Computer Science Colloquia
Monday, July 11, 2011
Advisor: Jack Stankovic
Attending Faculty: Sang Son (Chair); Kamin Whitehouse, Alf Weaver, and Alfredo Garcia
Olsson Hall, Room 236D, 9:00 AM
Ph.D Seminar Presentation
Design and Implementation of an Automatic, Robust, and Efficient Breadcrumb Sensor Network
In public safety and homeland security, reliable and robust communication between the first responders and an incident commander is crucial to real-time monitoring and control applications. Commanders desire to keep track of first responders and monitor their healthy status to safeguard their lives. Moreover, commanders may send first responders various types of information, such as 3D building floormaps and retreat commands. In order to support all these safety-critical applications, a communication infrastructure offering reliable communication links is indispensable .
However, such a reliable indoor infrastructure is not available nowadays. Predeployed sensor systems may be impaired and destroyed during disasters, and one-hop trunked radios such as the P25 systems  may cause first responders to lose their connections, because indoor environments often contain substantial amounts of metal and other reflective materials that affect the propagation of radio frequency signals in nontrivial ways, causing severe
multi-path effects, dead-spots, noise and interference . Hence, a new dynamic infrastructure, called breadcrumb sensor networks, has been emerging in recent years, which allows each first responder to carry a small dispenser filled with sensor nodes and deploy them one-by-one in a manner that guarantees reliable communication and extends transmission range.
To date, the state of art on breadcrumb sensor networks is still in its infant stage. There is a lack of systematic system design and effort to make the deployment process automatic and efficient. Therefore, in this dissertation I provide the first sophisticated study of automatic, reliable, and efficient breadcrumb sensor networks. It includes three parts. First, a holistic reliability model is proposed for both the deployment process and post-deployment link measurement. System components like redundancy degree, link monitoring, height effect solver, and adaptive power control are designed carefully and evaluated through extensive experiments in real buildings. Second, I explore the coordination among multiple first responders to efficiently utilize limited system resources. A novel utility function based coordination algorithm is proposed for global optimization of the resource assignment problem. Finally, I measure the impact of body shadowing on system reliability and efficiency when multiple first responder exist, and propose an intentional forwarding approach to solve this important problem.