Nrl
Publications at NRL

Search by Title

Search:

Search by Author

Search:


PPT Presentation

Title

New trends in pervasive computing: vehicular sensor platforms

Abstract

In traditional pervasive computing applications, the individual interacts with a highly instrumented environment and “unconsciously” connects with the various sensors embedded in the environment to extract important information. In fact, the “constellation” of devices that the individual carries on himself spontaneously “computes” in a “pervasive” way several metrics that could be of interest to its master. An example is the traveler that in the airport hurries towards the gate and is informed of last minute announcements, or, the vehicle in an “intelligent” highway that picks up navigation safety information from the curb. A new trend has emerged recently, which basically turns the tables around. Individuals now carry a number of sensing devices (video cameras, acoustic recorders, etc). In the future they may carry even more, for example radiation sensors, etc. The same can be said for cars, which carry video cameras, GPS and also various other types of sensors. The model now is no longer the mobile user that unconsciously extracts information form the environment. Rather, it is the environment (specifically, a few well placed access point in the urban mesh network, say) that continuously probes passing pedestrians and cars, in order to extract useful information. More generally, the environment has been extended to include mobile sensor platforms, which now are becoming a valuable source of “pervasive” information. This new mobile sensor and information platform model poses very interesting challenges. In this talk we will focus on cars in the urban grid as distributed platforms, and will address the problem of accessing the massive information that has been collected and is available in the car sensor fabric. Two approaches will be discussed. The first approach is for each car to “diffuse” the information it collected to all the other cars in the neighborhood using the “Epidemic Dissemination” method. Another approach is to maintain a distributed index (say, implemented with Distributed Hash Tables) that allows to track the desired content as well as the location of the vehicle that stores it. We will report on our experience with Ad Torrent (a scheme for disseminating advertisements to cars). We will discuss the two schemes and comment on their relative merits.

Download

Information & Date

ISWPC '06, Phuket, Thailand, January. 2006

Authors

Mario Gerla