| As the communication technology keeps evolving,Mobile Ad Hoc Network has become a hot topic in the study of the wireless communication system recently.Mobile Ad Hoc Network originating from the traditional wireless sensor network has attracted more and more attention due to its multiple application scenarios,especially in the military armed battlefield and disaster relief.The network organizes its self and requires no central entity.Each node can enter and exit the network dynamically in cooperation with each other.At the same time,it can transmit and process data packets independently.Therefore,it lays a foundation for the area-oriented Mobile Ad Hoc Network.However,there are the problems that the limited energy of nodes and a large amount of data transmission will increase the load on the nodes of the communication links in area-oriented Mobile Ad Hoc Network.Then,these problems will increase the average end-to-end delay,reduce the packet delivery rate,and reduce the throughput.Accordingly,in order to improve the communication performance of the area-oriented Mobile Ad Hoc Network,the thesis selects the Ad Hoc OnDemand Distance Vector Routing(AODV)as the research object for the following research.In this thesis,the bionic algorithm based on Sparrow Search Algorithm(SSA)heuristically optimizes Ad Hoc On-demand Distance Vector Routing(AODV).Thus,SSA-AODV routing protocol is proposed.The optimization process is as follows.Firstly,the optimal routing path in the network is selected based on the cyclic iteration of fitness function combined by the producer of the sparrow algorithm.Therefore,the nodes on the routing path are "producer nodes".In the selection of the fitness function,the hop number of nodes,the stability of nodes,the signal intensity,and the energy life of nodes are taken into consideration.Secondly,the scrounger routing backup path is created when the "producer node" on the producer path is made use of.When the "producer node" on the routing path becomes the "danger node",the information is sent to the upstream "producer node" to inquire about the scrounger path.The detection of "producer node" to "danger node" is based on the energy of the node,the strength of the signals received,and the number of packets in the queue of the node cache in comparison with their respective thresholds.In this way,the performance the nodes on the producer path is evaluated and the early warning is given.Quality of communication links between nodes on the routing path can be perceived dynamically in this way,effectively reducing the frequency of re-routing discovery caused by communication link interruption,and node congestion.Finally,the SSA-AODV routing protocol is simulated and verified by Network Simulator Version 2(NS2).The simulation results show the network performance of the improved SSA-AODV routing protocol is superior to the original AODV routing protocol.The improved SSA-AODV routing protocol improves the throughput and packet delivery rate,and reduces the average end-to-end delay and average node consumption.In addition,this thesis uses Long Short-Term Memory(LSTM)artificial neural network to predict how long the communication link of adjacent nodes in the Mobile Ad Hoc Network can last.Firstly,the NS2 is used to collect the dataset of the lifetime of the actual communication link between the nodes,their position,and their speed.Then the training dataset and test dataset are combed and obtained.Secondly,MATLAB software is used to program the LSTM network and adjust the network parameters to construct and train the model to predict the lifetime of the communication link between nodes.Finally,the simulation results of MATLAB show that it is feasible to use LSTM to predict the communication link lifetime of adjacent nodes.Therefore,this method can be used to improve the interval size of HELLO message in AODV routing protocol to make the energy utilization of nodes more efficient. |
