Fog computing, an extension of the Cloud Computing paradigm where routers themselves may provide the virtualisation infrastructure, aims at achieving fluidity when distributing in-network functions, in addition to allowing fast and scalable processing, and exchange of information. In this paper we present a fog computing architecture based on a “content island” which interconnects sets of “things” to exchange and process data among themselves or with other content islands. We then present a use case that focuses on a smartphone-based forward collision warning application for a connected vehicle scenario. This application makes use of the optical sensor of smartphones to estimate the distance between the device itself and other vehicles in its field of view. The vehicle travelling directly ahead is identified relying on the information from the GPS, camera, and inter-island communication. Warnings are generated at both content islands, if the driver does not maintain a predefined safe distance towards the vehicle ahead. Experiments performed with the application show that with the developed method, we are able to estimate the distance between vehicles, and the inter-island communication has a very low overhead, resulting in improved performance. On comparing our proposed solution based on edge/fog computing with a cloud-based api, it was observed that our solution outperformed the cloud-based api, thus making us optimistic of the utility of the proposed architecture.
Minimizing Age-of-Information for Fog Computing-supported Vehicular Networks with Deep Q-learning