Computation Offloading Game for Edge Computing with Strategic Local Pre-Processing Time-Length

In this paper, a novel computation offloading framework for edge computing is proposed. Unlike existing studies, this work considers that for offloading those computation-intensive tasks, mobile users are allowed to intentionally defer the declarations of their offloading requests and reserve some time for local pre-processing. By doing so, the offloading cost (including edge service charge and transmission cost) may be reduced because of less edge service demand, while the delay cost may increase due to later report. To strike the balance, each mobile user can strategically and selfishly determine a best timing of when to declare its offloading request (or the time-length of its local preprocessing). To characterize the resulted strategic interactions, a computation offloading game built upon a queueing model with strategic queue timing is formulated. Theoretical analyses and simulations evaluate the performance of the proposed equilibrium solution and demonstrate its superiority over counterparts.