Uncertainly Data Estimation Algorithm Research In WSN

Wireless sensor network(WSN)is composed of a large number of sensor nodes in monitoring area.The nodes are used to collect and manage the data in monitoring area,then they sent the data to the tester.The wireless sensor nodes have many advantages,including low cost and small volume,so it has been widely applied in many fields.Due to characteristics of WSN and influence of the external environment,some sensory data in WSN are missing.So estimating the missing data in WSN is crucial.A improve multiple-regression-model bases on the spatial correlation algorithm(ISM)is proposed.On the basis of multiple-regression-model,it uses the adaptive linear unit learning algorithm of neural network to optimize the coefficient of regression model.The estimated value has the properties of higher accuracy and stability.However,for temporal correlation of sensor data,it has the problem of slow-convergence speed if coefficient of the regression model is computed by using the coefficient optimization method of ISM algorithm.To improve the situation,a improve multiple-regression-model bases on the temporal correlation algorithm(ITM)is proposed by introducing Aitken accelerate convergence method.The experimental results show that ITM algorithm has higher accuracy and convergence speed.Finally,ISTM algorithm which bases on the space-time correlation is proposed by combing the temporal correlation and spatial correlation of sensor data.And it can reduce the errors obtained by the single algorithm.The experimental results show that the algorithm can accurately estimate the missing data in WSN under the condition of the low computational complexity.So ISTM algorithm has higher precision,quicker constringency speed and better stability performance.