Abstract:
In this era, the magnitude of data shared is enormous and raised the bar for the quality of service and maintenance it requires. This paved the road for the integration of Fog Computing, which is an extension of the Cloud. Fog Computing’s main advantage is the increased quantity in which it can be deployed while in close vicinity of the end-users, thus enhancing their Quality of Experience (QoE). The connected vehicles domain is one of many domains that can benefit from Fog Computing. Moreover, caching has been an area of study for many years by researchers that aim to increase cache hit rate and decrease request delays affecting Connected Vehicles networks. Many studies implemented Machine Learning models to enhance cache hit rate and request delays. In this thesis, we implemented cooperation between a Deep Reinforcement Learning (DRL) model and Federated Learning to improve caching in Connected Vehicles connected to fog nodes. Furthermore, the results showed the proposed model's effectiveness compared to traditional algorithms.
