MRPO: An Evolutionary Game-based Multiple Round Probabilistic Oracle for BIoV

Blockchain oracles enable reliable data interaction between blockchains and the external world. However, existing oracles are unsuitable for blockchain-based IoV scenarios due to the characteristics of vehicle mobility and data locality, which may result in unreliable data uploaded to the blockchain. To this end, we propose a probabilistic oracle protocol MRPO, which is realized through multi-round interactions. In MRPO, roadside units manage the blockchain, while vehicles reach consensus on the requested query through multiple rounds. To address misleading and blocking malicious attacks compromising MRPO's safety, we propose an incentive-based weighted voting mechanism and use evolutionary game theory to demonstrate its effectiveness. Additionally, we propose an asynchronous, verifiable random selection algorithm and a collection-based early termination mechanism, enabling MRPO to achieve seconds level finality for near-real-time or delay-tolerant BIoV oracle services. Experimental results show that MRPO withstands Sybil attacks with a Sybil replication factor up to 10. In the dynamic initial-attack setting, MRPO recovers from an initial malicious stake proportion of 90% after continuous vehicle turnover and honest inflow dilute the adversarial stake below 50%, whereas the evaluated baselines fail under the same setting. MRPO also remains resilient when the aggregate malicious stake is maintained at 40%.