Caching in Fog Radio Access Networks: Modeling, Analysis and Optimization

Existing 5G wireless networks comprises of hyper-connected users, machines, devices, Augmented/Extended Reality (AR/XR) requiring enhanced data rates, improved energy efficiency, seamless coverage and Ultra-reliable low-latency communication (URLLC). To cope with the challenges associated with the deployment of 5G wireless networks, power-hungry data-demanding applications such as real time interactive hologram services, immersive media, and virtual augmented reality require adaptive on-the-fly resources for interconnected devices, machines and users in a self-organized manner by using network slicing, artificial intelligence, blockchain and fog computing. Fog Radio Access Network (FRAN) provides an efficient platform to achieve the goals of 6G by incorporating cloud computing, fog computing, edge caching and artificial intelligence. Cloud computing enables a centralized cloud server to compute and serve a user. Different from cloud computing, fog computing relies on several distributed servers with limited computation capabilities to serve the users. On the other hand, edge caching strategically exploits the low-cost storage elements at edge nodes to offload different network elements and reduces network congestion. The key goal of this thesis is to propose different methodologies to jointly solve the problem of content caching and resource allocation in cloud-aided wireless networks under latency constraints. The problem of edge caching for wireless networks is mainly investigated under three cases: a theoretical analysis of content caching problem with storage-bandwidth trade off in Small Cell Networks (SCNs), a learning-based caching in cloud-aided wireless networks, and a latency-aware radio resource optimization in learning-based cloud-aided wireless networks. Towards achieving these goals, this dissertation makes a number of key contributions including three journal papers and one conference paper which all are published.