(Downtown San Francisco, taken from the top of the Twin Peaks in the summer of 2011.)

Hi, I am Zheng, a 5th-year PhD candidate in CyLab Mobility Research Center, Department of Electrical and Computer Engineering, Carnegie Mellon University. My research lies in the broad areas of ubiquitous computing, mobile sensing, and networked embedded systems. Most of my research projects focus on using machine learning and signal processing techniques to design and prototype new technologies for mobile systems. I am advised by Prof. Pei Zhang and Prof. Dan Siewiorek.

Prior to joining CMU, I received my B.E. (Top 1 GPA in the program)  and M.S. (with Honors) in Signal and Information Processing from Beijing University of Posts and Telecommunications, Beijing, China.

My resume.

My LinkedIn page.

I will be joining Google as a full-time software engineer starting from the summer of 2014.

Research Projects

Headio: Zero-Configured Heading Acquisition for Indoor Mobile Devices Through Multimodal Context Sensing
Headio is a novel approach to providing reliable device headings in indoor environments. It achieves this by aggregating ceiling images of an indoor environment, and using computer vision-based pattern detection techniques to provide directional references. Headio constantly provides accurate heading detection performance in diverse situations, achieving better than 1 degree average heading accuracy, up to 33X improvement over existing techniques that use a digital magnetometer. [Ubicomp 2013 paper] [Java Source Code (Open-sourced under the MIT License)]

Spartacus: Spatially-Aware Interaction for Mobile Devices Through Energy-Efficient Audio Sensing
Spartacus is a mobile system that enables spatially-aware device interactions with zero prior configuration. Using built-in microphones and speakers on commodity mobile devices, Spartacus uses a novel acoustic Doppler-effect based technique to enable users to accurately initiate an interaction with a neighboring device through a pointing gesture. Experimental results show that Spartacus achieves an average 90% device selection accuracy within 3m for most interaction scenarios. [MobiSys 2013 paper] [Introduction Video (Youtube Link)] [User Gesture Videos & Analysis Code (Matlab)]

SugarMap: Location-less Coverage for Micro-Aerial Sensing Swarms
SugarMap is a location-less coverage system that enables resource-constrained micro-aerial sensing nodes to achieve efficient sensing coverage. The self-establishing system uses approximate motion models of mobile nodes in conjunction with radio signatures from self-deployed stationary anchor nodes to create a common coverage map. Consequently, the system coordinates node movements to reduce sensing overlap and increase the speed and efficiency of coverage. The system uses particle filters to account for uncertainty in sensors and actuation of sensing nodes, and incorporates redundancy to guarantee coverage. Through large-scale simulations and a real implementation on the SensorFly MAV sensing platform, we show that SugarMap provides better coverage than the existing coverage approaches for MAV swarms. [IPSN 2013 paper]

Polaris: Accurate Indoor Orientations for Mobile Devices Using Ubiquitous Visual Patterns on Ceilings
Polaris is a system for providing accurate orientations for mobile phones in indoor environments. It achieves this by applying computer vision techniques on ceiling images, and uses constant ceiling patterns of a building as orientation references. Since ceiling patterns are universal and unrelated to magnetic fields, Polaris can provide accurate orientations for mobile devices even under severe magnetic interferences. The achieved accuracy is lower than 4.5 degree, 3.5X better than raw compass readings.
(Update 2013: A later improvement leverages online map services rather than crowd-sourcing to acquire building orientations. This improvement further increases Polaris' accuracy to lower than 1.5 degree, capable of supporting numerous ubiquitous and augmented reality applications. See our Ubicomp 2013 paper.) [HotMobile 2012 paper]

PANDAA:Physical Arrangement-Detection for Networked Devices through Ambient Sound Awareness
PANDAA is a zero-configuration spatial localization system for networked devices based on ambient sound sensing. After initial placement of the devices, ambient sounds, such as human speech, music, footsteps, finger snaps, hand claps, or coughs and sneezes, are used to autonomously resolve the spatial relative arrangement of devices using trigonometric bounds and successive approximation. Using only time difference of arrival measurements as a bound for successive estimations, PANDAA is able to achieve an average of 0.17 meter accuracy for device location in the meeting room deployment. Ubicomp 2011 Best Demo Award [Ubicomp 2011 paper][SIGCOMM 2011 demo paper][Ubicomp slides]

