Spatial-Temporal Value-of-Information Maximization for Mobile Crowdsensing in Wireless Sensor Networks

Abstract

Mobile crowdsensing is considered as a promising technology to exploit the computing and sensing capabilities of the decentralized wireless sensor nodes. Typically, the quality of information obtained from crowdsensing is largely affected by various factors, such as the diverse requirements of crowdsensing tasks, the varying quality of information across different crowd workers, and the dynamic changes of channels conditions and the sensing environment. In this paper, considering the dynamics’ of the crowd workers, we focus on a spatial-temporal crowdsensing model and aim to maximize the value of information at the point of interest, by optimizing the recruiting range and time duration for the crowd workers. In particular, the crowdsensing system includes a mobile access point (MAP) and a set of wireless sensor nodes. As the information requester, the MAP can broadcast its crowdsensing task and then estimate the value of information by collecting the responses from the sensing nodes. Each sensing node in the crowdsensing task will receive a payment from the MAP. We aim to maximize the utility of the information requester by optimizing the recruiting range and waiting time for the sensing nodes. We firstly define a set of value metrics to characterize the MAP’s value of information. The optimal recruiting range can be obtained in closed-form expressions. Furthermore, considering the aging effect, we propose a gradient-based method to maximize the spatial-temporal value of information. Specifically, we first determine the optimal recruiting time for the requester and then choose the optimal recruiting range within each time slot. Via simulation, we first compare the sum, max, and min values of information at the requester, and then verify the effectiveness of the gradient-based method to optimize the recruiting time and range to maximize the value of information.