Abstract
DiSenS: Scalable Distributed Sensor Network Simulation
by: Ye Wen, Rich Wolski, and Gregory Moore
Abstract:
Simulation can be a very useful tool for developing, evaluating and analyzing sensor network applications, especially when deploying a large scale sensor network remains an expensive and labor intensive endeavor. However, the difficulty of achieving both fidelity and scalability has limited its use as a design and analysis tool. In this paper, we describe DiSenS (DIstributed SENsor network Simulation) -- a complete scalable and extensible distributed simulation system for sensor networks. DiSenS provides a cycle-accurate device emulator that is extendable by various fidelity-enhancing models (radio, power, etc.) for tunable simulation accuracy. A key distinguishing feature of DiSenS is that it is implemented for distributed-memory parallel cluster systems. To achieve efficiency in this computational setting we have designed a simple and efficient distributed synchronization protocol and coupled it with a sophisticated node partitioning algorithm (not of our design) to achieve new scalability and performance levels.
Keywords:
sensor networks, distributed simulation
Date:
December 2005
Document: 2005-30