SensorFly: A Collaboratively-Mobile Aerial Sensor Network
SensorFly is a controlled-mobile aerial sensor network platform for indoor emergency response application. The miniature, low-cost sensor platform has capabilities to self deploy, achieve 3-D sensing, and adapt to node and network disruptions in harsh environments. In our IPSN paper, we describe hardware design trade-offs, the software architecture, and the implementation that enables limited-capability nodes to collectively achieve application goals. Through the indoor fire monitoring application scenario we validate that the platform can achieve coverage and sensing accuracy that matches or exceeds static sensor networks and provide higher adaptability and autonomy.
We are currently designing and prototyping the 5th generation of SensorFly hardware and software. Please stay tuned for an update of our new platform. Sensys 2009 Best Demo Award [IPSN 2011 paper][video]

CoughLoc: Location-Aware Indoor Acoustic Sensing for Non- Intrusive Cough Detection
CoughLoc is a ubiquitous acoustic sensing system for continuous cough detection using a wireless sensor network. CoughLoc shows how knowledge of sound source locations can be leveraged to improve the detection accuracy of sound events caused by mobile users. Experiments in indoor environments show this system achieves over 90% cough detection performance under quiet backgrounds, 1.6 times higher performance compared to a baseline approach without location information. [MobiSense 2011 paper][slides]


Imirok: Real-Time Imitative Robotic Arm Control for Home Robot Applications
Training home robots to behave like human can help people with their daily chores and repetitive tasks. Imirok is a system to remotely control robotic arms by user motion using low-cost, off-the-shelf mobile devices and webcam. The motion tracking algorithm detects user motion in real-time, without classifier training or predefined action set. I prototyped a robotic arm with 5 degree-of-freedom. Human gestures are recognized using motion tracking algorithms in OpenCV. A webcam is used to capture human gestures, then the inferred gesture sequences are transferred to a remote control unit (SunSPOT) through 802.15.4 radio, which remotely controls the robot arms.

Experimental results show that the system achieves 90% precision and recall rate on motion detection with blank background, and is robust under the change of cluttered background and user-to-camera distance. [PerCom 2010 paper]

Cortina: Collaborative Context-aware Indoor Positioning Employing RSS and RToF Techniques
Cortina is an energy-efficient indoor localization system, which leverages a wireless sensor network to support navigation and tracking applications. My colleagues and I developed a hybrid ranging system, which incorporates both RSS and RToF (round-trip time of flight)-based techniques.

To overcome indoor multi-path effects, we designed and implemented algorithms to take into account various context information. The system was evaluated in an indoor area over 2000 square meters instrumented with twenty-six fixed nodes. Evaluation results show the system achieved 2.5m accuracy in a pedestrian tracking application. [PerCom 2010 paper]


Publications

Journal Papers

Zheng Sun, Aveek Purohit, Philippe De Wagter, Irina Brinster, Chorom Hamm, and Pei Zhang. “PANDAA: A Physical Arrangement Detection Technique for Networked Devices through Ambient-Sound Awareness”. In ACM SIGCOMM Computer Communication Review, Volume 41 Issue 4, August 2011. [pdf]

Zheng Sun, Aveek Purohit, Raja Bose, and Pei Zhang. “Spartacus: Spatially-Aware Interaction for Mobile Devices Through Energy-Efficient Audio Sensing”. Invited journal article at ACM SIGMOBILE Mobile Computing and Communication Review, in preparation.

Conference and Workshop Papers

Zheng Sun, Shijia Pan, Yu-Chi Su, and Pei Zhang. “Headio: Zero-Configured Heading Acquisition for Indoor Mobile Devices Through Multimodal Context Sensing”. ACM Ubicomp 2013, Zurich, Switzerland, September 2013 (direct acceptance, acceptance rate = 23.4%). [pdf]

Zheng Sun, Aveek Purohit, Raja Bose, and Pei Zhang. “Spartacus: Spatially-Aware Interaction for Mobile Devices Through Energy-Efficient Audio Sensing”. ACM MobiSys 2013, Taipei, Taiwan, June 2013 (acceptance rate = 33/210 = 15.7%). [pdf]

Aveek Purohit, Zheng Sun, and Pei Zhang. “SugarMap: Location-less Coverage for Micro-Aerial Sensing Swarms”. ACM IPSN 2013, Philadelphia PA, April 2013 (IP track, acceptance rate = 24/115 = 20.9%). [pdf]

Aveek Purohit, Zheng Sun, Shijia Pan, and Pei Zhang. “ SugarTrail: Indoor Navigation in Retail Environments without Surveys and Maps”. IEEE SECON 2013, New Orleans, USA, June 2013 (acceptance rate = 51/173 = 29.5%). [pdf]

