Reproduce “Dynamic Hypergame for Task Assignment in Multi-platform Mobile Crowdsensing Under Incomplete Information” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Sumedh J. Dongare, Christo Kurisummoottil Thomas, Andrea Ortiz, Walid Saad, Anja Klein

Researchers from Technical University of Darmstadt and Virginia Tech developed a dynamic hypergame formulation and a decentralized learning framework, PACMAB, for task assignment in competitive multi-platform mobile crowdsensing environments with incomplete information. PACMAB achieves 99.1% of optimal social welfare and 99.8% task completion, demonstrating effective adaptation and resource allocation despite agents lacking full knowledge of preferences or costs.

Key insights

  • Hypergame theory enables agents to effectively model and update perceptions about competitors' strategies through repeated interactions, addressing incomplete information in competitive settings.
  • Integrating perception-aware action pruning into a UCB bandit algorithm helps manage the large action space for MCSPs, allowing efficient learning in complex competitive environments.
  • A fully decentralized two-sided learning framework allows agents to make independent decisions, enhancing scalability and robustness in multi-platform mobile crowdsensing systems.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.