Optimizing Collaborative Crowdsensing: A Graph Theoretical Approach to Team Recruitment and Fair Incentive...

Area of Science:

• Computer Science
• Distributed Systems
• Game Theory

Background:

• Collaborative crowdsensing leverages large networks for tasks in intelligent computing, federated learning, and blockchain.
• Traditional crowdsensing focuses on individual user capabilities, whereas collaborative crowdsensing emphasizes teamwork.
• User recruitment and incentive fairness are key challenges in collaborative crowdsensing, particularly due to free-riding behavior.

Purpose of the Study:

• To develop a graph-based approach for optimal user recruitment in collaborative crowdsensing.
• To introduce a mechanism (MR-SVIM) that addresses unfair incentives caused by free-riding.
• To ensure equitable earnings distribution based on user contributions and collaborative abilities.

Main Methods:

• User interactions are modeled as a graph, with recruitment utilizing an enhanced Prim algorithm to find the maximum spanning tree.
• A Gaussian mixture model and historical reputation predict task quality and direct reputation.
• Improved PageRank and aggregation functions assess user influence (local and global) for indirect reputation calculation.

Main Results:

• The MR-SVIM mechanism effectively calculates comprehensive user reputation by combining direct and indirect assessments.
• A contribution feature function is formulated, integrating collaborative capabilities.
• An enhanced Shapley value method is applied for equitable distribution of earnings, validated by real-world data.

Conclusions:

• The proposed graph-based recruitment and MR-SVIM mechanism significantly improve fairness in collaborative crowdsensing.
• The study demonstrates the effectiveness of integrating user collaboration, reputation, and contribution assessment.
• This work provides a robust solution for incentive allocation in decentralized collaborative environments.