Most research on target monitoring using sensor networks requires the geographic locations of the sensors to be known. However, in practice, it is not always possible for every sensor node to have a built-in GPS-like capability. On the other hand, applying a localization procedure to all the sensor nodes can be expensive for large networks. Even if the location information of many sensors can be computed, it may never be used because targets usually appear very sparse.
This project, therefore, tackles the target monitoring problem in the context of low-cost sensor networks, where a sensor node does not need built-in capability to determine its location. Instead, hopcount information serves as the primary source of information to track the target. The challenges result from the unknown mobility and nature of the target and the coarseness of the hopcount data. The matter is more sophisticated if there are multiple moving targets and it is not known whether they move in groups or independently.
The research objective of the project is to design and implement a novel solution framework to address these challenges. This framework, based on a machine-learning approach, can be applied to a wide range of sensor networks which can have modest resource capacities or be deployed in non-conventional physical settings such as under the water or on wet ground. Building on the research, an education plan is proposed that seeks to transfer knowledge to younger generations, improve student enrollment and retention, and encourage active participation from students of under-represented groups.