Abstract
Vehicular crowdsensing allows the rapid, predictable movement of vehicles, as well as their wide variety of sensors, to gather sensing data in crowdsensing applications. Recruitment algorithms are used to select a subset of participants in an area that will provide the most complete coverage. In this paper, we explore two variations of the vehicular recruitment problem. In the first problem, which we refer to as the priority based vehicle recruitment problem, we consider coverage areas in which subsets must be covered. In the multisensor variation, we consider coverage areas which require different types of sensors, in which participating vehicles have one or more sensor types onboard. For each, we implement a mixed integer programming model which returns optimal solutions, as well as a heuristic for obtaining approximate solutions. In the unbudgeted priority vehicular recruitment performance evaluation, our heuristic on average obtains only 0.05% lower utility at 1.78% higher recruitment cost. In the budgeted runs, our heuristic obtains on average only 0.02% lower utility at 0.59% higher recruitment costs. In the unbudgeted multisensor vehicular recruitment performance evaluation, our heuristic obtains only 0.04% lower utility at 1.10% higher recruitment cost, and in the budgeted runs we obtain 11.33% lower utility at 0.27% higher recruitment cost.