Trust-Aware Blockchain Reinforcement Learning for Secure Task Offloading in Edge-Assisted IoV

The Internet of Vehicles (IoV) is rapidly expanding, and vehicles continuously generate complex and interdependent computation tasks, which contain precedence constraints that complicate their real-time offloading. These tasks must be processed in real time and with strong security guarantees, which introduces significant challenges in both computation efficiency and trust management. To address these challenges, we investigate secure online offloading of large-scale dependent tasks in distributed edge-assisted IoV environments and develops a blockchain-assisted multi-level feedback queue framework that ensures transparent task registration, tamper-proof execution logs, and fairness among vehicular nodes. To optimize the offloading strategy under dynamic and uncertain system states, the problem is formulated as a multi-stage mixed-integer nonlinear program, and a trust-aware blockchain reinforcement learning algorithm (TaBRL) is developed. The blockchain consensus layer provides verifiable feedback to guide adaptive learning and strengthen decision reliability. Extensive experiments demonstrate that the proposed framework achieves superior task throughput, queue stability, and defense robustness, effectively ensuring secure, trustworthy, and efficient coordination in edge-assisted IoV environments.