A trust and bundling-based task allocation scheme to enhance completion rate and data quality for mobile crowdsensing

Abstract

In Mobile CrowdSensing (MCS), task bundling has shown promise in improving task completion rate by pairing unpopular tasks with popular ones. However, existing methods often assume truthful data from workers, an assumption misaligned with real-world MCS scenarios. Workers tend to submit low-quality or false data to maximize their rewards, particularly given the Information Elicitation Without Verification (IEWV) problem, which hinders the detection of dishonest behavior. To address this, we propose a Trust and Bundling-based Task Allocation (TBTA) scheme to enhance task completion rates and data quality at a low cost. The TBTA scheme includes three main strategies: (1) a trusted worker identification algorithm that evaluates workers' trust degrees by considering the IEWV challenge, allowing for the selection of reliable workers and thus ensuring higher data quality; (2) a task bundling method using the Non-dominated Sorting Genetic Algorithm II to bundle unpopular tasks with popular ones strategically, maximizing platform utility and completion rates; and (3) an optimal allocation algorithm that assigns trusted workers to tasks best suited to their capabilities, thus improving data reliability and minimizing costs. Experimental results demonstrate that compared to the state-of-the-art methods, the TBTA scheme achieves a 15.54 % improvement in task completion rate, and a 1.83 % reduction in worker travel distance.