PTA: A Predictive Tracking Algorithm in Wireless Multimedia Sensor Networks
Résumé
In this paper, we propose a new Predictive Tracking Algorithm for Wireless Multimedia Sensor Networks named PTA. PTA is a complete tracking algorithm that implements a five-step process: wake up, detection, localization, prediction, and next sensor selection. Each step has an important role in the tracking process. PTA attempts to find the trade-off between tracking accuracy and energy conservation. In this algorithm, the prediction phase is performed using a Kalman Filter, which is a recursive state estimator. Using simulations, we show the efficiency of the proposed algorithm in both trajectory prediction as well as energy saving. Moreover, we perform a comparative study between PTA and existing solutions: 1) BASIC solution where all the Camera Sensors are always active, 2) Optimal Camera Node Selection (OCNS) which is a cluster-based mechanism based on probabilistic node election. And Finally 3) PAM, another predictive scheme based on Autoregressive Model. Our results show that PTA increases the tracking accuracy up to 30% compared to existing solutions, while reducing energy consumption down to 589.16 Joules. Therefore, PTA yields an accurate upcoming position prediction, and is more efficient than existing predictive models.