SPREAD, a crowd sensing incentive mechanism to acquire better representative samples

Abstract

Crowd sensing is an approach to collect many samples of a phenomena of interest by distributing the sampling across a large number of individuals. While any one individual may not provide sufficient samples, aggregating samples across many individuals may provide high-quality and high-coverage measurements of a phenomena. In this work, we propose an incentive assignment mechanism for crowd sensing variable phenomena (e.g., temperature) that balances the goal of maximizing coverage of the area of interest, while at the same time staying within a budget constraint. This algorithm not only takes into account the area covered by the participants' sensors, but also the spread of these sensors through a target area. This characteristic enables more representative sampling than existing methods, assuming the same budget. Compared to existing methods, this algorithm improves the spread of the set of acquired samples by more than 56 % percent, without sacrificing the number of samples purchased from human sensors.