Santashil PalChaudhuri, Jean-Yves Le Boudec, and Milan Vojnović
The random trip model was recently proposed as a generic mobility model that contains many particular mobility models, including the widely-known random waypoint and random walks, and accommodates more realistic scenarios. The probability distribution of the movement of a mobile in all these models typically varies with time and converges to a “steady state" distribution (viz. stationary distribution), whenever the last exists. Protocol performance during this transient phase and in steady-state may differ significantly. This justifies the interest in perfect sampling of the initial node mobility state, so that the simulation of the node mobility is perfect, i.e. it is in steady state throughout a simulation. In this work, we describe implementation of the perfect sampling for some random trip models. Our tool produces a perfect sample of the node mobility state, which is then used as input to the widelyused ns-2 network simulator. We further show some simulation results for a particular random trip mobility model, based on a real-world road map. The performance metrics that we consider include various node communication properties and their evolution with time. The results demonstrate difference between transient and steady-state phases and that the transient phase can be long lasting (in the order of a typical simulation duration), if the initial state is drawn from a non steady-state distribution. The results give strong arguments in favor to running perfect simulations. Our perfect sampling tool is available to public at: http://www.rice.edu/~santa/research/mobility
|Publisher||Institute of Electrical and Electronics Engineers, Inc.|
© 2004 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.