2021/10/25 by Farzaneh Farhadi, Farhadi, Farzaneh, Maria Chli +3
Business, Management and Accounting · Computer Science · Decision Sciences · #Artificial Intelligence (cs.AI) #Auction Theory and Applications #Blockchain Technology Applications and Security #Computer Science and Game Theory (cs.GT) #FOS: Computer and information sciences #Mobile Crowdsensing and Crowdsourcing #Supply Chain and Inventory Management
paper · pdf · doi:10.48550/arxiv.2110.12846
openalex publication_date 2021/10/25 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
We consider an outsourcing problem where a software agent procures multiple\nservices from providers with uncertain reliabilities to complete a\ncomputational task before a strict deadline. The service consumer requires a\nprocurement strategy that achieves the optimal balance between success\nprobability and invocation cost. However, the service providers are\nself-interested and may misrepresent their private cost information if it\nbenefits them. For such settings, we design a novel procurement auction that\nprovides the consumer with the highest possible revenue, while giving\nsufficient incentives to providers to tell the truth about their costs. This\nauction creates a contingent plan for gradual service procurement that suggests\nrecruiting a new provider only when the success probability of the already\nhired providers drops below a time-dependent threshold. To make this auction\nincentive compatible, we propose a novel weighted threshold payment scheme\nwhich pays the minimum among all truthful mechanisms. Using the weighted\npayment scheme, we also design a low-complexity near-optimal auction that\nreduces the computational complexity of the optimal mechanism by 99% with only\nmarginal performance loss (less than 1%). We demonstrate the effectiveness and\nstrength of our proposed auctions through both game theoretical and numerical\nanalysis. The experiment results confirm that the proposed auctions exhibit 59%\nimprovement in performance over the current state-of-the-art, by increasing\nsuccess probability up to 79% and reducing invocation cost by up to 11%.\n