Research On Target Tracking Algorithm For Energy Efficiency Optimization In Wireless Sensor Network

As an important technology of wireless sensor network,target tracking has been widely used in military tracking,environmental surveillance and intruder tracking.Because sensor nodes are resource-constrained,the tracking accuracy and energy consumption of wireless sensor network become the primary challenge to target tracking.Therefore,in order to balance the contradiction between the tracking accuracy and energy consumption,this thesis is aiming at reducing the number of active sensor nodes and deploying the sensor nodes energy-efficiently on condition of the tracking performance is keeping meet.Due to the interference of process noise,the degree of the nonlinearity of the target is different when target is moving in the wireless sensor network.When the degree of nonlinearity of the target is smaller,the requirement for observations would be lower.So activating a fixed number sensor nodes for each observation will waste the energy of wireless sensor network.Therefore a target tracking algorithm for adjusting the number of sensor nodes dynamically is proposed.Firstly the algorithm estimates the error ellipse which target will occur in the next moment according to the prediction covariance.Then the average measurement residual error is regard as the error indicator to dynamically adjust the number of sensor nodes which will be activated in the next time.The simulation results that the algorithm of target tracking based on error ellipse and adaptive selection of sensor nodes can reduce the number of sensor nodes and dynamically adjust the number of sensor nodes at each moment while keeping the tracking accuracy is meet.In order to improve the tracking accuracy further,this thesis proposes a sensor node deployment scheme based on the updating process of information filtering.When the sensor nodes observe the target,the geometric distribution of the sensor-to-target would affect the information that is contained in the observations.The more the information contained in the observation,the more accurate of the target state.Therefore,based on the predicted covariance and the number of sensor which will be activated,the location of sensor nodes where could provide the most information will be obtained,and then the sensor nodes are scheduled to deploy after considering the remaining energy of the sensor node.Finally,activating the sensor node which deployed at the optimal observation position.The simulation results show that the deployment scheme of sensor nodes based on the maximum information can effectively improve the tracking accuracy and further reduce the number of sensor nodes which need to be activated.