Cooperation is a key enabler for delivering popular contents to users in today's centralized wireless communication networks where the cloud-computing base station (CBS) is connected with low power base-stations, also named as remote radio heads (RRHs), via fronthaul links. The performance of today's wireless communication networks is constrained by the scarcity of radio resources and limited-capacity of fronhaul links. Edge caching (EC) is a powerful networking technique, where popular contents are cached in the edge nodes, that reduces the burden on fronthaul links. State-of-the-art literature on content delivery focuses on allocating a single user to each radio resource-block (RRB). However, this needs a large numbers of RRBs to satisfy the tremendous increase in the number of users. Consequently, network coding (NC) is employed to mix users' contents and simultaneously allocate different users to the same RRB. This dissertation designs innovative NC schemes for effective resource allocation in cache-enabled 5G networks, where the goal is to optimize different metrics, such as throughput, CBS offloading, and completion time. We first study the cross-layer throughput maximization problem in cloud-radio access network (C-RAN). For this problem, we develop novel cross-layer NC (CLNC) frameworks that propose to optimize RRHs' transmit powers and users' rates in making the coding decisions. Subsequently, we study CBS offloading and users' quality of service (QoS)-guarantee trade-off for fog-RAN system subject to the cached contents, the required minimum rate, power control, and NC constraints. Next, we focus on developing NC schemes for device-to-device (D2D)-aided F-RAN system. The developed NC schemes exploit the aforementioned F-RAN's optimization factors and potential D2D communications to minimize the completion time of users. Finally, to circumvent the necessity for CBSs in the above-mentioned centralized systems, we develop a distributed and decentralized game-theoretical NC framework for partially connected D2D networks such that the number of required D2D transmissions until all users receive all their requested contents is minimized. Simulation results are provided to attest the effectiveness of the proposed frameworks against baseline schemes for each of the above systems.