The recent advances of computing, caching, and communication (C^3) can have significant impacts on the future network design. In addition, there are many new technologies for big data analysis and cross-layer optimization. In this talk, we limit our scope to three specific cases, while want to give insights. First, a new formulation for bandwidth allocation and routing problem is formulated for wireless network virtualization, and the decomposition algorithms are constructed using the alternating direction method of multipliers that can be implemented in a parallel and distributed fashion. Second, the caching based D2D communication scheme is investigated by taking social ties among users and common interests into consideration, and the hypergraph and matching frameworks are employed for optimization. Third, we consider mobile social networks together with C^3 networks using deep learning to study social networks and a novel deep reinforcement learning approach to automatically make a decision for optimally allocating the network resources.