Project Abstract:

Sensor network has emerged as one of the hottest research areas recently. It has a wide range of applications: security sensing in military defense systems, environment monitoring, manufacturing surveillance, human healthcare monitoring… In the project, we investigate two scenarios of deploying a sensor network. In the first scenario, we study the capability of using a terrain of acoustic sensors for enemy target detection and tracking. Sensor coverage areas can be overlapped. A sensor needs to exchange information about its battery life and direction of detected target to all nearby neighbors. The goal is to achieve maximal coverage area and highly accurate tracking while optimizing battery usage of the sensors. In the second scenario, we investigate target tracking with optical sensors. Each optical sensor has a surveillance cone which can be fully rotated around the sensor center point. A sensor can only tell the angle of approaching target but not the exact distance. When seeing a target, the optical sensor alarms its neighbors and co-operate with them to do the tracking. Stanfield algorithm is used to combine one or more directional information from sensors to form positional information in terms of coordinates of best point estimate and ellipse error.

This is a joint-project between DSO National Labs and SMA CS Adaptive Computing Lab. After testing the scenarios with software simulation, we implement hardware prototypes with Crossbow Cricket, Micaz sensor mote kit and Canon communication cameras.