Raja Jurdak,

University of California at Irvine –

Abstract:

Wireless sensor networks are emerging as a new class of ad hoc networks that differ considerably from conventional wired or wireless networks. The distinguishing characteristic of sensor networks is the lack of previously deployed infrastructure, which poses the design challenge of distributing network management among the nodes. Nodes in ad hoc sensor networks also have limited resources, so the distributed algorithms and protocols that manage network behavior must also enable fine-grained optimization at the resource-limited nodes. In addition, sensor networks have a wide range of potential applications, each of which requires custom performance guarantees, suggesting the need for a flexible methodology to optimize network behavior.

In this talk, I propose a flexible cross-layer framework for optimizing the performance of sensor networks that enables each node to greedily and locally optimize its behavior in order to achieve a global performance goal. The framework provides a customizable definition of node states and it enables the network designer to select the network mechanisms at any layer of the communication stack that require optimization. The framework\\\\\\\’s applicability and optimization potential are validated through two diverse case studies: (1) minimization of energy consumption in a sensor network test-bed using radio frequency signals; (2) maximization of battery lifetime and minimization of power consumption in an acoustic underwater sensor network. Each case study adopts a customized version of the framework by tailoring state definitions and by defining mechanisms that require optimization in accordance with the case study\\\\\\\’s performance requirements. In both case studies, the framework yields significant improvement in network performance.

 

Date: 2005-Dec-19     Time: 14:00:00     Room: 336

For more information:

  • 213100228