Abstract
Fog computing (FC) and multi-access edge computing (MEC) are two promising technologies that have been emerged to solve problems related to the access to the cloud computing (CC), mainly high latency and high bandwidth consumption. These two paradigms consist in enabling the cloud closer to users at the edge of the network. The pool of vehicular resources provided by the vehicular cloud (VC) can be exploited to process and store end users data instead of accessing remote servers. The combination of these three concepts can considerably augment the edge resources. In this context, we propose a multi-access edge based vehicular fog computing architecture on the internet of vehicles where vehicles are the fog nodes. In this paper, we present a detailed description of our suggested architecture and its modules. Then, we focus on a particular module which is the gateways selection module. The role of this module is the election of suitable fog nodes (i.e. vehicles) to access the MEC servers and the conventional cloud in order to reduce communication costs (e.g. bandwidth use, delay). The proposed selection approach has two steps. The first step consists in selecting a set of candidate gateways based on fuzzy logic. The second step allows the optimization of the number of selected gateways. We formulate it as a multi-objective optimization problem, and we solve it using ant colony optimization. The obtained simulation results show the efficiency of our proposed approach in terms of the number of selected gateways and connected fog nodes. In both static and mobile scenarios, the number of selected gateways is reduced up to 82% and 92%, respectively, compared to the fuzzy step. The ratio of connected vehicles is more than 94% in the static scenario.