Cellular networks in the 5G epoch need to satisfy numerous challenging requirements such a low content delivery latency, a traffic reduction in the backhual links and others. Cache-enabled wireless networks is one of the technologies that can reduce the content delivery latency as well as decrease the traffic in the backhaul links. This thesis focuses on the study of how to exploit caching and allocate radio resources in 5G cellular networks more effectively. In particular, the proposed techniques aim to recognize main challenges in the design of resource allocation and content caching schemes that meet these new needs, and to provide effective solutions to address these challenges. Firstly, we investigated a joint beamforming and admission control design in cacheenabled cloud radio access networks (Cloud-RANs). The main goal is to minimize the total network cost which is a combination of transmission power cost and fronthaul cost while admitting as many users as possible to satisfy certain considerations. The single objective optimization problem has been formulated as a mixed-integer second order cone programming, and then an efficient branch and bound algorithm has been proposed. Secondly, we proposed a learning-based method to overcome the complexity issues which are emerging due to using mathematical optimization methods when designing beamforming and admission control in cache-enabled Cloud-RANs in the previous problem. A determinantal point process learning framework to obtain the subset of admitted users for cache-enabled Cloud-RANs is developed. Finally, we studied the impact of cooperative caching and coordinated transmission techniques in a cognitive radio networks to improve the system performance. Depending on the availability and placement of the requested content, different transmission techniques and caching strategies are suggested.
