EADS-EHWSNs: Efficient Energy-Based Adaptive Duty Cycle Scheme for Energy-Harvested Wireless Sensor Networks

In dynamic energy harvesting wireless sensor network systems (EHWSNs), the amount of harvested energy depends on the radiation parameters of the antenna. A radio frequency (RF) energy-harvesting antenna must be portable, compact, and integrable to enable easy embedding in the network which is a new advancement for EHWSNs. In the literature, most medium access control (MAC) protocols for EHWSNs were studied without taking into consideration the effect of the dynamic nature of energy harvesting. Most of the existing works didn't study the benefits that both the predicted energy harvesting factor and the integration of the duty cycle (DC) of the sensor nodes with the predicted amount of harvested energy might provide for enhancing the system performance. This paper presents EADS-EHWSNs, a MAC protocol that is prediction-based energy harvesting with an adaptive duty cycle (ADC) depending on the predicted EH amount and the geographical distance to the cluster head or sink in case the node is the cluster head. The proposed EADS-EHWSNs incorporate a machine learning model, namely Deep Learning Neural Network (DNN), to achieve good prediction accuracy under variant EH amounts. Simulation has been carried out using GreenCastalia to evaluate the network performance of the proposed EADS-EHWSNs. The network performance metrics include packet delay, network throughput, packet delivery ratio, and network energy consumption. In comparison with the three state-of-the-art EHWSNs protocols, results show that EADS-EHWSNs provide the lowest average end-to-end delay for the high and low-priority packets, network energy consumption by the approximate average of 21%, 18%, and 74%, respectively. In regard to network throughput and packet delivery ratio, EADS-EHWSNs outperforms the three protocols by an approximate average of 39% and 44%, respectively.