Dynamic Edge Computation Offloading and Scheduling for Model Task in Energy Capture Network

As an emerging and promising technique, mobile edge computing (MEC) can significantly speed up the execution of tasks and save device energy by offloading the computation-intensive tasks from resource-constrained mobiles to the MEC servers. Besides, technological advances have promoted the emergence of novel applications task with a model framework. These model frameworks are indispensable and reusable: if the model task wants to execute on MEC, both the model and data need to offload; the model can store in the cache for the later execution of the same type of tasks. What’s more, consider the limited capacity cache, it is a great challenge to replace the model dynamically to meet long-time requirements. In this paper, we jointly consider radio frequency (RF) energy capturing, computation offloading, and task scheduling in a multi-user cache-assisted MEC system. We formulate a replace algorithm and a global replacement scheduling algorithm (GRSA) to solve the mixed discrete-continuous optimization problem, which minimizes the total execution time subject to energy consumption and channel conflict. The simulation results show that our algorithm can effectively reduce computation latency.