A Localization Algorithm Based on Dynamic Grid Division for Mobile Wireless Sensor Networks

Localization is extremely critical for many applications in wireless sensor networks. Without the location of sensor nodes, collected information is valueless. Meanwhile, location information is also helpful for many network operations such as clustering, topology control, and geographical routing. Localization is an extensively studied problem in wireless sensor networks. Some localization algorithms for static wireless sensor networks have been proposed. However, little study has been done about the localization in mobile wireless sensor networks. A Monte-Carlo localization algorithm is presented based on dynamic grid division for wireless sensor networks, in which the nodes can move randomly. In the presented algorithm, when the number of received one-hop anchors is lager than a threshold, a farthest distance selecting algorithm is used. Only these selected anchors take part in localization and data transmitting, and they can conserve some energy. Then sampling area is created based on selected or all received anchor information, a grid division is made, and the maximum sampling number is computed with cell counts. Next, sampling is made in the created area and filtering is done with a mobility model of error compensation, which can improve the sampling efficiency and reduce the computing overhead. The simulation demonstrates that the proposed algorithm provides better performance in localization precision, computing overhead, and energy consumption.