granularity. Second, I will introduce Dia, a collaborative acoustic sensing application that enables precise I/O synchronization of multiple smartphones to emulate a microphone array, thereby realizing distributed audio beamforming and speaker localization. Both UbiK and Dia point to the potential of enhancing mobile sensing with application-level signal processing. Besides, I will briefly introduce other ongoing projects in my research group, including network MIMO, visible light networking and millimeter-wave networks.