Harnessing Context for Budget-Limited Crowdsensing With Massive Uncertain Workers

Crowdsensing is an emerging paradigm of ubiquitous sensing, through which a crowd workers are recruited to perform sensing tasks collaboratively. Although it has stimulated many applications, open fundamental problem how select among massive number given task under limited budget. Nevertheless, due the proliferation smart devices equipped with various sensors, very difficult profile in terms ability. uncertainties can be addressed by conventional Combinatorial Multi-Armed Bandit (CMAB) framework trade-off between exploration and exploitation, we do not have sufficient allowance directly explore exploit Furthermore, since sensor usually quite resources, may bounded capabilities only few times, further restricts our opportunities learn uncertainty. To address above issues, propose Context-Aware Worker Selection (CAWS) algorithm this paper. By leveraging correlation context information their abilities, CAWS aims at maximizing expected cumulative revenue efficiently both budget constraint capacity constraints respected, even when uncertain massive. The efficacy verified rigorous theoretical analysis extensive experiments.