Data-Driven Capacity Planning for Vehicular Fog Computing

The strict latency constraints of emerging vehicular applications make it unfeasible to forward sensing data from vehicles the cloud for processing. To shorten network latency, fog computing (VFC) moves computation edge Internet, with extension support mobility distributed entities [a.k.a nodes (FNs)]. In other words, VFC proposes complement stationary FNs co-located cellular base stations mobile ones carried by moving (e.g., buses). Previous works on mainly focus optimizing assignments tasks among available FNs. However, capacity planning, which decides where and how much resources deploy, remains an open challenging issue. complexity this problem results spatiotemporal dynamics traffic, varying resource demand generated applications, solve above challenges, we propose a data-driven planning framework that optimizes deployment minimize installation operational costs under quality-of-service constraints, taking into account variation in both supply. Using real-world traffic application profiles, analyze cost efficiency potential long term. We also evaluate impacts patterns plans savings. find high density significant hourly would lead dense create more savings