Lyapunov-Guided Optimal Service Placement in Vehicular Edge Computing

Vehicular Edge Computing(VEC)brings the computational resources in close proximity to the service requestors and thus supports explosive com-puting demands from smart vehicles.However,the limited computing capability of VEC cannot simul-taneously respond to large amounts of offloading re-quests,thus restricting the performance of VEC sys-tem.Besides,a mass of traffic data can incur tremen-dous pressure on the front-haul links between vehicles and the edge server.To strengthen the performance of VEC,in this paper we propose to place services beforehand at the edge server,e.g.,by deploying the services/tasks-oriented data(e.g.,related libraries and databases)in advance at the network edge,instead of downloading them from the remote data center or of-floading them from vehicles during the runtime.In this paper,we formulate the service placement prob-lem in VEC to minimize the average response latency for all requested services along the slotted timeline.Specifically,the time slot spanned optimization prob-lem is converted into per-slot optimization problems based on the Lyapunov optimization.Then a greedy heuristic is introduced to the drift-plus-penalty-based algorithm for seeking the approximate solution.The simulation results reveal its advantages over others in terms of optimal values and our strategy can satisfy the long-term energy constraint.