Offloading Floating Car Data through V2V Communication

During a century of existence, vehicles transformed from mechanical master pieces into consumer products with more and more computing power and sensing capabilities. Current vehicles include tens or even hundreds of Electronic Control Units (ECUs) that gather important amounts of information from the embedded sensors, and use this information to control different parts of the vehicles or to provide relevant messages to the driver. Until recently, all this data was only partially and temporarily stored inside the vehicles. However, with the advent of traffic information system or fleet management applications, the trend is to recover this floating car data (FCD), through a cellular network, at a central server where the FCD is treated and the results are potentially disseminated back to the vehicles. While the current penetration ratio of vehicles equipped with a cellular interface remains low, the number of FCD-enabled vehicles is expected to grow rapidly, triggered especially by a smartphone-vehicle integration. This inflation of applications and equipped vehicles risks to result in a non-negligible load on the cellular uplink access. Considering that the mobile demand has already reached the capacity limits of 3G networks, and that even the upcoming LTE infrastructure is deemed unable to cover such a gap, offloading FCD through direct vehicle-to-vehicle (V2V) communication could only benefit the cellular network operation. In this scenario, only a subset of the vehicles would use the cellular uplink, while also being in charge of data aggregation and, potentially, fusion (e.g. if the application requiring the data only needs an average value for a certain parameter). The FCD offloading problem therefore maps to identifying, in an efficient, distributed way, the vehicles in charge of performing the data fusion and upload after gathering the FCD from their neighbors. Clearly, we wish such a set to be as small as possible, since the fewer the vehicles performing the local fusion and upload, the lower the uplink traffic load on the cellular network. At the same time, however, we do not want the offload process to reduce the quality of the overall FCD information. In other words, the objective becomes that of collecting FCD from the whole network using the least amount of vehicles, so as to minimize the number of uploads and maximize the local FCD fusion. The FCD offloading problem can thus be formulated as a classical graph-theory Minimum Dominating Set (MDS) problem, whose solution yields the minimal set of vehicles …