ECFTF

I. Code description

The code includes two parts: energy allocation algorithms and energy prediction algorithms. The following algorithms were implemented:

Energy allocation

• (1) FTF: Fast Time Fair Energy Allocation algorithm of our paper.
• (2) MAllEC: implementation of Maximum Allowed Energy Consumption algorithm from "Mallec: Fast and optimal scheduling of energy consumption for energy harvesting devices," IEEE Internet of Things Journal, vol. 5, pp. 5132–5140, Dec 2018.
• (3) buchli: implementation of Periodic Optimal Control algorithm from "Optimal power management with guaranteed minimum energy utilization for solar energy harvesting systems", Buchli et al, DCOSS 2015
• (4) gorlatova: implementation of Progressive Filling algorithm from "Networking low-power energy harvesting devices: Measurements and algorithms", Gorlatova et al, INFOCOM, 2011.

Attention: The implementation of algorithm (2)(3)(4) refers to the source code eh_python of V. Cionca.

Energy Prediction

• LSTM based predictor.

Dataset

We collect 200 hours of metro vehicle vibration energy harvesting data in total at intervals of 2 min.

II. Code File structure

• /algorithms -- Source code of algorithms.
• /data -- The metro vibration energy harvesting dataset we used.
• /LSTM -- LSTM energy predictor based on tensorflow.
• /saved_models -- Well trained LSTM model.
• /simtest -- Code for offline and online experiments
  • /offline -- Energy allocation algorithms + offline dataset
  • /online -- Energy allocation algorithms + Energy preditor

III. Code environment requirements

• Windows 10, IDE pycharm 2016.2.2 or higher
• python 3.6.6
• Tensorflow 1.10.0
• keras 2.2.4
• numpy 1.16.2
• pandas 0.23.4