Zheng Sun, Aveek Purohit, Shijia Pan, Frank Mokaya, Raja Bose, and Pei Zhang. “Polaris: Getting Accurate Indoor Orientations for Mobile Devices Using Ubiquitous Visual Patterns on Ceilings”. ACM HotMobile 2012, San Diego, CA, February 2012 (acceptance rate = 14/68 = 20.6%). [pdf]

Zheng Sun, Aveek Purohit, Kaifei Chen, Shijia Pan, Trevor Perring, and Pei Zhang. “PANDAA: Physical Arrangement Detection of Networked Devices through Ambient-Sound Awareness”.ACM Ubicomp 2011, Beijing, China, September 2011. (acceptance rate = 50/302 = 16.6%. Also won Best Demo Award at Ubicomp 2011.) [pdf] [slides]

Zheng Sun, Aveek Purohit, Kathleen Yang, Neha Pattan, Dan Siewiorek, Asim Smailagic, Ian Lane, and Pei Zhang. “CoughLoc: Location-Aware Indoor Acoustic Sensing for Non-Intrusive Cough Detection”. MobiSense workshop of Pervasive 2011, San Francisco, CA, June 2011.[pdf] [slides]

Aveek Purohit, Zheng Sun, Frank Mokaya, and Pei Zhang. 2010. “SensorFly: Controlled-mobile Sensing Platform for Indoor Emergency Response Applications”. ACM IPSN 2011, Chicago, IL, April 2011. (SPOTS track, acceptance rate = 9/35 = 25.7%, Also won Best Demo Award at SenSys 2009.) [pdf]

Zheng Sun, Rick Farley, Telis Kaleas, Judy Ellis, and Kiran Chikkappa, “Cortina: Collaborative Context-aware Indoor Positioning Employing RSS and RToF Techniques”. IEEE PerCom 2011, Seattle, WA, March 2011.  [pdf]

Heng-Tze Cheng, Zheng Sun, and Pei Zhang, “Imirok: Real-Time Imitative Robotic Arm Control for Home Robot Applications”. IEEE PerCom 2011, Seattle, WA, March 2011. [pdf]

Aveek Purohit, Zheng Sun and Pei Zhang, “SensorFly: A Collaboratively-Mobile Sensor Network”. Cylab Partners Conference, Carnegie Mellon University, Pittsburgh, PA, September 2010.

Zheng Sun, Zhiqiang He, Ruochen Wang, and Kai Niu, “A Heuristic Scheduling Scheme in Multiuser OFDMA Networks”, IEEE VTC ’08 Fall, Calgary, Canada, September, 2008. [pdf]

Ruochen Wang, Zhiqiang He, Zheng Sun, Shan Lu, and Kai Niu, “A Revenue-Based Low-Delay and Efficient Downlink Scheduling Algorithm in OFDMA Systems”, IEEE VTC ’08 Fall, Calgary, Canada, September, 2008.

Zheng Sun, Wenjun Xu, Zhiqiang He, and Kai Niu, “Criteria on Utility Designing of Convex Optimization in FDMA Networks”, the 3rd BWAW workshop of IEEE ICC 2008, Beijing, China, May, 2008. (Acceptance rate = 16/73 = 21.9%) [pdf]

Zheng Sun, Xiaohong Huang, and Yan Ma, “Load Balancing Strategies to Solve Flowshop Scheduling on Parallel Computing”, IC-BNMT 2007, Beijing, China, September, 2007.

Posters and Demos

Shijia Pan, Yulai Shen, Zheng Sun, Priya Mahajan, Lin Zhang, and Pei Zhang, "Saving Energy in Smart Commercial Buildings through Social Gaming". ACM Ubicomp 2013, Zurich, Switzerland, September, 2013.

Zheng Sun, Aveek Purohit, Philippe De Wagter, Irina Brinster, Chorom Hamm, and Pei Zhang. “PANDAA: A Physical Arrangement Detection Technique for Networked Devices through Ambient-Sound Awareness”. In ACM SIGCOMM '11.

Zheng Sun, Aveek Purohit, KathleenYang, Neha Pattan, Dan Siewiorek, Asim Smailagic, Ian Lane, and Pei Zhang, “VMA: Indoor Acoustic Sensing Platform for In-home Patient Monitoring”. ACM MobiSys 2010, San Francisco, CA, June, 2010.



Media Coverage

My research has been featured in the following tech news and reviews.



Useful Links

Here is a list of useful links to great people, fun stuff, or inspiring ideas that I have been collecting over the years.

Surviving Ph.D.
PhD Thesis Structure and Content