Research On The Detection Performance Of Wireless Sensor Networks When The Communication Is Unstable

As an intelligent grid system,wireless sensor network is composed of a large number of sensor nodes deployed in the monitoring area,which can perceive and detect objects in the detection area.Wireless sensor networks are widely used in industrial monitoring,office automation,medical services,environmental protection,logistics management,military security and other aspects due to their reliability and other advantages.In the related research and application of wireless sensor networks,distributed target detection is an important research area.The sensor nodes in the distributed detection network will initially process the observations obtained through the detection,and the processed information will be sent to other nodes for further information fusion,and the fusion center will make the final fusion decision based on the collected information.The detection network of this structure helps to remove redundant data in the detection,reduce energy consumption and communication bandwidth requirements,and extend the service life of the detection network.This thesis mainly studies the distributed target detection problem based on hypothesis testing in the wireless sensor network.A large number of studies have shown that if there is unstable communication in the wireless sensor network,the detection performance will be affected to different degrees with the different factors that cause the unstable communication.This thesis focuses on the detection performance of the M-ary tree wireless sensor detection network when the communication is unstable.The main content includes the following two aspects:1.An M-ary tree detection network with channel noise is analyzed,and the bit-error probability is used to simulate the influence of channel noise on the network.A feedback structure is introduced and the detection performance is analyzed.It is proved that the error probability of the node will gradually converge with the increase of the feedback time,and the convergence rate is preliminary analyzed.Next,the single-step evolution nature of the node error probability is analyzed.In order to make the error probability of the fusion center converge to a fixed value greater than 0 and less than 1/2,the constraint conditions that the bit-error probability should follow are given.After introducing the feedback structure,the detection network can more easily follow this condition,and when the detection network follows this condition,the convergence value of the error probability of the fusion center after feedback is introduced will be smaller than that without feedback,which proves that the detection network with feedback has a certain inhibitory effect on channel noise.2.Considering other factors that cause unstable communication,this thesis studies the Mary tree detection network with underreporting information.The link failure probability is used to simulate the influence of information underreporting on the network,and a feedback structure that does not increase the communication bandwidth is introduced.As the feedback time increases,the error probability values of all nodes in the network gradually stabilize.When the height of the tree is big enough,the fusion center can make a decision that the error probability is 0.In the subsequent analysis,it is pointed out that after a sufficiently long feedback time,the error probability of the fusion center is reduced compared with the case without feedback,so that the detection performance is improved.