The annual conference, which this year took place in Orlando, Florida, April 17–20, focuses on IoT solutions to everyday problems. The other members of the Columbia team were Stephen Xia (EE PhD student), Wendy P. Fernandez (visiting undergraduate student from City College, NY), and Professors Xiaofan (Fred) Jiang and Peter R. Kinget.
The winning project digitally extracts arrival-time differences from the analog output signals of a microphone array, digitizing only the relevant features. Traditional systems digitize a sensor’s entire analog output and then use digital signal processing to extract the features. Direct analog-to-feature extraction is extremely power efficient, as it avoids the need to deal with power-hungry multi-bit analog-to-digital convertors, large memory blocks, or digital computational blocks. The immediate motivation for the project was to improve the power consumption of a pedestrian safety headset that Columbia graduate students are developing in collaboration with graduate students from the University of North Carolina in a project sponsored by the National Science Foundation.