LazySync: A New Synchronization Scheme for Distributed Simulation of Sensor Networks

摘要

To meet the demands for high simulation fidelity and speed, parallel and distributed simulation techniques are widely used in building wireless sensor network simulators. However, accurate simulations of dynamic interactions of sensor network applications incur large synchronization overheads and severely limit the performance of existing distributed simulators. In this paper, we present LazySync, a novel conservative synchronization scheme that can significantly reduce such overheads by minimizing the number of clock synchronizations during simulations. We implement and evaluate this scheme in a cycle accurate distributed simulation framework that we developed based on Avrora, a popular parallel sensor network simulator. In our experiments, the scheme achieves a speedup of 4% to 53% in simulating single-hop sensor networks with 8 to 256 nodes and 4% to 118% in simulating multi-hop sensor networks with 16 to 256 nodes. The experiments also demonstrate that the speedups can be significantly larger as the scheme scales with both the number of packet transmissions and sensor network size.