Sleeping scheduling has been widely employed in target tracking due to its energy conservation. However, the randomness of target's trajectory makes it difficult to implement with accuracy and real time guarantee. We propose TTVT, a novel, simple and efficient tracking technique. TTVT first constructs a Voronoi based network model, then makes nodes in the Voronoi polygon that the target is in work and others sleep. The target is hence detected by nodes closest to it. TTVT further presents a weighted centriod based algorithm to locate the target with the chosen nodes and reduce the influence of data noise on localization accuracy. We have implemented TTVT, and our extensive simulation show that it outperforms similar schemes with much lower location error.