V2X多节点协同分布式卸载策略

为了应对车联网中计算资源密集、可分离型任务的卸载环境动态变化和不同协同节点通信、计算资源存在差异的问题,提出了一种在V2X下多协同节点串行卸载、并行计算的分布式卸载策略。该策略利用车辆可预测的行驶轨迹,对任务进行不等拆分,分布式计算于本地、MEC及协同车辆,建立系统时延最小化的优化问题。为求解该优化问题,设计了博弈论的卸载机制,以实现协同节点串行卸载的执行顺序;鉴于车联网的动态时变特性,利用序列二次规划算法,给出了最优的任务不等拆分。仿真结果表明,所提策略能够有效减少计算任务系统时延,且当多协同节点分布式卸载服务时,所提策略在不同的参数条件下仍然能够保持稳定的系统性能。

Multi-node cooperative distributed offloading strategy in V2X scenario

In order to cope with the dynamic changes of the offloading environment for computing resource-intensive and separable tasks in Internet of vehicle and deal with the problem that different collaborative nodes had different communication and computing resources, a distributed offloading strategy that multiple collaborative nodes had serial offloading mode and parallel computing mode in vehicle to everything(V2 X) scenario was proposed. Utilizing the predictable motion trajectories of vehicle, the tasks were split into unequal parts, finally each part was computed on itself, mobile edge server, and vehicles in parallel. Then an optimization problem of the system time delay minimization was established. To solve the optimization problem, an offloading scheme based on the game theory was designed to determine the serial offloading execution order of the cooperative nodes. Considering the dynamic characteristics of Internet of vehicles, a sequential quadratic programming(SQP) algorithm was adopted to optimally split tasks. Finally, the simulation results show that the proposed strategy can effectively reduce system delay, and when multiple cooperative nodes offload in parallel, the proposed strategy can still maintain the stable system performance under the different parameter conditions.