Personal Homepage of

Marcus Chang

Marcus Chang, PhD
Postdoctoral Fellow
Hopkins interNetworking Research Group (HiNRG)
Johns Hopkins University
3400 North Charles Street, Shaffer 200D
Baltimore, MD 21218
 Research Interests
My main research interests revolve around cyber-physical systems in genereal and embedded wireless sensor networks programming in particular. As wireless computational devices that interact with humans and the environment proliferate through our physical world these devices will continue to have a growing impact on our society. Especially the continuing miniaturization and increasing processing power available on consumer devices enables new and exciting research areas within mobile computing.
 Selected Previous Projects
  • HealthOS: A Platform for Pervasive Health Applications Jong Hyun Lim, Andong Zhan, Evan Goldschmidt, JeongGil Ko, Marcus Chang, Andreas Terzis, Second International Workshop on Mobile Systems, Applications, and Services for Healthcare, November 2012 (PDF).
    One of the challenges faced in electronic healthcare is the lack of interoperability between devices from different manufactures. We devised a framework that acts as both information storage and access hub but also a set of drivers to help facilitate cross-vendor integration and a framework library to ease driver development. This will enable physicians to set up trend finders that correlate data from multiple sources and automatically receive notifications when certain criteria are met.
  • Accurate Caloric Expenditure of Bicyclists using Cellphones Andong Zhan, Marcus Chang, Yin Chen, Andreas Terzis, Proceedings of the 10th ACM Conference on Embedded Networked Sensor Systems, November 2012 (PDF).
    One problem faced by people interested in keeping track of their workout is that calorie burn cannot be measured directly but only inferred indirectly, for example through oxygen consumption or heart rate monitoring. However, using only an accelerometer and GPS we were able to create a software based calorie sensor running on a smartphone. Due to my background in physics, I was able to find a model that would allow us to fuse our measurement with the local weather report and high resolution elevation maps into a single energy estimate. By leveraging the flexibility of the software based sensor we were able to devise a sensor that could easily be calibrated to different bikes and bikers.
  • Meeting Ecologists Requirements with Adaptive Data Acquisition. Marcus Chang, Philippe Bonnet, 8th ACM Conference on Embedded Networked Sensor Systems (SenSys 2010), November 2010 (PDF).
    A major part of my PhD thesis was the interpretation of high-level objectives and the following sub-tasking of an environmental monitoring sensor network. The problem I wanted to address was the steep learning curve faced by ecologist. Ideally, the ecologists should only have to worry about what to sample and the sensor network should be intelligent enough to handle the how. To accomplish this I devised an AI controller that would take the ecologists high-level sampling requirements and adaptively task a sensor network to achieve this goal, given the system's current energy and communication constraints.
  • Characterizing Mote Performance: A Vector-Based Methodology. Martin Leopold, Marcus Chang, Philippe Bonnet, 5th European conference on Wireless Sensor Networks (EWSN 2008), January 2008 (PDF).
    During my PhD studies I worked on characterizing software components' computational load and power consumption across different platforms. We devised a method to accurately measure the power consumption of individual software components on one device and predict the energy consumption on another. This is an important step in understanding the different software and hardware layers' power consumptions and in finding the most suitable devices in terms of power consumption for any given workload.

© Marcus Chang, 2013.  Valid HTML 4.01 Transitional Valid CSS!