Two-Layer Blockchain-Based Secure Task Offloading of Vehicle Edge Computing

Abstract

Blockchain provides a secure and trusted environment for task offloading in Vehicle Edge Computing (VEC). Nevertheless, single-layer blockchain architecture exhibits low transaction processing efficiency and insufficient scalability in VEC environments. Accordingly, we present a two-layer blockchain-based VEC (TBVEC) task offloading framework, which extends consensus nodes to parked vehicles (PVs) to improve the scalability of edge computing. Specifically, a Hotstuff–BLSPBFT two-layer blockchain is constructed by employing Hotstuff for the lower-layer blockchain and BLSPBFT for the upper-layer blockchain, thereby improving transaction efficiency of the blockchain system. Then, sharding is performed on consensus nodes of the lower-layer blockchain, and an improved K-Means algorithm is adopted for grouping optimization. Furthermore, utility functions of user vehicles and Road Side Units (RSUs) are formulated based on a two-layer blockchain. A game-based Deep Reinforcement Learning (DRL) model, equipped with two Soft Actor-Critic (SAC) agents, is developed to optimize user and RSU utility functions. The model determines the optimal task offloading strategies and transaction pricing policies, while jointly optimizing the number of consensus node groups and the quantity of nodes within each group to maximize the utility function. Experimental results validate that the proposed scheme not only ensures the security of task offloading but also significantly enhances overall system efficiency.