Recruiting Trustworthy Crowdtesters in AIoT: A Cheat-Proof and Budget-Feasible Mechanism

Artificial intelligence of things (AIoT) offers unprecedented connectivity and intelligence for smart city applications. Yet, it also raises significant challenges in security, privacy, and interoperability that require thorough testing to guarantee a seamless user experience across diverse devices and platforms. Crowdtesting emerges as a vital solution, providing a cost-effective and comprehensive evaluation mechanism for AIoT by leveraging a global pool of testers. This paper proposes a novel social-aware recommend-then-recruit crowdtesting framework to optimally recruit trustworthy crowdtesters in AIoT applications. Specifically, we first develop an intelligent crowdtester recommendation mechanism that integrates social effects with collaborative filtering, effectively matching tasks with suitable and reliable candidates. We then design a trust model to evaluate crowdtesters' trustworthiness, incorporating both social and feedback aspects. Furthermore, a cheat-proof and individually rational auction mechanism is devised to ensure high-quality crowdtesting outcomes under budget constraints. Extensive simulations validate the superiority of the proposed scheme in terms of task quality and social welfare, compared with conventional schemes.