This project focuses on processing the data generated by Monjolo nodes. Each energy-harvesting Monjolo node activates at a rate that is proportional to the energy source it is harvesting from, and increments a counter at each wakeup. The node periodically wireless transmits this counter value to a central receiver. Each node is uncalibrated and therefore the activation frequency does not provide accurate power measurements. However, when many of these streams are combined with knowledge of the “power tree” (which measurements are aggregates of others) and a calibrated total, we hypothesize that we can disaggregate the power draw of each individual load.
In order to test our system we need real-world power traces and ground truth to compare our results to. Therefore, I have been instrumenting my kitchen and living room with Watts Up? power meters and our Monjolo sensors. While not only helping with the Delta Flow project, this also illustrates the negatives of currently available submeters and the need for a better solution.