A Self-organizing Price-based Mechanism for Frame Rate Optimization in Cloud Gaming Networks Considering Quality of Experience

Abstract

One of the most important issues in cloud services, especially in cloud gaming, is the limitations relevant to the hardware capabilities of end-users who are using mobile devices. In the cloud gaming terminology, the users who have poor resources are referred to as “Thin Clients”. On the other hand, cloud service providers should be able to provide minimum frame rate required by a computer game, simultaneously, to several thin clients at the same time, in such a way that meets the users' Quality of Experience (QoE). Generally, computer games have different degrees of render settings by default, in terms of three levels: high, medium, and low. These setting levels, in turn, can be translated into hardware and operating system capabilities, such as maximum textual details, brightness, contrast, and so on. Cloud game service providers use rendering settings as a means for transferring the frames of the game to thin clients. They also try to maximize their own profits with pricing on game frames. This paper addresses the subject of resource optimization using a pricing mechanism. The designed mechanism is able to allocate game frames in such a way that leads to achieving high user QoE and also maximizing the profit of the cloud server. We set the price of the frames considering their types. Each type has its own processor, memory, GPU requirements, etc. Our goal is to find the optimal combination of types for each user. Our proposed method is able to dynamically adapt itself to the requirements of both sides of the network. We also consider the scheduling issues of the cloud server in order to transfer the frames to the requested users. When the number of users is very high, the proposed mechanism applies the settings in such a way that the QoE is maintained at the desired level and the user's satisfaction is not violated at all. Simulation results show that the proposed priced-based approach attains lower bandwidth consumption values compared to the case in which no pricing mechanism is used. Also, the social welfare of the users is maximized.