Kinetic Energy Measurements
Contributors: Mina Cong, Kanghwan Kim, Maria Gorlatova, John Sarik.
The long-term goal of this project is to create small, flexible, energetically self-reliant devices that can be attached to traditionally non-networked objects. These objects will communicate with one another using ultra-wideband impulse radio technology, thus forming rechargeable, wireless sensor networks. To help us better understand the properties of various energy sources and their impact on energy harvesting adaptive algorithms, we collected acceleration traces from different participants. The volunteers were asked to perform normal daily routines while comfortably carrying SparkFun Electronics ADXL345 accelerometer boards. For our long-term studies, we collected over 200 hours of acceleration information in 25 days from 5 participants. The data simulates the natural motions that participants' belongings (keys, phone, or wallet) experience on a daily basis.
The description of the energy measurements and some analysis appear in:
- M. Gorlatova, J. Sarik, G. Grebla, M. Cong, I. Kymissis, and G. Zussman, Movers and Shakers: Kinetic Energy Harvesting for the Internet of Things, in Proc. ACM SIGMETRICS '14 (to appear) June 2014
We would appreciate it if you cite this paper when publishing results that use the provided measurements.
Setup Description and Data Format
For the logging of measurements, we use ADXL345 tri-axis accelerometers, Atmega328P microcontrollers and microSD cards located on SparkFun ADXL345 evaluation boards. The accelerometer sensors record acceleration along the x, y, and z axes with a 100 Hz sampling frequency. Experiments were conducted with the sensing units placed in multiple locations.
The format for the trace files is as follows:
- The first column is the time since the measurement started (in seconds).
- The second column is the acceleration along x axis (in g) .
- The third column is the acceleration along y axis (in g) .
- The fourth column is the acceleration along z axis (in g) .
Measurements were collected from five participants of different ages, physiques and means of transportation to the same laboratory work location. The participants included two undergraduate students who commuted by foot as well as an undergraduate student, graduate student, and software developer who commuted by train. As mentioned earlier, the five participants were asked to carry the sensing units in a comfortable manner over a period of 25 days. The dominant motion frequency of all collected traces ran in the range of 1.92-2.8 Hz, which corresponded to human walking. Using collected data, we were able to calculate the average power a harvester could generate over a trace's length as well as over a 24-hour period ( [ link to paper PDF ] ).
A collection of .zip trace files for all participants is available at https://www.dropbox.com/sh/wrd0te977rb5pw2/UoJkFwRbOm. The traces correspond to the participants listed in Table 3 in [ link to Paper PDF ].
- Participant M1 (undergraduate student, lived on campus):
Days: 5, Hours: 60.4.
- Participant M2 (undergraduate student, commuted to campus):
Days: 3, Hours: 27.7.
- Participant M3 (undergraduate student, lived on campus):
Days: 9, Hours: 62.0.
- Participant M4 (graduate student, commuted to campus):
Days: 7, Hours: 80.1.
- Participant M5 (software developer, commuted to office):
Days: 1, Hours: 11.0.
Related datasets include additional measurement sets as well as papers on related studes.
Questions? Please contact Maria Gorlatova, maria.gorlatova at caa dot columbia dot edu.