Online task scheduling for edge computing based on repeated stackelberg game

A key function of an edge service provider (ESP) is to dynamically allocate resources to tasks existing at the edges upon request. This is, however, a challenging task due to a number of several factors: real-time decision-making without any prior knowledge of future arrivals, tasks' satisfactions provided by requests, and utilization of resources. To address these challenges, we propose an online scheduling that maps various tasks to the given relevant resources based on a repeated Stackelberg game. First, we model this problem as a long-term vs. short-term repeated Stackelberg game. In particular, for each round of the game, acting as a short-term leader, a user with a request first decides the unit prices for processing tasks within the relevant budget to maximize current total satisfaction of tasks. Then, based on the prices offered by different users in different rounds, to maximize the long-term profits earned from users, the ESP acts as the follower whose strategy is matching resources with tasks, and splitting those tasks among different edge centers owning various types of resources (edge mobile devices). The Stackelberg equilibrium between the ESP and the users is obtained using our proposed algorithms. Finally, we evaluate the effectiveness of our proposal, in terms of task distributions.