Research On IoTs For Environmental Monitoring Based On Wind Harvesting And Neural Network

It is necessary to select an appropriate routing algorithm to collect,transfer and process the environmental information quickly and accurately.And the clustering routing algorithm is a hot research direction currently.At the same time,with the technology of energy harvesting is maturing gradually,network nodes can utilize the environmental energy to supplement nodes' energy and break the constraints of nodes' battery.At present,scholars have applied the technology of energy harvesting to the clustering routing algorithm successfully.In addition,when the technology of data fusion is applied in the IoTs for environmental monitoring,it can not only eliminate a large number of redundant data,but also improve the efficiency and accuracy of data collection.Moreover,data fusion based on neural networks is the hot research in this field.Therefore,this paper applies the technology of energy harvesting and data fusion based on neural networks to the clustering routing algorithm to improve the whole performance of the IoTs for environmental monitoring.In this paper,a clustering routing algorithm based on wind harvesting and neural networks is designed and implemented.In the monitoring area the wind energy is used to supplement the energy of sensor nodes,and the collecting data are fused from two dimensions of temporal correlation and spatial correlation.First of all,this paper proposes a clustering routing algorithm based on wind harvesting,which is divided into four parts.Firstly,the wind power forecasting algorithm is introduced into the clustering routing algorithm to measure the ability of wind harvesting.And the forecasting models are established by support vector machine,random forest and Back Propagation neural network respectively.Because of the better performance,we select BP neural network forecasting model in the end.Secondly,we take the distance,the residual energy of battery and the ability of wind harvesting into account and select the node which is closer to the sink node,having more residual energy and stronger ability of wind harvesting to be the head node.Thirdly,the data fusion rate is used as a balance factor to select the head node that the energy consumption which is from the normal node to the head node and from the head node to the sink node is minimum.Fourth,the sleep-awake mechanism is set for nodes to utilize the wind harvesting energy reasonably.Then,the collecting data are fused from two dimensions of temporal and spatial correlation.In normal nodes,Recurrent neural network is used to fuse the temporal features of the collecting data.In head nodes,Convolutional Neural Network is used to fuse the spatial features of the collecting data.And then we use Gated Linear Unit to improve the CNN.Finally,we use the network simulation tool,ONE,to implement the clustering routing algorithm proposed in this paper,which is compared with PHC clustering routing algorithm in three aspects: the number of awake nodes,the whole network residual energy and the sink node receiving data packets.We use the forest fire data set and the air quality data set in the open data resources of the University of California(UCI)Machine Learning Center to implement the experiments.The experimental results show that the proposed fusion algorithm based on temporal and spatial correlation can achieve 82% classification prediction accuracy and best regression prediction results on fire data set,as well as 87% classification prediction accuracy and best regression prediction results on the air quality data set.Meanwhile,the clustering routing algorithm proposed by this paper performs better than the PHC clustering routing algorithm in the three evaluation index.In summary,the above results show that the clustering routing algorithm proposed in this paper is highly feasible and excellent in performance.