Abstract: Recently, internet of things, big data and cloud computing provide new information technologies for a variety of large-scale service systems and for their resource management and task scheduling, and they also lead to many crucial characteristics, such as, large scale, loose coupling, cross organizations, heterogeneity and so on. To study the large-scale service systems, the adversarial networks becomes a hot and interesting research direction through a simply deterministic hypothesis and an upper bound estimation for system stability. In this paper, we firstly analyze a directed ring of service adversarial network with priority, and develop an upper bound estimation method by means of a mathematical modeling and analysis, hence also propose a new discussion of system stability. Then, we study the performance measures of this network, and also provide an upper bound of the running time that each customer is served in the adversarial network. Finally, we give some numerical examples to demonstrate how the upper bounds of the total number of customers and of the running time of each customer depend on some key parameters of this adversarial network.

Key words: adversarial service network, priority, scheduling mechanism, stability