PRIME: A Privacy-Enhanced Framework for Efficient Unknown Worker Recruitment in Heterogeneous Mobile Crowdsensing

Abstract

With the rapid advancement of sensor technologies and ubiquitous mobile intelligence, Mobile Crowdsensing (MCS) has become an effective paradigm for large-scale sensing tasks. It overcomes the limitations of traditional fixed sensing infrastructures while enabling low-cost and flexible data collection. However, practical MCS deployment still faces significant challenges. Conventional privacy-preserving approaches often suffer from high computational overhead or degraded data utility. Most existing recruitment methods assume prior knowledge of worker reliability, rendering them ineffective for unknown workers. Moreover, static task allocation strategies fail to adapt to the heterogeneous and characteristics of real-world sensing tasks. To address these limitations, we propose PRIME (Privacy-preserving Recruitment and Incentive-driven Multi-task allocation with freshness awareness), a privacy-enhanced framework for efficient unknown worker recruitment in heterogeneous Mobile Crowdsensing. The framework incorporates a lightweight privacy-preserving mechanism based on encoding and masking techniques to ensure both data integrity and secure transmission. In static scenarios, the recruitment process is modeled as a Combinatorial Multi-Armed Bandit (CMAB) problem with a bidirectional arm structure that jointly considers task requirements and worker preferences. A collateral function is introduced to mitigate the risk of malicious data submission. In dynamic scenarios, an optimal task allocation strategy is designed to account for system constraints and heterogeneous worker behaviors, thereby maximizing overall system utility. Extensive experiments on both synthetic and real-world datasets demonstrate that PRIME outperforms state-of-the-art baselines in privacy protection, recruitment effectiveness, and overall system performance.