Nonlinear Dependent Task Offloading Strategy in Mobile Edge Computing

The emerging MEC computing paradigm can effectively relieve the computational pressure on mobile devices by offloading computationally intensive tasks to edge servers, which in turn reduces energy consumption and extends device lifetime. However, due to the large decision space, nonlinear dependent tasks with complex subtask topology can only be considered as a whole for coarse-grained offloading decisions. In order to minimize the device energy consumption, in this paper we propose an algorithm named Edge Computing Offloading with All Topo Sorting by fully considering the topology of nonlinear dependent tasks. First, we can transform the nonlinear topology of a local application into a set of linear topologies by all topology sorting algorithm. Then for tasks with linear topologies, the local computation frequency, transmission power and computation offloading decisions of the mobile device are optimized to minimize its task execution energy consumption. Simulation results show that the proposed algorithm for nonlinear dependent tasks can effectively save energy consumption with low time complexity.