Efficient Continuous Object Tracking with Virtual Grid in Wireless Sensor Networks

For tracking a continuous object, mass communications are generated in the sensor field because of their huge scale and extensive diffusing property. Due to the severe resource constraints of sensor nodes, the redundant information and communications should be reduced in order to prolong the life time of WSNs. Although there have been numerous studies on continuous object tracking, most of the proposed schemes have focused on reducing the redundant boundary nodes and control messages through static/dynamic clustering in space domain. However, it is not enough to reduce the redundancy only with space domain. Redundantly sampled reporting packets could be a cause of generating redundant long range transmission, which is an important source of energy consumption. For reducing the redundancy, we need to consider not only the method to reduce the redundancy in space domain but also an adaptive sampling scheme in time domain. However, adaptive sampling is difficult to design in continuous object tracking because local sensor nodes hard to figure out how far the continuous object is diffused. To solve this problem, we suggest a new scheme which picturizes the diffusing object by dividing the sensor field into several cells like a pixel on TV and samples reporting time based on pixel image. In addition, we use pictured image of diffusing object for selecting the boundary information of diffusing object in space domain to reduce the redundant boundary information. To present the sensor field as an image, we adopt a virtual grid which divides the sensor field into several cells under the static clustering WSN architecture. The performance of proposed scheme is verified with simulation and the results show that the total number of control messages, reporting packets and boundary nodes can be reduced.