Design And Optimization Of MAC Protocol For Energy Harvesting Wireless Sensor Networks

Thanks to the progress of computers,wireless communication and low power sensors,wireless sensor networks(WSN)develop rapidly.However,the development of battery technology is slower than that of communication technology,so energy shortage has become an important factor to restrict the performance of traditional sensor networks.With energy harvesting technology having become matured,energy harvesting wireless sensor networks(EH-WSN),which can extract energy from the environment to make sensor nodes achieve unlimited life,has become a research hotspot.In WSNs with energy constrained,MAC protocols are designed to save energy in order to prolong the network lifetime.EH-WSN is different,and the design idea of protocols in EH-WSN is to maximize run performance of the network under given environmental energy limiting conditions.This paper proposes two MAC protocols-CSMA with the ability to adjust back-off window based on energy harvesting capability(EH-CSMA)and MAC protocol with Duty-cycle Scheduling based on Solar Energy Prediction(DSEP-MAC).They provide new mechanisms in EH-WSN to manage multiple nodes to access channels and to adjust the duty cycle of nodes dynamically.EH-CSMA decides the back-off window size based on relative position of energy harvesting capability in the entire network and residual energy of nodes,and enhances the ability of nodes with fewer energy to access channel.EH-CSMA improves the equity index of network while maintaining throughput performance.According to the simulation results,EH-CSMA is suitable for deployment in the EH-WSN where energy distribution is extremely uneven and equity index is paid attention to.On the basis of algorithm named Duty-cycle Scheduling based on Residual energy(DSR),DSEP-MAC adjusts the duty cycle of nodes to be more in line with the changing trend of energy acquisition by predicting the next period of time of solar energy.DSEP-MAC protocol uses the BP(Back Propagation)neural network training solar model,and is able to predict the energy harvesting power more accurately.It adjusts the time of nodes to sleep according to the size of current residual energy.Nodes with higher energy have shorter time to sleep,which is for the purpose of increasing activity and improving network performance.At the same time,protocol also provides conversion strategy between DSEP-MAC and DSR-MAC to reduce the negative impact on the node when BP neural network model makes a large error on predicting the solar energy.Simulation results show that DSEP-MAC has better network performance.