Modeling and performance evaluation of Advanced Diffusion with Classified Data in vehicular sensor networks

In this paper, we propose a newly distributed protocol called Advanced Diffusion of Classified Data (ADCD) to manage information harvesting and distribution in vehicular sensor networks. ADCD aims at reducing the generated overhead, avoiding network congestions as well as long latency to deliver the harvested information. The concept of ADCD is based on the characterization of sensed information (i.e., based on its importance, location, and time of collection) and the diffusion of this information accordingly. Furthermore, ADCD uses an adaptive broadcasting strategy to avoid overwhelming users with messages in which they have no interest. Also, we propose in this paper a new probabilistic model for ADCD based on Markov chain. This one aims to optimally tune the parameters of ADCD, such as the optimal number of broadcaster nodes. The analytical and simulation results based on different metrics, such as the overhead, the delivery ratio, the probability of a complete transmission, and the minimal number of hops, are presented. These results illustrate that ADCD allows mitigating the information redundancy and its delivery with an adequate latency while making the reception of interesting data for the drivers (related to their location) more adapted. Moreover, the ADCD protocol reduces the overhead by 90% compared with the classical broadcast and an adapted version of MobEyes. The ADCD overhead is kept stable whatever the vehicular density. Copyright © 2011 John Wiley & Sons, Ltd.