Fog computing reduces network latency by moving computational resources close to where the data is generated. Vehicular fog computing (VFC) is an emerging computing paradigm where fog nodes deployed on moving vehicles (i.e., vehicular fog nodes (VFNs)) complement stationary fog nodes (e.g., the ones co-located with cellular base stations) to satisfy the spatiotemporally varying demand for computing resources in a cost-efficient manner. On-demand VFC (ODVFC) supports dynamic routing of VFNs, with the aim of fulfilling the spatiotemporally varying demand for computational resources in a cost-efficient manner. Different from previous works on capacity planning and vehicle routing that utilize compute-intensive optimization methods such as integer linear programming (ILP), this paper explores the feasibility of applying reinforcement learning to dynamic capacity planning in a time-efficient manner. Specifically, we propose to apply multi-agent reinforcement learning (MARL) with actor-critic methods to train the VFN routing policies. This approach allows distributed VFNs to cooperatively maximize the techno-economic performance of ODVFC. For evaluation, we built an open-source VFC simulation platform that integrates vehicular traffic simulation with 5G NR V2X and MARL environment. Compared with decentralized learning (i.e., each VFN independently learns its routing policy), centralized learning (i.e., using a global agent for VFN routing), and ILP methods, our proposal proves to achieve 8.3% higher revenue and 13.2% higher number of served tasks than decentralized training; and it has 40.6% and 83% lower execution time than centralized learning and ILP, respectively, with only 14% lower revenue than both. It is also scalable to real-life scenarios with a great number of users and VFNs.

Emerging compute-intensive and latency-sensitive vehicular applications are expected to be deployed at the edge instead of the cloud to shorten the network latency. Mobile fog nodes carried by moving vehicles have been proposed to complement the stationary fog nodes co-located with base stations to handle the spatio-temporal variations of the demand in a cost-efficient way. Existing works on capacity planning for such vehicular fog computing (VFC) scenarios assume that the vehicular traffic follows certain spatio-temporal patterns, which may change in different seasons, and create capacity plans accordingly. In other words, they consider long-term capacity planning, leaving the adaptation to temporary changes or unexpected variations out of scope. In this work, we propose an integer linear programming (ILP) based framework to optimize the routing strategy of vehicular fog nodes (VFNs) in order to maximize the profit received by the service provider, taking into account the quality of service (QoS) received by the users and service level agreement (SLA) of various applications. To adapt to the temporal variations in demand, we predict the traffic flow and resource consumption from the users with feedback from service evaluation. To reduce the computational time and enable parallel processing, we create the capacity plan in two steps, namely global planning and regional planning. Through simulations, we show that the proposed solution achieves an 85% higher profit and a 20% higher service rate compared to the strategy where the VFNs randomly travel and serve the surrounding users without demand prediction. It achieves similar network latency compared to the strategy using only stationary fog nodes, but with a higher cost-efficiency. We also evaluate the impacts of number of VFNs, cost parameters, and regional size on the capacity plan. We find that a high number of VFNs, a small regional size, a high penalty cost, and low traveling and rental costs will lead to a high service rate; while a large regional size and low traveling, rental, and penalty costs will result in a high profit.

Edge/fog computing is a key enabling technology in 5G and beyond for fulfilling the tight latency requirements of emerging vehicle applications, such as cooperative and autonomous driving. Vehicular Fog Computing (VFC) is a cost-efficient deployment option that complements stationary fog nodes with mobile ones carried by moving vehicles. To plan the deployment and manage the VFC resources in the real world, it is essential to take into account the spatio-temporal variations in both demand and supply of fog computing capacity and the trade-offs between achievable Quality-of-Services and potential deployment and operating costs. Concerning the complexity and the economic load of real-world measurements, simulation becomes a better option at the early research phase to validate capacity and resource management solutions in various urban environments. The existing simulation platforms cannot provide a realistic techno-economic investigation to analyze the implications of VFC deployment options, due to the simplified network models in use, the lack of support for fog node mobility, and limited testing scenarios. In this paper, we propose an open-source simulator VFogSim that allows real-world data as input for simulating the supply and demand of VFC in urban areas. It follows a modular design to evaluate the performance and cost-efficiency of different deployment scenarios under various vehicular traffic models, and the effectiveness of the diverse network and computation schedulers and prioritization mechanisms under user-defined scenarios. Compared with the existing edge/fog computing simulators, such as IFogSim, IoTSim, and EdgeCloudSim, to the best of our knowledge, our platform is the first one that supports the mobility of fog nodes and provides realistic modeling of V2X in 5G and beyond networks in the urban environment. Furthermore, we validate the accuracy of the platform using a real-world 5G measurement and demonstrate the functionality of the platform taking VFC capacity planning as an example.

The strict latency constraints of emerging vehicular applications make it unfeasible to forward sensing data from vehicles to the cloud for processing. To shorten network latency, Vehicular fog computing (VFC) moves computation to the edge of the Internet, with the extension to support the mobility of distributed computing entities. In other words, VFC proposes to complement stationary fog nodes co-located with cellular base stations with mobile ones carried by moving vehicles. Previous works of VFC mainly focus on optimizing the assignments of computing tasks among available fog nodes. However, capacity planning, which decides where and how much capacity to deploy, remains an open and challenging issue. The complexity of this problem comes from the mobility of vehicles, the spatio-temporal dynamics of vehicular traffic, and the computing resource demand generated by varying vehicular applications. To solve the above challenges, we propose a data-driven capacity planning framework that optimizes the deployment of stationary and mobile fog nodes to minimize the installation and operational costs under the quality-of-service constraints, taking into account the spatio-temporal variation in computing demand. Through real-world experiments, we analyze the cost efficiency potential of VFC in long term and demonstrate that the performance loss of VFC is below $6\%$ compared to stationary deployment with equal network capacity. We also analyze the impacts of traffic patterns on the potential cost saving. The results show when the traffic density is higher, more operational costs will be saved in the long run due to more dense deployment of mobile fog nodes.