The recent advances in the areas of both wireless communication networks and machine learning
applications together with the ever-increasing smart devices call a paradigm shift on the how the
intelligence is located in the network. The conventional approaches locates the intelligence at the
core of the network such as content storage or processing. However, rapid increases in the content
streaming as well as increasing size of the computational process in parallel to available date, make
transition from centralized to distributed setup indispensable. Hence, in this dissertation we analyze
how this transition can be done efficiently, in both areas of content storage and computation. Accordingly, the dissertation consist of two main parts; first, where we study collaborative content storage and delivery, second, we study distributed computation.
In the first part, we analyze collaborative content storage and delivery in heterogeneous cellular networks with a particular focus on the user mobility and content popularity. In the second part, we
analyze the distributed computation frameworks with particular focus on the failure-tolerant strategies.