Predictive lifetime-based node management in wireless sensor networks

This paper presents a dynamic tracking scheme that exploits the predictability of nodes´ activity in the wireless sensor networks. A node future activity is predicted by itself. To more realistically represent the various degrees of activity correlation in time, a Gauss-Markov prediction model is used. On the basis of result of the theoretic analysis, we develop a scalable and robust algorithm to cope with the sensor node choice problem for target location under requirements of the cost limitation and high coverage rate. The experimental results reveal that, for not very large sensor fields, the algorithm can solve the node choice problem very well under the requirement of the cost limitation and the complete coverage and also extend the network´s living time greatly. Moreover, even in larger sensor fields, it can also find the way to solve the problem with little more time spending.