Delay-Sensitive Server Selection System for Online Multiuser Collaboration

到2020年，全球有27亿电脑游戏玩家，市场规模达1593亿美元。游戏现在是娱乐行业最大、最受欢迎的市场，在COVID-19大流行期间，它的受欢迎程度进一步提高。多用户在线游戏平台面临一个主要挑战：延迟。未能达到延迟阈值是游戏服务提供商失去用户和收益的主要原因。每100毫秒的延迟就会导致25%的玩家性能下降，即使是第三人称视角游戏，这种游戏通常可以忍受500毫秒的延迟。确保延迟不违反游戏类型的延迟阈值需要一个组织良好的资源管理系统，以便在分布式云/边缘数据中心中部署更靠近玩家的大量游戏服务器。但这也带来了另一个挑战：如何为玩家分配最好的服务器。在5G时代，这一挑战更加严峻，因为5G提供了非常低的网络延迟，所以在5G之前，服务器只是在玩家的延迟范围之外，现在因为5G而在这个范围内。在服务器数量众多的情况下，设计一种最优的服务器选择方法成为一个难以有效解决的多元优化问题。

There were 2.7 billion computer gamers globally in 2020, leading to a market of US$159.3 Billion. Gaming is now easily the largest and most popular market of the entertainment sector, and its popularity has increased even more during the COVID-19 pandemic. A multiuser online gaming platform has one major challenge: delay. Failure to meet the delay threshold is the main reason game service providers lose subscribers and revenue. Every 100 msec of latency results in about a 25% decrease in player performance even for third person perspective games which can generally tolerate delays of up to 500 msec. Ensuring the delay does not violate the game genre's delay threshold requires a well-organized resource management system for deploying a large number of game servers closer to the players in distributed cloud/edge datacenters. But this leads to another challenge: how to assign the best server to a player. This challenge is exacerbated in the age of 5G, because 5G provides very low network delays, so servers that before 5G were just outside a player's delay range are now within that range because of 5G. With so many servers, designing an optimum server selection method becomes a multivariate optimization problem that is difficult to solve efficiently. The goal of this project is to achieve real-world deployability and transfer the technology of an invention by uOttawa's research team to Swarmio Inc. The research prototype of the invention, called Delay-Sensitive Server Selection System (DS4) has been developed since September 2017 under an NSERC Strategic project in partnership with Swarmio, and has been successfully verified at the University's lab. DS4 is a server selection method ensuring that not only the delay threshold is met, but also the variance of delay among players is minimized, leading to fair gaming. This fairness is very important in the case of online video gaming because players with higher latencies perform worse than those with lower latencies and could lose the game. With further research in this project we plan to make DS4 a robust system that works beyond the lab setting and in the company's gaming platform Swarmio MatrixTM.