DTD: An Intelligent Data and Bid Dual Truth Discovery Scheme for MCS in IIoT
Yunchuan Kang, Anfeng Liu, Naixue Xiong, Shaobo Zhang, Tian Wang, Mianxiong Dong
Abstract
Mobile crowdsensing (MCS) is a crucial component in the Industrial Internet of Things (IIoT), mainly due to its role in collecting data and enhancing applications. Nonetheless, it faces challenges in maintaining data quality and cost efficiency. Low-quality workers and their deceptive data bids undermine the trustworthiness of MCS data collection. Despite this, prior studies have not sufficiently scrutinized the validity of data and bids. These issues could render MCS services ineffective and unaddressed, hindering IIoT development. In response, we propose an Intelligent Data and Bid dual truth discovery (DTD) scheme. Initially, the scheme applies a detection algorithm to identify the features of ground truth data sensed by unmanned aerial vehicles. The approach uses features to evaluate the data trust from unknown workers and filter out low-quality workers. Subsequently, the scheme evaluates the bid trust from reliable workers by calculating their bid confidence intervals. Upon completing this assessment, the scheme identifies high-quality workers. This process hinges on a contribution value incorporating both data and bid trust. Ultimately, the scheme assigns these high-quality workers to sense the subsequent tasks. This approach significantly improves the data quality and reduces costs for MCS in IIoT. The experimental results demonstrated that the DTD scheme outperforms the existing main schemes in terms of sensitivity (improvement 44%), specificity (improvement 5%), accuracy (improvement 18%), and F1-score (improvement 47%). It also reduced data bias by 24 percentage points and reduced costs by 38 percentage points.