Modeling and Trade-Off for Mobile Communication, Computing and Caching Networks

This paper considers a new mobile edge computing (MEC) model where the MEC server has the input and output data of all computation tasks and communicates with multiple caching-and-computing- enabled mobile devices via a shared wireless link. Each mobile device can pre-store the input or output data of a task and also execute a task locally. We aim to investigate the impact of local caching and computing at mobile devices as well as content-centric multicast transmission on the saving of required bandwidth on the wireless link. To this end, we first formulate a joint caching and computing decision optimization problem to minimize the required transmission bandwidth subject to latency, caching and energy constraints at each mobile device in the general case. The joint policy optimization problem is shown to be NP-hard. Based on equivalent transformation and exact penalization of the problem, a stationary point is obtained via concave convex procedure. In the special case where all the computation tasks are symmetric and user requests are uniform, we obtain the closed- form expressions for the local caching gain, local computing gain, and multicasting gain. Our results indicate that exploiting the computing and caching resources at mobile devices can provide significant bandwidth savings.