Relay Selection Strategy For Energy Harvesting Wireless Sensor Network Based On Neural Network

With the development of Io T technology,wireless sensor network are increasingly used in the military,environmental monitoring and other fields.Sensor nodes can monitor information such as humidity,temperature,precipitation to provide more timely and effective environmental data for various departments.The reliable and efficient wireless sensor network technology is the key to ensure the operation of the monitoring system.Relay Collaboration technology solves the problem of direct communication between sensor nodes and destination node due to long distances.Simultaneous Wireless Information and Power Transfer technology completes the energy exchange between nodes while transmitting information,providing a reliable technical guarantee for sensor nodes to replenish energy and extending the life cycle of the network.The application of Solar Energy technology makes the wireless sensor network energy source diversification,reducing the dependence on the power grid supply and more in line with the concept of green communication.Microwave Power Directional Emission technology allows the energy node adjust the sensor node charging strategy according to the energy sources to achieve the purpose of energy saving.The development of Artificial Intelligence technology for the wireless sensor network relay selection strategy to provide intelligent solutions,making the implementation process with better real-time performance and lower computational complexity.This dissertation analyzes the current status of research on relay selection algorithms for wireless sensor network at home and abroad at first,introduces the relevant technologies,analyzes their functions and characteristics,finally proposes an improved strategy.The main work is as follows:(1)The energy supply source and relay node selection algorithm of energy harvesting wireless sensor network(EH-WSN)are the key factors restricting the network life cycle.In order to improve the renewable energy utilization of EH-WSN,an energy node(EN),which equipped with solar panels and power grids,is added to the EH-WSN.The multi-relay cooperative communication model of EH-WSN is constructed by the decode-and-forward relay protocol and the improved power split receiver structure.The directional energy supply from EN to relay nodes(RN)is realized based on two-dimensional linear phased array antennas.According to the different energy sources of EN,the charging strategy is dynamically adjusted and the optimal relay selection algorithm under the condition of maximizing the network life cycle is proposed.The simulation results show that the network life cycle using the optimized relay selection algorithm is longer than the Simultaneous Wireless Information and Power Transfer Wireless Sensor Network(SWIPT-WSN),and the renewable energy utilization rate can reach a maximum of 21% in a day.(2)In order to improve the relay selection efficiency of EH-WSN,the relay selection model based on feed-forward artificial neural network(ANN)and gated recurrent neural network(GRU)are established,which through the back propagation algorithm and cross entropy function to correct the model structure.The simulation results show that the correctness of the relay selection based on the ANN model is up to 90%,and the selection efficiency is improved by 92%.Compared with the ergodic relay selection algorithm,the proposed strategy has the advantages of low computational complexity and high real-time performance.The relay selection model based on GRU model has an accuracy of 71%,where the output results have a certain predictive nature due to the input pattern setting of the training data,which provides a further research direction for the relay selection algorithm intelligence.