In floating car data services (FCD services), sensor data which is generated in vehicles is transmitted to a service center. Combining data from different vehicles, the service center draws conclusions depending on the kind of FCD service and sends these conclusions back to the service subscribers. Typical services are detection of traffic jams and information on traffic flow. FCD services are expected to become a mass market in the future, resulting in a very high number of cars taking part. If no restricting means are implemented, many cars will send data to the service center if an event occurs, causing a great deal of redundant data. The algorithms described in this paper solve this problem by delaying transmissions to the service center for a randomly chosen time. The service center controls the sending behavior by prescribing the probability distribution function and its parameters. The number of uplink transmissions needed for recognition of the event by the service center, statistical security conditions, and the time period which the service center is given to recognize the event are taken into account. Furthermore, the fact that the vehicles typically do not detect an event at the same time is considered. So, with the proposed algorithms the number of transmissions from vehicles to the service center can be reduced dramatically. Additionally, since the vehicles participating in the FCD service do not need to be addressed individually, the computing load of the service center can be decreased significantly. The performance of the proposed algorithms is shown through statistical analysis.
Abstract