Abstract
Prompted by the remarkable progress in both cloud computing and GPU virtualization, cloud gaming has become increasingly more popular in the gaming industry. In cloud gaming, games are stored and run on cloud servers and the gamers interact with games through thin clients. Cloud gaming service providers generally employ multiple geographically distributed data centers to deliver their services. The main challenge for cloud gaming service providers is to find the best tradeoff between two contradicting objectives: reducing the infrastructure operating costs and increasing the quality of player's experience. In this paper, we address a virtual machine provisioning problem for multiplayer cloud gaming with the objective of minimizing both the inter-player delay among interacting players and the electricity costs of cloud gaming service providers, while providing the good-enough response delay to gamers. We formulate the problem into a constrained multiobjective optimization problem and propose an improved grey wolf algorithm to solve the problem. The performance of our proposed algorithm are assessed by simulation experiments based on the real-world parameters. The results show the superior performance of the proposed approach in comparison with the state-of-the-art approaches applied to similar problems.
Highlights
As the gaming industry matures, games became more and more complex and demand for the latest hardware such as multicore processors and high-end graphic cards for fluent game playing
We focus on the provisioning of the rendering servers (VMs) for multiplayer cloud gaming (MCG) in geographically distributed data centers with the objective of minimizing both the inter-player delay among interacting players and the electricity costs of MCG providers, while providing the good-enough response delay to gamers
Chen et al [7] addressed the virtual machine (VM) provisioning problem that aims at minimizing the inter-player delay, while preserving good-enough response delay experienced by players
Summary
As the gaming industry matures, games became more and more complex and demand for the latest hardware such as multicore processors and high-end graphic cards for fluent game playing. We focus on the provisioning of the rendering servers (VMs) for MCG in geographically distributed data centers with the objective of minimizing both the inter-player delay among interacting players and the electricity costs of MCG providers, while providing the good-enough response delay to gamers. We formulate the VM provisioning problem into a constrained multiobjective optimization problem to answer the following questions: a) how to provision VMs without violating the resource capacity and responsiveness constraints, b) how to adjust the number of active servers in each data center, c) how to fully exploit the geographical heterogeneity of electricity prices and data center PUEs and d) how to achieve a trade-off between the fairness and the electricity cost. Our simulation results show that, compared with other alternatives, our proposed algorithm can achieve lower electricity costs and better fairness, while providing the good-enough response delay to gamers.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
