A multi-objective task allocation scheme with privacy-preserving and regional heat in mobile crowdsensing

Abstract

In Mobile Crowdsensing, platforms typically require all users to upload their location information during the user recruitment phase, then select a subset of users to perform tasks based on location and reputation. However, this approach results in users who upload their location information but do not participate in tasks essentially providing their location data without compensation, posing a risk of location data leakage. If users repeatedly upload location information without receiving compensation for tasks, they may lose confidence in the platform and consequently leave it. Therefore, this paper proposes a multi-objective task allocation scheme based on differential privacy and regional heat, named MTADPRH. During the user recruitment phase, the MTADPRH scheme uses the Optimized Unary Encoding (OUE) mechanism to statistically analyze the distribution of all users, providing privacy protection that meets local differential privacy. In the location upload phase, the scheme adds planar Laplace noise to the location coordinates of participating users to achieve geo-indistinguishability. During the task allocation phase, MTADPRH employs the multi-objective evolutionary algorithm C3M to find Pareto optimal solutions, aiming to maximize the reward per unit distance for users and the revenue for the platform. The experimental results show that, with privacy protect, the MTADPRH scheme achieves the best results in terms of platform revenue, task completion rate, and per-unit distance compensation for users, and it provides a superior Pareto solution.