Prediction Model for Tiered Accuracy Requirements in Large-scale Object Tracking Sensor Networks

In large-scale object tracking wireless sensor networks, multiple mobile nodes will bring large amount of communication overheads in maintaining the accuracy of localization tracking, which would possibly affect the collection and dissemination of the tracking data, seriously. Existing seminar works purely focus on achieving optimal tracking accuracy and suffer from large amount of protocol overhead, due to lacking of adaptations on various accuracy requirements for different tracking applications. In this paper, we propose a prediction based object tracking protocol for tiered accuracy requirements. The tired accuracy requirements are perceived by each entity so as to establish a unified predicted model for the moving objects and the sink nodes. Moreover, the network entities could share the corresponding model parameters to others related, thus, all the interested entities use the same model to predict the moving objects and the corresponded sink location, so as to achieve a precise location. In this approach, the system achieves efficient data distribution and information sharing to reduce communication overheads, such that tiered application requirements are satisfied. An optimal model on vehicular mobility in analytical form is made, employing the trace generated by kinematic model and the data collected in real experiments. In that, we set up a simulation environment in evaluating our protocol using Qualnet, a validated commercial simulation platform. Simulation results show that, the proposed algorithm can effectively improve network efficiency and successfully satisfy different level of application requirements.