Multi-target Real-time Tracking Based On Distributed Measurements Of WSNs

Wireless sensor networks (WSNs) technology has integrated with sensor technology,embedded system technology, short-range wireless communication technology, anddistributed data fusion technology, which acquire, percept and monitor environmental changeand information by embedded systems, and transmit data packet through a multi-hoprouting or over the Internet to a remote user terminal. The application prospect of WSNs isvery bright. This technology can be used in military, industrial, agricultural and other fields,which is considered to be the first item of the future emerging technologies.Location and tracking of the moving target is an important application in wirelesssensor networks, due to the demanding real-time, which is also a technically challengingresearch topic. In this research filed, the predecessors have made a lot of work and haveachieved a lot, but many of them are limited to theoretical results, without the verification ofthe physical platform. In this article, we build a multi-objective real-time tracking platformbased on wireless sensor network. The entire system consists of a group of distributed nodeswhich are equipped with a set of ultrasonic sensors, a monitoring base station and multiwheeled robots as moving targets. The experiment performance shows that the platform caneffectively track multi-target simultaneously.In order to avoid conflict caused by the multi-objective real-time tracking network, thispaper presents an adaptive clustering and inter-cluster negotiation node scheduling policy. Thewireless communication of the entire system adopted ZigBee protocol. By one-waybroadcasting synchronization, the entire system achieves clock synchronization between thebase station and all nodes. In the indoor environment, ZigBee communication attenuationmodel parameter has been identified and packet loss model in the dynamic tracking process isanalyzed by two state Markov chain model. During the Multi-target tracking process,inter-cluster time division transfer protocol can decrease the probability of conflict broughtabout by the channel multiplexing, and reduce the packet loss rate of the entire wirelessnetwork. For the characteristics of distributed measurement, asynchronous measurement datasynchronization and extended Kalman filter localization algorithm is applied to thissystem. Adaptive clustering scheduling policy separates measurement data into differentclusters accordance with the moving targets, to achieve real-time tracking of multiple targetsin the entire network.The experimental results show that the Kalman filter can effectively track the movingtarget without velocity jump. Mobile robot in the grid change will bring the switching of the observation model. Comparing to the least-squares localization algorithm, Kalman filterlocalization algorithm can achieve a smooth handoff. When three robot are moving at thespeed of0.5m/s in the monitored area, the system can effectively track three targets, andtransmit positioning results to robots. This system can navigate the three targets and lay thefoundation of formation feedback control.