User willingness aware task allocation for cloud–edge–terminal collaborative crowdsensing system

Cloud–Edge–Terminal Collaborative Crowdsensing (CETCS) has emerged as a novel research paradigm in the field of Mobile Crowdsensing (MCS). By leveraging edge servers for task computation, CETCS effectively mitigates communication delays and request congestion caused by the increasing scale of sensing tasks and growing data complexity. However, in real-world deployments, edge servers are characterized by resource and service heterogeneity. The Heterogeneous Edge Servers based Task Allocation (HESTA) problem has been formally formulated and proven to be NP-hard. Previous studies have largely overlooked two critical aspects: users’ willingness to execute tasks and the complexity involved in task offloading decisions. To address these limitations, we propose a unified framework that integrates Willingness-Aware Repair with a Probability Genetic Algorithm and Proximal Policy Optimization with the Dynamically Masked Action Space to jointly optimize task allocation, offloading, and computation during the task execution process. Our work differs from previous works in the following aspects: (1) We develop a comprehensive optimization framework that explicitly incorporates user willingness into the task allocation process to maximize overall platform utility; (2) We systematically categorize potential scenarios arising during task offloading and design corresponding utility functions to guide decision-making; (3) We propose a novel task offloading and computation selection algorithm aimed at maximizing the average remaining time of all tasks, thereby enhancing system responsiveness and efficiency. The extensive simulations are conducted on both synthetic and real-world datasets to demonstrate the effectiveness and superiority of the proposed approach.