Influence before Hiring: A Two-tired Incentive Compatible Mechanism for IoT-based Crowdsourcing in Strategic Setting

In crowdsourcing, a group of common people is asked to execute the tasks and in return will receive some incentives. In this article, one of the crowdsourcing scenarios with multiple heterogeneous tasks and multiple IoT devices (as task executors) is studied as a two-tiered process. In the first tier of the proposed model, it is assumed that a substantial number of IoT devices are not aware of the hiring process and are made aware by utilizing their social connections. Each of the IoT devices reports a cost (private value) that it will charge in return for its services. The participating IoT devices are rational and strategic. The goal of the first tier is to select the subset of IoT devices as initial notifiers so as to maximize the number of IoT devices notified with the constraint that the total payment made to the notifiers is within the budget. For this purpose, an incentive compatible mechanism is proposed. In the second tier, a set of quality IoT devices is determined by utilizing the idea of single-peaked preferences. The next objective of the second tier is to hire quality IoT devices for the floated tasks. For this purpose, each quality IoT device reports private valuation along with its favorite bundle of tasks. In the second tier, it is assumed that the valuation of the IoT devices satisfies gross substitute criteria and is private. For the second tier, the truthful mechanisms are designed independently for determining the quality IoT devices and for hiring them and deciding their payment respectively. Theoretical analysis is carried out for the two tiers independently and is shown that the proposed mechanisms are computationally efficient, truthful, correct, budget feasible, and individually rational. Further, the probabilistic analysis is carried out to estimate the expected number of IoT devices got notified about the task execution process.