A Quality Aware Multiunit Double Auction Framework for IoT-Based Mobile Crowdsensing in Strategic Setting

Abstract

Crowdsourcing with the intelligent agents carrying smart devices is becoming increasingly popular in recent years. It has opened up meeting an extensive list of real-life applications such as measuring air pollution levels, road traffic information, etc. In literature, this is known as <i>mobile crowdsourcing</i> or <i>mobile crowdsensing</i>. In this paper, the discussed set-up consists of multiple task requesters (or task providers) and multiple IoT devices (as <i>task executors</i>), where each of the task providers has multiple homogeneous sensing tasks. Each task requester reports a bid and the number of homogeneous sensing tasks to the platform. On the other side, multiple IoT devices report the ask (the charge for imparting its services) and the number of sensing tasks they can execute. The valuations of task requesters and IoT devices are <i>private</i> information, and both might act <i>strategically</i>. One assumption that is made in this paper is that the bids and asks of the agents (<i>task providers</i> and <i>IoT devices</i>) follow <i>decreasing marginal returns</i> criteria. Given the above-discussed scenario, the objectives are: (1) to determine a set of quality IoT devices for each of the tasks held by the task requesters, and (2) to select a subset of quality IoT devices from among the available quality IoT devices for each of the sensing tasks. In this paper, a truthful mechanism is proposed for allocating the IoT devices to the sensing tasks carried by task requesters that also keep into account the quality of IoT devices. Through theoretical analysis, it is shown that the mechanism is <i>truthful</i>, <i>budget balanced</i>, <i>individually rational</i>, <i>computationally efficient</i>, <i>correct</i>, and <i>prior-free</i>. Further, probabilistic analysis is carried out to estimate the average number of tasks that get executed for any task requester. The simulations are carried out to measure the performance of the proposed mechanism against the benchmark mechanisms on the ground of <i>truthfulness</i>, <i>budget balance</i>, <i>satisfaction level</i>, <i>average incentive</i>, and <i>computational efficiency</i>.