Routing Method For Opportunistic Networks Based On Deep Learning

With the rapid development of wireless network,more and more applications have gradually penetrated into people's work and life.In order to meet the communication needs,Opportunistic Networks(Opportunistic Networks)emerged.In Opportunistic Networks(oppnet),messages are transmitted in the way of receiving,carrying and forwarding.Due to the uncertainty of node movement,it leads to lower transmission rate and larger delay.Therefore,an intelligent and secure storage and forward technology is needed.This paper proposes a routing method based on deep learning,which improves the message passing rate,average hops and communication overhead rate.In order to determine the next best hop in the path,a route selection method based on Long Short Term Model(LSTM)is proposed to select the most suitable node for message transmission.It consists of input stage,learning stage,decision-making stage and transmission stage.In the input stage,five parameters are registered for the nodes near the source node.In the learning stage,the network is trained with the input parameter set defined above.In the decision-making and transmission phase,the decision is made whether adjacent nodes forward messages to specific nodes.The experimental data set is simulated on Opportunistic Networking Environment(ONE).The results show that the network model designed in this paper can improve the message passing rate,average hops and communication overhead rate.With the increase of injected network messages,the previous messages recorded by LSTM model will be covered.A route selection method based on Bi-directional Long Short Term Model(BiLSTM)is proposed.The model calculates the input data forward and backward twice,and finally outputs the final results at the corresponding time.By simulating the experimental data set on ONE platform,we can see that BILSTM needs to be trained more times to make the network stable,but the accuracy is obviously higher than LSTM,and the stability is better.