To Frag or to Be Fragged - An Empirical Assessment of Latency in Cloud Gaming

Abstract

With the emergence of cloud computing, diverse types of Information Technology services are increasingly provisioned through large data centers via the Internet. A relatively novel service category is cloud gaming, where video games are executed in the cloud and delivered to a client as audio/video stream. While cloud gaming substantially reduces the demand of computational power on the client side, thus enabling the use of thin clients, it may also affect the Quality of Service through the introduction of network latencies. In this work, we quantitatively examine this effect, using a self-developed measurement tool and a set of actual cloud gaming providers. For the two providers and three games in our experiment, we find absolute increases in latency between approximately 40 ms and 150 ms, or between 85% and 800% in relative terms.

On the Latency of USB-Connected Input Devices

“…A video shows latency values for the G300 which are in the 3 -5 ms range, roughly similar to the ones we measured. Another blogger, sousu, conducted absolute, end-to-end latency measurements for over 50 mice with another destructive method 12 . They soldered a phototransistor to the contacts of the mouse button and glued it to a computer screen.…”

“…High latency disproportionally increases task completion time for pointing tasks by slowing down feedback loops [18]. In certain computer games, such as first-person shooters or real-time strategy games, the amount of latency a user experiences decides over virtual life or death [12]. In experimental psychology, researchers conducting reaction time studies are concerned about the latency of computers tainting the results [24].…”

Yet Another Latency Measuring Device

“…This so called end-to-end latency directly influences task difficulty [18] and should therefore be minimized to ensure optimal user experience. The influence of latency is especially important for real time applications, as it can decide about virtual life and death in video games [7,15,29] or influence the outcome of psychological experiments [16,23]. Shneiderman et al [28] recommend a system response time of 50 to 150 milliseconds for simple tasks like typing or pointing and Seow [27] classifies response times of 200 milliseconds and shorter as immediate.…”

DispLagBox

“…They conclude that even latencies starting from 16 to 60 ms can negatively affect the performance of users for a large number of tasks. Consistently low latency is especially important when performing critical tasks such as remote-controlling robots or vehicles -but also the video game community has a high demand for fast and reliable systems [6]. Thus, input devices and displays for gamers are optimized for low latency.…”