Stackelberg Game-Based Joint Computing Resource Allocation and Task Offloading Method in Edge Computing

Abstract

Edge computing (EC) has emerged as an important technology to support the low-delay request of massive devices nowadays. Task offloading is an essential part in EC because it can influence the use of network resources and network performance dramatically. Most existing task offloading works are only from the view of users. To effectively considering the features and objectives of both users and edge nodes from their different perspectives, a Stackelberg game-based joint computing resource allocation and task offloading method is proposed in this paper. For the nature in EC where edge nodes and users play different roles, the problem is formulated as a bi-level optimization model with multiple leaders and multiple followers. The edge nodes can be seen as leaders and the users are followers. When jointly allocating computing resource and offloading tasks, edge nodes and users have different objectives. The objective of edge nodes is to achieve the most revenue and least energy cost, and the objective of users is to obtain short delay, consume little energy and pay less. Further, considering the particular features of EC, unlike existing Stackelberg game-based task offloading research, we focus on the computing resource allocation rather than pricing. The edge nodes decide the amount of computing resources to be allocated to each user. The users will then react according to such allocation to decide task offloading strategies. Interference, delay, energy, and payoff are all considered. Evolutionary optimization method BLEAQ-II is applied to solve the designed Stackelberg game-based task offloading model. Numerical results have shown the effectiveness of the proposed method.