An Offloading Scheme for Diverse Inter-Dependent Tasks in Vehicle Edge Computing

Vehicle Edge Computing (VEC) is a promising paradigm for efficiently processing massive amounts of data from sensors and stable responses to diverse applications from vehicles. In the usual scenario, the data needs to be transmitted between different nodes and the applications can be represented as inter-dependent tasks. Moreover, the subtasks in each task have diverse interdependencies and each task has a different deadline. Therefore, in order to make the task offloading decision, two questions must be answered: when and where each subtask should be offloaded according to its dependencies. This paper formulates the execution sequence of subtasks by modeling the relationships between subtasks with Directed Acyclic Graph (DAG). Meanwhile, to satisfy the deadline of applications, the computing resources of idle vehicles are sufficiently utilized, and the computing resources on the VEC server are efficiently allocated to tasks. To optimize the overall performance of the system, this paper constructs an optimization problem of minimizing average task completion delay, and proposes the Optimal Offloading Strategy and Resource Allocation for Multi-Dependent Tasks (OSRA-MDT) scheme to solve this optimization problem. Simulation results show that our proposed OSRA-MDT scheme has achieved up to $24 \%$ improvement in reducing the average task completion latency of the system compared to the baseline scheme.