Communication-efficient Tracking Model Selection Methods for Multi-model Based Object Tracking Sensor Networks

The tracking technique is one of the most critical techniques in wireless sensor network applications. The methodology of most recently proposed tracking schemes is to predict the object locations based on a single object moving model, and periodically activate nearby sensors to monitor the target. In most real-world situations, a target moves with different movement patterns, and thus, accurately describing the movement of a target needs multiple moving models. Multi-model based Object Tracking Architecture (MOTA) is presented very recently to allow tracking irregularly moving objects by adopting different tracking models regarding movement properties of objects. However, MOTA only considered the framework of dynamically adopting tracking models in wireless sensor networks, but did not minimize the communication cost for data exchange among sensor nodes during monitoring objects. In this paper, we propose two communicationefficient tracking model selection strategies, called Bitmap Min-Max (BMM) Strategy and Aggregation-based Weighted Moving Average (AWMA) Strategy, to minimize the data transmitted in wireless sensor networks based on bitmap and aggregation techniques, respectively. We conduct a set of experiments to compare communication cost for the proposed methods. The results reveal that the proposed methods indeed efficiently minimize communication cost compared with the existing methods.