Prototype Development for Adaptive Solar Tracking and Optimization of Data Communication Protocol

To provide maintenance-free operation and a longer system lifetime of internet-of-things (IoT) network, utilizing solar energy for powering data transceivers in IoT device is important. A prototype for an adaptive solar tracking and efficient data communication system empowered by the harvested solar energy was developed by the capstone project team at Marshall University. In general, the harvested renewable (solar) energy can be random in time and in amount. The team tackled the randomness of the harvested energy with an efficient algorithm assisted by artificial intelligence techniques. The prototype is designed on Raspberry Pi and Arduino development boards, and the overall system comprises a solar tracking module, data transmitter, and receiver supported by omni-directional antennas, power banks, web-portal at the receiver, etc. The harvested energy powers up an integrated sensor-mounted data collection and wireless transmitter module. A long short-term memory (LSTM) layer incorporated deep neural network (DNN) accurately predicts the optimal time instants at which the data should be transmitted to a receiver module over a wireless channel in order to maximize the data throughput. The algorithm for the optimal control of solar tracking was developed in Python using open source DNN modules such as TensorFlow (Lite) to implement LSTM-assisted DNN for optimization of data transmission instances. A web-portal with a database shows the real-time data collection at the receiver end and exhibits dynamic performance analysis based on the collected information. The experimental results show the developed prototype significantly enhanced the efficiency of energy harvesting as an atomic part of a massive internet-of-things (IoT) network, an integral part of the fifth-generation (5G) cellular communication systems. Future enhance is discussed.