Data Exchange Strategies for Aggregating Geographical Distribution of Demands for Location-Dependent Information Using Soft-State Sketches in VANETs

When we drive a car, we often want to know the situation of roads near our destination. Ishihara et. al. proposed a VANET based information sharing system named Real-Time Visual Car Navigation System, in which a driver can obtain Location-Dependent Information (LDI), such as pictures or videos of his/her Point of Interest (POI) by telling the automotive equipment, e.g. car navigation device, the POI. The simplest way to provide LDI to vehicles is flooding the LDI to all vehicles in the VANET, but unneeded LDI may be disseminated, and network resources may be wasted. We have proposed a data dissemination scheme based on Demand map for effectively disseminating LDI to multiple vehicles that need it. In this scheme, each vehicle has Demand map (Dmap), a data set representing the geographical distribution of the strength of demands for LDI. Each vehicle exchanges a subset of data constituting a Dmap (Dmap Information: DMI) with other vehicles. According to the information in the Dmap, each vehicle preferentially sends new LDI or forwarded LDI strongly demanded to the area containing many vehicles that demand the LDI. In this paper, we propose strategies for controlling the frequency of sending DMI and strategies for selecting DMI to be sent for reducing the communication traffic without loss of the accuracy of information in Dmaps. Simulation results show that one of the proposed strategies, LTZ strategy, achieves high accuracy of Dmaps with a small amount of communication traffic.