Joint Optimization of Task Offloading and Resource Allocation of Fog Network by Considering Matching Externalities and Dynamics

Abstract

How to jointly optimize task offloading and resource allocation to minimize the task failure rate and task payments remains an unresolved challenge in fog networks. Focusing on this problem, this research formulates a novel task offloading and resource allocation model with two offloading modes and on-demand virtual resource units (VRUs). This model is decomposed into two sub-problems to solve: a joint task offloading and resource allocation optimization problem and a matching problem with externalities and dynamics. First, for a given terminal node (TN) and fog node (FN), this research theoretically derives the optimal offloading ratio and resource allocation strategy to minimize the payment of TNs for two offloading modes, i.e., immediate and queued offloading. Second, in the multi-TNs and multi-FNs scenario, the problem of making the task offloading decision is transformed into a many-to-one matching game by considering externalities and dynamics. Finally, a Deferred acceptance-based Loss ratio and Payment Minimized task Offloading and resource Allocation optimization (DLPMOA) algorithm is proposed to derive a stable and Pareto-optimal match. The simulation results show that the proposed DLPMOA has better performance in terms of task failure rate, task average payment, fog computing resource utilization, and fairness than the state-of-the-art methods.