Article Preview
TopSocial Welfare-Based Task Assignment In Mobile Crowdsensing
Mobile terminal devices have significantly improved with the rapid advancement of science and technology (Bradai et al., 2019; Guo et al., 2015). Mobile crowdsensing, a novel paradigm employing smart data collection devices, has become increasingly popular. Mobile terminals now integrate a wide range of embedded sensors such as magnetometers, IR, GPS, and gyroscopes. These sensors enable individuals to complete arduous tasks and address various social issues such as medical treatment (Pryss et al., 2015), transportation, environmental concerns, public safety, and commerce through almost ubiquitous wireless networks (Bellavista et al., 2015; Ganti et al., 2011; Guo et al., 2015; Kantarci et al., 2016; Lane et al., 2010). Mobile crowdsensing relies on sensors in individuals’ mobile devices rather than deploying static sensors, thereby reducing sensing costs and improving efficiency. The crowdsensing system consists of three components (J. Wang et al., 2018; X. Wang et al., 2018) as depicted in Figure 1.
Figure 1. Cloud-based mobile crowdsensing system
Service requesters post sensing tasks, and the platform selects specific users to perform them. At the same time, workers collect sensing data by completing the tasks. The cloud platform plays a vital role in task allocation, wherein the platform reasonably assigns tasks based on requirements such as the task’s type or location. In practice, the cloud platform identifies the appropriate worker to execute the task sent by the requester. There have been studies on task assignments. Zhang et al. (2023) proposed the group-oriented adjustable bid-based task assignment (GO-ABTA) algorithm to address the group-oriented bilateral preference-matching problem. First, the algorithm selects initial leaders and their collaborative groups in the social network using a group-oriented collaboration approach. Then, the algorithm completes the adjustable-bid task assignment process based on preference matching while considering budget constraints. Li et al. (2023) combined edge computing with the mobile crowdsensing system and proposed a two-stage task assignment-optimization approach with limited computational resources. This approach uses deep reinforcement learning to select the optimal edge server for task deployment and then uses a greedy adaptive stochastic algorithm to recruit sensing participants. These efforts, however, do not consider the characteristics of the parties involved in the real world, making task allocation and social welfare optimization inapplicable. Although there are sensing platforms, such as Campaigner and Medusa, these need to consider the characteristics of workers or cloud platforms to allocate tasks efficiently. Therefore, designing an approach that considers real task assignments in real-world scenarios is crucial. Regarding sensing modes, current task assignment methods utilize either participatory or opportunistic sensing modes (Capponi et al., 2019; Gong et al., 2018). These methods often focus on ideal scenarios and may have limitations. To make task allocation more applicable, it is crucial to consider the psychological and moral characteristics of both workers and cloud platforms.