SEAL: A Strategy-Proof and Privacy-Preserving UAV Computation Offloading Framework
Yuntao Wang,Zhou Su,Tom H. Luan,Jiliang Li,Qichao Xu,Ruidong Li
DOI: https://doi.org/10.48550/arXiv.2305.08691
2023-05-15
Abstract:Due to the limited battery and computing resource, offloading unmanned aerial vehicles (UAVs)' computation tasks to ground infrastructure, e.g., vehicles, is a fundamental framework. Under such an open and untrusted environment, vehicles are reluctant to share their computing resource unless provisioning strong incentives, privacy protection, and fairness guarantee. Precisely, without strategy-proofness guarantee, the strategic vehicles can overclaim participation costs so as to conduct market manipulation. Without the fairness provision, vehicles can deliberately abort the assigned tasks without any punishments, and UAVs can refuse to pay by the end, causing an exchange dilemma. Lastly, the strategy-proofness and fairness provision typically require transparent payment/task results exchange under public audit, which may disclose sensitive information of vehicles and make the privacy preservation a foremost issue. To achieve the three design goals, we propose SEAL, an integrated framework to address strategy-proof, fair, and privacy-preserving UAV computation offloading. SEAL deploys a strategy-proof reverse combinatorial auction mechanism to optimize UAVs' task offloading under practical constraints while ensuring economic-robustness and polynomial-time efficiency. Based on smart contracts and hashchain micropayment, SEAL implements a fair on-chain exchange protocol to realize the atomic completion of batch payments and computing results in multi-round auctions. In addition, a privacy-preserving off-chain auction protocol is devised with the assistance of the trusted processor to efficiently protect vehicles' bid privacy. Using rigorous theoretical analysis and extensive simulations, we validate that SEAL can effectively prevent vehicles from manipulating, ensure privacy protection and fairness, improve the offloading efficiency.
Computer Science and Game Theory